XC4000 Series
Field Programmable Gate Arrays
September 18, 1996 (Version 1.04)
Product Specification
XC4000-Series Features
Additional XC4000EX/XL Features
Note: XC4000-Series devices described in this data sheet
include the XC4000E, XC4000EX, XC4000L, and
XC4000XL. This information does not apply to the older
Xilinx families: XC4000, XC4000A, XC4000D or XC4000H.
For information on these devices, see the Xilinx WEBLINX
at http://www.xilinx.com.
•
•
•
•
•
•
•
•
•
•
Third Generation Field-Programmable Gate Arrays
- Select-RAMTM memory: on-chip ultra-fast RAM with
- synchronous write option
- dual-port RAM option
- Fully PCI compliant (speed grades -3 and faster)
- Abundant flip-flops
- Flexible function generators
- Dedicated high-speed carry logic
- Wide edge decoders on each edge
- Hierarchy of interconnect lines
- Internal 3-state bus capability
- 8 global low-skew clock or signal distribution
networks
System Performance to 66 MHz
Flexible Array Architecture
Systems-Oriented Features
- IEEE 1149.1-compatible boundary scan logic
support
- Individually programmable output slew rate
- Programmable input pull-up or pull-down resistors
- 12-mA sink current per XC4000E output (4 mA per
XC4000L output)
Configured by Loading Binary File
- Unlimited reprogrammability
Readback Capability
Backward Compatible with XC4000 Devices
XACTstep Development System runs on '386/'486/
Pentium-type PC, Sun-4, and Hewlett-Packard 700
series
- Interfaces to popular design environments
- Fully automatic mapping, placement and routing
- Interactive design editor for design optimization
- RAM/ROM compiler
•
•
•
•
•
•
•
•
Highest Capacity — Over 130,000 Usable Gates
Additional Routing Over XC4000E
- almost twice the routing capacity for high-density
designs
Buffered Interconnect for Maximum Speed
New Latch Capability in Configurable Logic Blocks
Improved VersaRingTM I/O Interconnect for Better Fixed
Pinout Flexibility
Flexible New High-Speed Clock Network
- 8 additional Early Buffers for shorter clock delays
- 4 additional FastCLKTM buffers for fastest clock input
- Virtually unlimited number of clock signals
Optional Multiplexer or 2-input Function Generator on
Device Outputs
High-Speed Parallel ExpressTM Configuration Mode
Improved I/O Setup and Clock-to-Output with FastCLK
and Global Early Buffers
4 Additional Address Bits in Master Parallel
Configuration Mode
Introduction
XC4000-Series high-performance, high-capacity Field Programmable Gate Arrays (FPGAs) provide the benefits of
custom CMOS VLSI, while avoiding the initial cost, long
development cycle, and inherent risk of a conventional
masked gate array.
The result of eleven years of FPGA design experience and
feedback from thousands of customers, these FPGAs combine architectural versatility, on-chip Select-RAM memory
with edge-triggered and dual-port modes, increased speed,
abundant routing resources, and new, sophisticated software to achieve fully automated implementation of complex, high-density, high-performance designs.
The XC4000 Series currently has 19 members, as shown
in Table 1.
Low-Voltage Versions Available
•
•
•
Low-Voltage Devices Function at 3.0 - 3.6 Volts
XC4000L: Low-Voltage Versions of XC4000E devices
XC4000XL: Low-Voltage Versions of XC4000EX
devices
September 18, 1996 (Version 1.04)
4-5
XC4000 Series Field Programmable Gate Arrays
Table 1: XC4000-Series Field Programmable Gate Arrays
Device
XC4003E
XC4005E/L
XC4006E
XC4008E
XC4010E/L
XC4013E/L
XC4020E
XC4025E
XC4028EX/XL
XC4036EX/XL
XC4044EX/XL
XC4052XL
XC4062XL
Max Logic Max. RAM
Typical
Total
Number
Gates
Bits
Gate Range
CLB
Logic
of
(No RAM) (No Logic) (Logic and RAM)*
Matrix
Blocks Flip-Flops
3,000
3,200
2,000 - 5,000
10 x 10
100
360
5,000
6,272
3,000 - 9,000
14 x 14
196
616
6,000
8,192
4,000 - 12,000
16 x 16
256
768
8,000
10,368
6,000 - 15,000
18 x 18
324
936
10,000
12,800
7,000 - 20,000
20 x 20
400
1,120
13,000
18,432
10,000 - 30,000
24 x 24
576
1,536
20,000
25,088
13,000 - 40,000
28 x 28
784
2,016
25,000
32,768
15,000 - 45,000
32 x 32
1,024
2,560
28,000
32,768
18,000 - 50,000
32 x 32
1,024
2,560
36,000
41,472
22,000 - 65,000
36 x 36
1,296
3,168
44,000
51,200
27,000 - 80,000
40 x 40
1,600
3,840
52,000
61,952
33,000 - 100,000
44 x 44
1,936
4,576
62,000
73,728
40,000 - 130,000
48 x 48
2,304
5,376
Larger Devices Available in the First Half of 1997
Max.
Decode
Inputs
per side
30
42
48
54
60
72
84
96
96
108
120
132
144
Max.
User I/O
80
112
128
144
160
192
224
256
256
288
320
352
384
* Max values of Typical Gate Range include 20-30% of CLBs used as RAM.
Note: Throughout the functional descriptions in this document, references to the XC4000E device family include the
XC4000L, and references to the XC4000EX device family
include the XC4000XL, unless explicitly stated otherwise.
References to the XC4000 Series include the XC4000E,
XC4000EX, XC4000L, and XC4000XL families. All functionality in low-voltage families is the same as in the corresponding 5-Volt family, except where numerical references
are made to timing, power, or current-sinking capability.
Description
XC4000-Series devices are implemented with a regular,
flexible, programmable architecture of Configurable Logic
Blocks (CLBs), interconnected by a powerful hierarchy of
versatile routing resources, and surrounded by a perimeter
of programmable Input/Output Blocks (IOBs). They have
generous routing resources to accommodate the most
complex interconnect patterns.
The devices are customized by loading configuration data
into internal memory cells. The FPGA can either actively
read its configuration data from an external serial or byteparallel PROM (master modes), or the configuration data
4-6
can be written into the FPGA from an external device
(slave, peripheral and Express modes).
XC4000-Series FPGAs are supported by powerful and
sophisticated software, covering every aspect of design
from schematic or behavioral entry, floorplanning, simulation, automatic block placement and routing of interconnects, to the creation, downloading, and readback of the
configuration bit stream.
Because Xilinx FPGAs can be reprogrammed an unlimited
number of times, they can be used in innovative designs
where hardware is changed dynamically, or where hardware must be adapted to different user applications.
FPGAs are ideal for shortening design and development
cycles, and also offer a cost-effective solution for production rates well beyond 5,000 systems per month. For lowest
high-volume unit cost, a design can first be implemented in
the XC4000E or XC4000EX, then migrated to one of Xilinx’
compatible HardWire mask-programmed devices.
Table 2 shows density and performance for a few common
circuit functions that can be implemented in XC4000-Series
devices.
September 18, 1996 (Version 1.04)
Table 2: Density and Performance for Several Common Circuit Functions in XC4000E1
Design Class
Memory
Logic
Note:
Function
256 x 8 Single Port (read/modify/write)
32 x 16 bit FIFO
simultaneous read/write
MUXed read/write
9 bit Shift Register (with enable)
16 bit Pre-Scaled Counter
16 bit Loadable Counter
16 bit Accumulator
8 bit, 16 tap FIR Filter sample rate
parallel
serial
8 x 8 Parallel Multiplier
single stage, register to register
16 bit Address Decoder (internal decode)
9 bit Parity Checker
CLBs Used
72
XC4000E-3
63
XC4000E-2
80
Units
MHz
48
32
5
8
8
9
63
63
170
142
65
65
80
80
200
170
76
76
MHz
MHz
MHz
MHz
MHz
MHz
400
68
55
8.1
65
10
MHz
MHz
73
3
1
37
4.7
4.3
30
3.9
2.7
ns
ns
ns
1. Most functions are faster in XC4000EX due to faster carry logic, direct connects, and other additional interconnect.
Taking Advantage of Reconfiguration
FPGA devices can be reconfigured to change logic function
while resident in the system. This capability gives the system designer a new degree of freedom not available with
any other type of logic.
Hardware can be changed as easily as software. Design
updates or modifications are easy, and can be made to
products already in the field. An FPGA can even be recon-
September 18, 1996 (Version 1.04)
figured dynamically to perform different functions at different times.
Reconfigurable logic can be used to implement system
self-diagnostics, create systems capable of being reconfigured for different environments or operations, or implement
multi-purpose hardware for a given application. As an
added benefit, using reconfigurable FPGA devices simplifies hardware design and debugging and shortens product
time-to-market.
4-7
XC4000 Series Field Programmable Gate Arrays
XC4000E and XC4000EX Families
Compared to the XC4000
For readers already familiar with the XC4000 family of Xilinx Field Programmable Gate Arrays, the major new features in the XC4000-Series devices are listed in this
section.
The biggest advantages of XC4000E and
XC4000EX devices are significantly increased system
speed, greater capacity, and new architectural features,
particularly Select-RAM memory. The XC4000EX devices
also offer many new routing features, including special
high-speed clock buffers that can be used to capture input
data with minimal delay.
Any XC4000E device is pinout- and bitstream-compatible
with the corresponding XC4000 device. An existing
XC4000 bitstream can be used to program an XC4000E
device. However, since the XC4000E includes many new
features, an XC4000E bitstream cannot be loaded into an
XC4000 device.
Most XC4000EX devices have no corresponding XC4000
devices, because of the larger CLB arrays. The XC4028EX
has the same array size as the XC4025 and XC4025E, but
is not bitstream-compatible.
However, the XC4025,
XC4025E, and XC4028EX are all pinout-compatible.
Improvements in XC4000E and XC4000EX
Increased System Speed
Delays in FPGA-based designs are layout dependent.
There is a rule of thumb designers can consider—the system clock rate should not exceed one third to one half of the
specified toggle rate. Critical portions of a design, such as
shift registers and simple counters, can run faster—approximately two thirds of the specified toggle rate.
XC4000E and XC4000EX devices can run at synchronous
system clock rates of up to 66 MHz, and internal performance can exceed 150 MHz. This increase in performance
over the previous families stems from improvements in both
device processing and system architecture.
XC4000Series devices use a sub-micron triple-layer metal process.
In addition, many architectural improvements have been
made, as described below.
Select-RAM Memory: Edge-Triggered, Synchronous
RAM Modes
The RAM in any CLB can be configured for synchronous,
edge-triggered, write operation. The read operation is not
affected by this change to an edge-triggered write.
Dual-Port RAM
A separate option converts the 16x2 RAM in any CLB into a
16x1 dual-port RAM with simultaneous Read/Write.
The function generators in each CLB can be configured as
either level-sensitive (asynchronous) single-port RAM,
edge-triggered (synchronous) single-port RAM, edge-triggered (synchronous) dual-port RAM, or as combinatorial
logic.
Configurable RAM Content
The RAM content can now be loaded at configuration time,
so that the RAM starts up with user-defined data.
H Function Generator
In XC4000-Series devices, the H function generator is
more versatile than in the XC4000. Its inputs can come not
only from the F and G function generators but also from up
to three of the four control input lines. The H function generator can thus be totally or partially independent of the
other two function generators, increasing the maximum
capacity of the device.
IOB Clock Enable
The two flip-flops in each IOB have a common clock enable
input, which through configuration can be activated individually for the input or output flip-flop or both. This clock
enable operates exactly like the EC pin on the XC4000
CLB. This new feature makes the IOBs more versatile, and
avoids the need for clock gating.
Output Drivers
XC4000-Series -3 and faster speed grades are fully PCI
compliant. XC4000E and XC4000EX devices can be used
to implement a one-chip PCI solution.
The output pull-up structure defaults to a TTL-like totempole. This driver is an n-channel pull-up transistor, pulling
to a voltage one transistor threshold below Vcc, just like the
XC4000 outputs. Alternatively, XC4000-Series devices can
be globally configured with CMOS outputs, with p-channel
pull-up transistors pulling to Vcc. Also, the configurable pullup resistor in the XC4000 Series is a p-channel transistor
that pulls to Vcc, whereas in the XC4000 it is an n-channel
transistor that pulls to a voltage one transistor threshold
below Vcc.
Carry Logic
Input Thresholds
The speed of the carry logic chain has increased dramatically. Some parameters, such as the delay on the carry
chain through a single CLB (TBYP), have improved by as
much as 50% from XC4000 values. See “Fast Carry Logic”
on page 21 for more information.
The input thresholds can be globally configured for either
TTL (1.2 V threshold) or CMOS (2.5 V threshold), just like
XC2000 and XC3000 inputs. The two global adjustments of
input threshold and output level are independent of each
other.
PCI Compliance
4-8
September 18, 1996 (Version 1.04)
Global Signal Access to Logic
Faster Input and Output
There is additional access from global clocks to the F and G
function generator inputs.
A fast, dedicated early clock sourced by global clock buffers
is available for the IOBs. To ensure synchronization with
the regular global clocks, a Fast Capture latch driven by the
early clock is available. The input data can be initially
loaded into the Fast Capture latch with the early clock, then
transferred to the input flip-flop or latch with the low-skew
global clock. A programmable delay on the input can be
used to avoid hold-time requirements. See “IOB Input Signals” on page 24 for more information.
Configuration Pin Pull-Up Resistors
During configuration, the three mode pins, M0, M1, and
M2, have weak pull-up resistors. For the most popular configuration mode, Slave Serial, the mode pins can thus be
left unconnected.
The three mode inputs can be individually configured with
or without weak pull-up or pull-down resistors after configuration.
The PROGRAM input pin has a permanent weak pull-up.
Soft Start-up
Like the XC3000A, XC4000-Series devices have “Soft
Start-up.” When the configuration process is finished and
the device starts up, the first activation of the outputs is
automatically slew-rate limited. This feature avoids potential ground bounce when all outputs are turned on simultaneously. Immediately after start-up, the slew rate of the
individual outputs is, as in the XC4000 family, determined
by the individual configuration option.
XC4000 and XC4000A Compatibility
Existing XC4000 bitstreams can be used to configure an
XC4000E device. XC4000A bitstreams must be recompiled for use with the XC4000E due to improved routing
resources, although the devices are pin-for-pin compatible.
Additional Improvements in XC4000EX
Only
Increased Routing
New interconnect in the XC4000EX includes twenty-two
additional vertical lines in each column of CLBs and twelve
new horizontal lines in each row of CLBs. The twelve
“Quad Lines” in each CLB row and column include optional
repowering buffers for maximum speed. Additional highperformance routing near the IOBs enhances pin flexibility.
September 18, 1996 (Version 1.04)
Latch Capability in CLBs
Storage elements in the XC4000EX CLB can be configured
as either flip-flops or latches. This capability makes the
FPGA highly synthesis-compatible.
IOB Output MUX From Output Clock
A multiplexer in the IOB allows the output clock to select
either the output data or the IOB clock enable as the output
to the pad. Thus, two different data signals can share a single output pad, effectively doubling the number of device
outputs without requiring a larger, more expensive package. This multiplexer can also be configured as an ANDgate to implement a very fast pin-to-pin path. See “IOB
Output Signals” on page 27 for more information.
Express Configuration Mode
A new slave configuration mode accepts parallel data input.
Data is processed in parallel, rather than serialized internally. Therefore, the data rate is eight times that of the six
conventional configuration modes.
Additional Address Bits
Larger devices require more bits of configuration data. A
daisy chain of several large XC4000EX devices may
require a PROM that cannot be addressed by the eighteen
address bits supported in the XC4000E. The XC4000EX
family therefore extends the addressing in Master Parallel
configuration mode to 22 bits.
4-9
XC4000 Series Field Programmable Gate Arrays
Table 3: CLB Count of Selected XC4000-Series Soft Macros
7400 Equivalents
CLBs
‘138
5
‘139
2
‘147
5
‘148
6
‘150
5
‘151
3
‘152
3
‘153
2
‘154
16
‘157
2
‘158
2
‘160
5
‘161
6
‘162
8
‘163
8
‘164
4
‘165s
9
‘166
5
‘168
7
‘174
3
‘194
5
‘195
3
‘280
3
‘283
8
‘298
2
‘352
2
‘390
3
‘518
3
‘521
3
Explanation of RAM nomenclature
s = single-port edge-triggered
d = dual-port edge-triggered
no extension = level-sensitive
4-10
Barrel Shifters
brlshft4
brlshft8
4-Bit Counters
cd4cd
cd4cle
cd4rle
cb4ce
cb4cle
cb4re
8- and 16-Bit Counters
cb8ce
cb8re
cc16ce
cc16cle
cc16cled
Identity Comparators
comp4
comp8
comp16
Magnitude Comparators
compm4
compm8
compm16
RAMs
ram16x4
ram16x4s
ram16x4d
CLBs Multiplexers
CLBs
4 m2-1e
1
13 m4-1e
1
m8-1e
3
m16-1e
5
3 Registers
5 rd4r
2
6 rd8r
4
3 rd16r
8
6
5
Shift Registers
6 sr8ce
4
10 sr16re
8
9 Decoders
9 d2-4e
2
21 d3-8e
4
d4-16e
16
1
2
5
Explanation of counter nomenclature
cb = binary counter
4
cd = BCD counter
9
cc = cascadable binary counter
20
d = bidirectional
l = loadable
e = clock enable
r = synchronous reset
c = asynchronous clear
2
2
4
September 18, 1996 (Version 1.04)
Detailed Functional Description
XC4000-Series devices achieve high speed through
advanced semiconductor technology and improved architecture. The XC4000E and XC4000EX support system
clock rates of up to 66 MHz and internal performance in
excess of 150 MHz. Compared to older Xilinx FPGA families, XC4000-Series devices are more powerful. They offer
on-chip edge-triggered and dual-port RAM, clock enables
on I/O flip-flops, and wide-input decoders. They are more
versatile in many applications, especially those involving
RAM. Design cycles are faster due to a combination of
increased routing resources and more sophisticated software.
Basic Building Blocks
Xilinx user-programmable gate arrays include two major
configurable elements: configurable logic blocks (CLBs)
and input/output blocks (IOBs).
•
•
CLBs provide the functional elements for constructing
the user’s logic.
IOBs provide the interface between the package pins
and internal signal lines.
Three other types of circuits are also available:
•
•
•
3-State buffers (TBUFs) driving horizontal longlines are
associated with each CLB.
Wide edge decoders are available around the periphery
of each device.
An on-chip oscillator is provided.
Programmable interconnect resources provide routing
paths to connect the inputs and outputs of these configurable elements to the appropriate networks.
The functionality of each circuit block is customized during
configuration by programming internal static memory cells.
The values stored in these memory cells determine the
logic functions and interconnections implemented in the
FPGA.
Each of these available circuits is described in this section.
Configurable Logic Blocks (CLBs)
Configurable Logic Blocks implement most of the logic in
an FPGA. The principal CLB elements are shown in
Figure 1. The number of CLBs needed to implement
selected soft macros is shown in Table 3.
Two 4-input function generators (F and G) offer unrestricted
versatility. Most combinatorial logic functions need four or
fewer inputs. However, a third function generator (H) is provided. The H function generator has three inputs. Either
zero, one, or both of these inputs can be the outputs of F
and G; the other input(s) are from outside the CLB. The
CLB can, therefore, implement certain functions of up to
nine variables, like parity check or expandable-identity
comparison of two sets of four inputs.
Each CLB contains two storage elements that can be used
to store the function generator outputs. However, the storage elements and function generators can also be used
independently. These storage elements can be configured
as flip-flops in both XC4000E and XC4000EX devices; in
the XC4000EX they can optionally be configured as
latches. DIN can be used as a direct input to either of the
two storage elements. H1 can drive the other through the H
function generator. Function generator outputs can also
drive two outputs independent of the storage element outputs. This versatility increases logic capacity and simplifies
routing.
Thirteen CLB inputs and four CLB outputs provide access
to the function generators and storage elements. These
inputs and outputs connect to the programmable interconnect resources outside the block.
Function Generators
Four independent inputs are provided to each of two function generators (F1 - F4 and G1 - G4). These function generators, with outputs labeled F’ and G’, are each capable of
implementing any arbitrarily defined Boolean function of
four inputs. The function generators are implemented as
memory look-up tables. The propagation delay is therefore
independent of the function implemented.
A third function generator, labeled H’, can implement any
Boolean function of its three inputs. Two of these inputs
can optionally be the F’ and G’ functional generator outputs. Alternatively, one or both of these inputs can come
from outside the CLB (H2, H0). The third input must come
from outside the block (H1).
Signals from the function generators can exit the CLB on
two outputs. F’ or H’ can be connected to the X output. G’
or H’ can be connected to the Y output.
A CLB can be used to implement any of the following functions:
•
any function of up to four variables, plus any second
function of up to four unrelated variables, plus any third
function of up to three unrelated variables1
• any single function of five variables
• any function of four variables together with some
functions of six variables
• some functions of up to nine variables.
1. When three separate functions are generated, one of the function outputs must be captured in a flip-flop internal to the CLB. Only two
unregistered function generator outputs are available from the CLB.
September 18, 1996 (Version 1.04)
4-11
XC4000 Series Field Programmable Gate Arrays
C1 • • • C4
4
H1
D IN /H 2
EC
SR/H 0
G4
G3
G2
S/R
CONTROL
Bypass
DIN
F'
G'
H'
LOGIC
FUNCTION G'
OF
G1-G4
YQ
D
SD
Q
G1
LOGIC
FUNCTION
OF
H'
F', G',
AND
H1
EC
RD
G'
H'
1
Y
F4
F3
F2
Bypass
S/R
CONTROL
DIN
F'
G'
H'
LOGIC
FUNCTION F'
OF
F1-F4
XQ
D
SD
Q
F1
EC
RD
K
(CLOCK)
1
H'
F'
X
Multiplexer Controlled
by Configuration Program
X6692
Figure 1: Simplified Block Diagram of XC4000-Series CLB (RAM and Carry Logic functions not shown)
Implementing wide functions in a single block reduces both
the number of blocks required and the delay in the signal
path, achieving both increased capacity and speed.
The versatility of the CLB function generators significantly
improves system speed. In addition, the design-software
tools can deal with each function generator independently.
This flexibility improves cell usage.
Flip-Flops
The CLB can pass the combinatorial output(s) to the interconnect network, but can also store the combinatorial
results or other incoming data in one or two flip-flops, and
connect their outputs to the interconnect network as well.
The two edge-triggered D-type flip-flops have common
clock (K) and clock enable (EC) inputs. Either or both clock
inputs can also be permanently enabled. Storage element
functionality is described in Table 4.
Table 4: CLB Storage Element Functionality
(active rising edge is shown)
Mode
Power-Up or
GSR
Flip-Flop
Latch
Both
Legend:
X
__/
SR
0*
1*
K
EC
SR
D
Q
X
X
X
X
SR
X
__/
0
1
0
X
X
1*
X
1*
1*
0
1
0*
0*
0*
0*
0*
X
D
X
X
D
X
SR
D
Q
Q
D
Q
Don’t care
Rising edge
Set or Reset value. Reset is default.
Input is Low or unconnected (default value)
Input is High or unconnected (default value)
Latches (XC4000EX only)
The CLB storage elements can also be configured as
latches. The two latches have common clock (K) and clock
enable (EC) inputs. Storage element functionality is
described in Table 4.
4-12
Clock Input
Each flip-flop can be triggered on either the rising or falling
clock edge. The clock pin is shared by both storage elements. However, the clock is individually invertible for each
storage element. Any inverter placed on the clock input is
automatically absorbed into the CLB.
September 18, 1996 (Version 1.04)
Clock Enable
Data Inputs and Outputs
The clock enable signal (EC) is active High. The EC pin is
shared by both storage elements. If left unconnected for
either, the clock enable for that storage element defaults to
the active state. EC is not invertible within the CLB.
The source of a storage element data input is programmable. It is driven by any of the functions F’, G’, and H’, or by
the Direct In (DIN) block input. The flip-flops or latches drive
the XQ and YQ CLB outputs.
Set/Reset
Two fast feed-through paths are available, as shown in
Figure 1. A two-to-one multiplexer on each of the XQ and
YQ outputs selects between a storage element output and
any of the control inputs. This bypass is sometimes used
by the automated router to repower internal signals.
An asynchronous storage element input (SR) can be configured as either set or reset. This configuration option
determines the state in which each flip-flop becomes operational after configuration. It also determines the effect of a
Global Set/Reset pulse during normal operation, and the
effect of a pulse on the SR pin of the CLB. All three set/
reset functions for any single flip-flop are controlled by the
same configuration data bit.
The set/reset state can be independently specified for each
flip-flop. This input can also be independently disabled for
either flip-flop.
The set/reset state is specified by using the INIT attribute,
or by placing the appropriate set or reset flip-flop library
symbol.
Control Signals
Multiplexers in the CLB map the four control inputs (C1 - C4
in Figure 1) into the four internal control signals (H1, DIN/
H2, SR/H0, and EC). Any of these inputs can drive any of
the four internal control signals.
When the logic function is enabled, the four inputs are:
•
•
EC — Enable Clock
SR/H0 — Asynchronous Set/Reset or H function
generator Input 0
DIN/H2 — Direct In or H function generator Input 2
H1 — H function generator Input 1.
SR is active High. It is not invertible within the CLB.
•
•
Global Set/Reset
When the memory function is enabled, the four inputs are:
A separate Global Set/Reset line (not shown in Figure 1)
sets or clears each storage element during power-up,
reconfiguration, or when a dedicated Reset net is driven
active. This global net (GSR) does not compete with other
routing resources; it uses a dedicated distribution network.
•
•
•
•
Each flip-flop is configured as either globally set or reset in
the same way that the local set/reset (SR) is specified.
Therefore, if a flip-flop is set by SR, it is also set by GSR.
Similarly, a reset flip-flop is reset by both SR and GSR.
Using FPGA Flip-Flops and Latches
GSR can be driven from any user-programmable pin as a
global reset input. To use this global net, place an input
pad and input buffer in the schematic or HDL code, driving
the GSR pin of the STARTUP symbol. (See Figure 2.) A
specific pin location can be assigned to this input using a
LOC attribute or property, just as with any other user-programmable pad. An inverter can optionally be inserted
after the input buffer to invert the sense of the Global Set/
Reset signal.
Alternatively, GSR can be driven from any internal node.
STARTUP
PAD
IBUF
GSR
GTS
Q2
Q3
Q1Q4
CLK DONEIN
EC — Enable Clock
WE — Write Enable
D0 — Data Input to F and/or G function generator
D1 — Data input to G function generator (16x1 and
16x2 modes) or 5th Address bit (32x1 mode).
The abundance of flip-flops in the XC4000 Series invites
pipelined designs. This is a powerful way of increasing performance by breaking the function into smaller subfunctions and executing them in parallel, passing on the results
through pipeline flip-flops. This method should be seriously
considered wherever throughput is more important than
latency.
To include a CLB flip-flop, place the appropriate library
symbol. For example, FDCE is a D-type flip-flop with clock
enable and asynchronous clear. The corresponding latch
symbol (for the XC4000EX only) is called LDCE.
In XC4000-Series devices, the flip flops can be used as
registers or shift registers without blocking the function generators from performing a different, perhaps unrelated task.
This ability increases the functional capacity of the devices.
The CLB setup time is specified between the function generator inputs and the clock input K. Therefore, the specified
CLB flip-flop setup time includes the delay through the
function generator.
X5260
Figure 2: Schematic Symbols for Global Set/Reset
September 18, 1996 (Version 1.04)
4-13
XC4000 Series Field Programmable Gate Arrays
Using Function Generators as RAM
RAM Configuration Options
Optional modes for each CLB make the memory look-up
tables in the F’ and G’ function generators usable as an
array of Read/Write memory cells. Available modes are
level-sensitive (similar to the XC4000/A/H families), edgetriggered, and dual-port edge-triggered. Depending on the
selected mode, a single CLB can be configured as either a
16x2, 32x1, or 16x1 bit array.
The function generators in any CLB can be configured as
RAM arrays in the following sizes:
Supported CLB memory configurations and timing modes
for single- and dual-port modes are shown in Table 5.
XC4000-Series devices are the first programmable logic
devices with edge-triggered (synchronous) and dual-port
RAM accessible to the user. Edge-triggered RAM simplifies system timing. Dual-port RAM doubles the effective
throughput of FIFO applications. These features can be
individually programmed in any XC4000-Series CLB.
•
•
One F or G function generator can be configured as a 16x1
RAM while the other function generators are used to implement any function of up to 5 inputs.
Additionally, the XC4000-Series RAM may have either of
two timing modes:
•
•
Advantages of On-Chip and Edge-Triggered RAM
The on-chip RAM is extremely fast. The read access time
is the same as the logic delay. The write access time is
slightly slower. Both access times are much faster than
any off-chip solution, because they avoid I/O delays.
Edge-triggered RAM, also called synchronous RAM, is a
feature never before available in a Field Programmable
Gate Array. The simplicity of designing with edge-triggered
RAM, and the markedly higher achievable performance,
add up to a significant improvement over existing devices
with on-chip RAM.
Three application notes are available from Xilinx that discuss edge-triggered RAM: “XC4000E Edge-Triggered and
Dual-Port RAM Capability,” “Implementing FIFOs in
XC4000E RAM,” and “Synchronous and Asynchronous
FIFO Designs.” All three application notes apply to both
XC4000E and XC4000EX RAM.
Table 5: Supported RAM Modes
Single-Port
Dual-Port
16
x
1
√
√
16
x
2
√
32
x
1
√
EdgeTriggered
Timing
√
√
LevelSensitive
Timing
√
Two 16x1 RAMs: two data inputs and two data outputs
with identical or, if preferred, different addressing for
each RAM
One 32x1 RAM: one data input and one data output.
Edge-Triggered (Synchronous): data written by the
designated edge of the CLB clock. WE acts as a true
clock enable.
Level-Sensitive (Asynchronous): an external WE signal
acts as the write strobe.
The selected timing mode applies to both function generators within a CLB when both are configured as RAM.
The number of read ports is also programmable:
•
•
Single Port: each function generator has a common
read and write port
Dual Port: both function generators are configured
together as a single 16x1 dual-port RAM with one write
port and two read ports. Simultaneous read and write
operations to the same or different addresses are
supported.
RAM configuration options are selected by placing the
appropriate library symbol.
Choosing a RAM Configuration Mode
The appropriate choice of RAM mode for a given design
should be based on timing and resource requirements,
desired functionality, and the simplicity of the design process. Recommended usage is shown in Table 6.
The difference between level-sensitive, edge-triggered,
and dual-port RAM is only in the write operation. Read
operation and timing is identical for all modes of operation.
Table 6: RAM Mode Selection
Use for New
Designs?
Size (16x1,
Registered)
Simultaneous
Read/Write
Relative
Performance
4-14
LevelSensitive
EdgeTriggered
Dual-Port
EdgeTriggered
No
Yes
Yes
1/2 CLB
1/2 CLB
1 CLB
No
No
Yes
X
2X
2X (4X
effective)
September 18, 1996 (Version 1.04)
C1 • • • C4
4
D1
WE
EC
D0
DIN
WRITE
DECODER
4
G1 • • • G4
16-LATCH
ARRAY
MUX
G'
4
1 of 16
LATCH
ENABLE
READ
ADDRESS
WRITE PULSE
DIN
WRITE
DECODER
4
F1 • • • F4
16-LATCH
ARRAY
MUX
F'
4
1 of 16
LATCH
ENABLE
K
(CLOCK)
WRITE PULSE
READ
ADDRESS
X6752
Figure 3:
16x2 (or 16x1) Edge-Triggered Single-Port RAM
C1 • • • C4
4
EC
WE
D1/A4
EC
D0
DIN
WRITE
DECODER
G1 • • • G4
F1 • • • F4
4
16-LATCH
ARRAY
MUX
G'
4
1 of 16
LATCH
ENABLE
WRITE PULSE
READ
ADDRESS
H'
DIN
WRITE
DECODER
4
16-LATCH
ARRAY
MUX
F'
4
1 of 16
K
(CLOCK)
LATCH
ENABLE
WRITE PULSE
READ
ADDRESS
X6754
Figure 4: 32x1 Edge-Triggered Single-Port RAM (F and G addresses are identical)
September 18, 1996 (Version 1.04)
4-15
XC4000 Series Field Programmable Gate Arrays
RAM Inputs and Outputs
TWPS
The F1-F4 and G1-G4 inputs to the function generators act
as address lines, selecting a particular memory cell in each
look-up table.
WCLK (K)
The functionality of the CLB control signals changes when
the function generators are configured as RAM. The DIN/
H2, H1, and SR/H0 lines become the two data inputs (D0,
D1) and the Write Enable (WE) input for the 16x2 memory.
When the 32x1 configuration is selected, D1 acts as the
fifth address bit and D0 is the data input.
The contents of the memory cell(s) being addressed are
available at the F’ and G’ function-generator outputs. They
can exit the CLB through its X and Y outputs, or can be captured in the CLB flip-flop(s).
Configuring the CLB function generators as Read/Write
memory does not affect the functionality of the other portions of the CLB, with the exception of the redefinition of the
control signals. In 16x2 and 16x1 modes, the H’ function
generator can be used to implement Boolean functions of
F’, G’, and D1, and the D flip-flops can latch the F’, G’, H’, or
D0 signals.
Single-Port Edge-Triggered Mode
Edge-triggered (synchronous) RAM simplifies timing
requirements. XC4000-Series edge-triggered RAM timing
operates like writing to a data register. Data and address
are presented. The register is enabled for writing by a logic
High on the write enable input, WE. Then a rising or falling
clock edge loads the data into the register, as shown in
Figure 5.
Complex timing relationships between address, data, and
write enable signals are not required, and the external write
enable pulse becomes a simple clock enable. The active
edge of WCLK latches the address, input data, and WE signals. An internal write pulse is generated that performs the
write. See Figure 3 and Figure 4 for block diagrams of a
CLB configured as 16x2 and 32x1 edge-triggered, singleport RAM.
The relationships between CLB pins and RAM inputs and
outputs for single-port, edge-triggered mode are shown in
Table 7.
The Write Clock input (WCLK) can be configured as active
on either the rising edge (default) or the falling edge. It
uses the same CLB pin (K) used to clock the CLB flip-flops,
but it can be independently inverted. Consequently, the
RAM output can optionally be registered within the same
4-16
TWHS
TWSS
WE
TDSS
TDHS
TASS
TAHS
DATA IN
ADDRESS
TILO
DATA OUT
TWOS
OLD
TILO
NEW
X6461
Figure 5:
Edge-Triggered RAM Write Timing
CLB either by the same clock edge as the RAM, or by the
opposite edge of this clock. The sense of WCLK applies to
both function generators in the CLB when both are configured as RAM.
The WE pin is active-High and is not invertible within the
CLB.
Note: The pulse following the active edge of WCLK (TWPS
in Figure 5) must be less than one millisecond wide. For
most applications, this requirement is not overly restrictive;
however, it must not be forgotten. Stopping WCLK at this
point in the write cycle could result in excessive current and
even damage to the larger devices if many CLBs are configured as edge-triggered RAM.
Table 7: Single-Port Edge-Triggered RAM Signals
RAM Signal
D
A[3:0]
A[4]
WE
WCLK
SPO
(Data Out)
CLB Pin
D0 or D1
(16x2, 16x1)
D0 (32x1)
F1-F4 or
G1-G4
D1 (32x1)
WE
K
F’ or G’
Function
Data In
Address
Address
Write Enable
Clock
Single Port Out
(Data Out)
September 18, 1996 (Version 1.04)
Dual-Port Edge-Triggered Mode
In dual-port mode, both the F and G function generators
are used to create a single 16x1 RAM array with one write
port and two read ports. The resulting RAM array can be
read and written simultaneously at two independent
addresses. Simultaneous read and write operations at the
same address are also supported.
Dual-port mode always has edge-triggered write timing, as
shown in Figure 5.
Figure 6 shows a simple model of an XC4000-Series CLB
configured as dual-port RAM. One address port, labeled
A[3:0], supplies both the read and write address for the F
function generator. This function generator behaves the
same as a 16x1 single-port edge-triggered RAM array. The
RAM output, Single Port Out (SPO), appears at the F function generator output. SPO, therefore, reflects the data at
address A[3:0].
The other address port, labeled DPRA[3:0] for Dual Port
Read Address, supplies the read address for the G function
generator. The write address for the G function generator,
however, comes from the address A[3:0]. The output from
this 16x1 RAM array, Dual Port Out (DPO), appears at the
G function generator output. DPO, therefore, reflects the
data at address DPRA[3:0].
RAM16X1D Primitive
Therefore, by using A[3:0] for the write address and
DPRA[3:0] for the read address, and reading only the DPO
output, a FIFO that can read and write simultaneously is
easily generated. Simultaneous access doubles the effective throughput of the FIFO.
The relationships between CLB pins and RAM inputs and
outputs for dual-port, edge-triggered mode are shown in
Table 8. See Figure 7 for a block diagram of a CLB configured in this mode.
Note: The pulse following the active edge of WCLK (TWPS
in Figure 5) must be less than one millisecond wide. For
most applications, this requirement is not overly restrictive;
however, it must not be forgotten. Stopping WCLK at this
point in the write cycle could result in excessive current and
even damage to the larger devices if many CLBs are configured as edge-triggered RAM.
Table 8: Dual-Port Edge-Triggered RAM Signals
RAM Signal
CLB Pin
D
D0
A[3:0]
F1-F4
DPRA[3:0]
WE
WCLK
SPO
G1-G4
WE
K
F’
DPO
G’
DPO (Dual Port Out)
WE
D
DPRA[3:0]
WE
D
D
Q
Registered DPO
AR[3:0]
AW[3:0]
Function
Data In
Read Address for F,
Write Address for F and G
Read Address for G
Write Enable
Clock
Single Port Out
(addressed by A[3:0])
Dual Port Out
(addressed by
DPRA[3:0])
G Function Generator
SPO (Single Port Out)
WE
D
A[3:0]
D
Q
Registered SPO
AR[3:0]
AW[3:0]
F Function Generator
WCLK
X6755
Figure 6: XC4000-Series Dual-Port RAM, Simple
Model
September 18, 1996 (Version 1.04)
4-17
XC4000 Series Field Programmable Gate Arrays
C1 • • • C4
4
D1
WE
EC
D0
DIN
WRITE
DECODER
16-LATCH
ARRAY
G'
MUX
4
1 of 16
LATCH
ENABLE
G1 • • • G4
WRITE PULSE
4
READ
ADDRESS
DIN
WRITE
DECODER
F1 • • • F4
4
K
(CLOCK)
16-LATCH
ARRAY
F'
MUX
4
1 of 16
LATCH
ENABLE
WRITE PULSE
READ
ADDRESS
X6748
Figure 7: 16x1 Edge-Triggered Dual-Port RAM
4-18
September 18, 1996 (Version 1.04)
Single-Port Level-Sensitive Timing Mode
Note: Edge-triggered mode is recommended for all new
designs. Level-sensitive mode, also called asynchronous
mode, is still supported for XC4000-Series backward-compatibility with the XC4000 family.
Level-sensitive RAM timing is simple in concept but can be
complicated in execution. Data and address signals are
presented, then a positive pulse on the write enable pin
(WE) performs a write into the RAM at the designated
address. As indicated by the “level-sensitive” label, this
RAM acts like a latch. During the WE High pulse, changing
the data lines results in new data written to the old address.
Changing the address lines while WE is High results in spurious data written to the new address—and possibly at
other addresses as well, as the address lines inevitably do
not all change simultaneously.
The user must generate a carefully timed WE signal. The
delay on the WE signal and the address lines must be carefully verified to ensure that WE does not become active until
after the address lines have settled, and that WE goes inactive before the address lines change again. The data must
be stable before and after the falling edge of WE.
In practical terms, WE is usually generated by a 2X clock. If
a 2X clock is not available, the falling edge of the system
clock can be used. However, there are inherent risks in this
approach, since the WE pulse must be guaranteed inactive
before the next rising edge of the system clock. Several
older application notes are available from Xilinx that discuss the design of level-sensitive RAMs. These application
notes include XAPP031, “Using the XC4000 RAM Capability,” and XAPP042, “High-Speed RAM Design in XC4000.”
However, the edge-triggered RAM available in the XC4000
Series is superior to level-sensitive RAM for almost every
application.
Figure 8 shows the write timing for level-sensitive, singleport RAM.
The relationships between CLB pins and RAM inputs and
outputs for single-port level-sensitive mode are shown in
Table 9.
Figure 9 and Figure 10 show block diagrams of a CLB configured as 16x2 and 32x1 level-sensitive, single-port RAM.
Initializing RAM at Configuration
Both RAM and ROM implementations of the XC4000Series devices are initialized during configuration. The initial contents are defined via an INIT attribute or property
attached to the RAM or ROM symbol, as described in the
schematic library guide.
If not defined, all RAM contents are initialized to all zeros,
by default.
RAM initialization occurs only during configuration. The
RAM content is not affected by Global Set/Reset.
Table 9: Single-Port Level-Sensitive RAM Signals
RAM Signal
D
A[3:0]
WE
O
CLB Pin
D0 or D1
F1-F4 or
G1-G4
WE
F’ or G’
Function
Data In
Address
Write Enable
Data Out
T WC
ADDRESS
TAS
T WP
T AH
WRITE ENABLE
T DS
DATA IN
T DH
REQUIRED
X6462
Figure 8: Level-Sensitive RAM Write Timing
September 18, 1996 (Version 1.04)
4-19
XC4000 Series Field Programmable Gate Arrays
C1 • • • C4
4
WE
D1
EC
D0
DIN
Enable
16-LATCH
ARRAY
WRITE
DECODER
4
G1 • • • G4
G'
MUX
1 of 16
4
READ ADDRESS
DIN
Enable
WRITE
DECODER
4
F1 • • • F4
16-LATCH
ARRAY
F'
MUX
1 of 16
4
READ ADDRESS
X6746
Figure 9: 16x2 (or 16x1) Level-Sensitive Single-Port RAM
C1 • • • C4
4
WE
D1/A4
D0
EC
DIN
Enable
G1 • • • G4
F1 • • • F4
4
WRITE
DECODER
16-LATCH
ARRAY
MUX
G'
1 of 16
4
READ ADDRESS
H'
DIN
Enable
4
WRITE
DECODER
16-LATCH
ARRAY
MUX
F'
1 of 16
4
READ ADDRESS
X6749
Figure 10: 32x1 Level-Sensitive Single-Port RAM (F and G addresses are identical)
4-20
September 18, 1996 (Version 1.04)
Fast Carry Logic
Each CLB F and G function generator contains dedicated
arithmetic logic for the fast generation of carry and borrow
signals. This extra output is passed on to the function generator in the adjacent CLB. The carry chain is independent
of normal routing resources.
Dedicated fast carry logic greatly increases the efficiency
and performance of adders, subtractors, accumulators,
comparators and counters. It also opens the door to many
new applications involving arithmetic operation, where the
previous generations of FPGAs were not fast enough or too
inefficient. High-speed address offset calculations in
microprocessor or graphics systems, and high-speed addition in digital signal processing are two typical applications.
The two 4-input function generators can be configured as a
2-bit adder with built-in hidden carry that can be expanded
to any length. This dedicated carry circuitry is so fast and
efficient that conventional speed-up methods like carry
generate/propagate are meaningless even at the 16-bit
level, and of marginal benefit at the 32-bit level.
The dedicated carry logic is discussed in detail in Xilinx
document XAPP 013: “Using the Dedicated Carry Logic in
XC4000.” This discussion also applies to XC4000E
devices, and to XC4000EX devices when the minor logic
changes are taken into account.
The fast carry logic can be accessed by placing special
library symbols, or by using Xilinx Relationally Placed Macros (RPMs) that already include these symbols.
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
This fast carry logic is one of the more significant features
of the XC4000 Series, speeding up arithmetic and counting
into the 70 MHz range.
The carry chain in XC4000E devices can run either up or
down. At the top and bottom of the columns where there
are no CLBs above and below, the carry is propagated to
the right. (See Figure 11.) In order to improve speed in the
high-capacity XC4000EX devices, which can potentially
have very long carry chains, the carry chain travels upward
only, as shown in Figure 12. This restriction should have little impact, because the smallest XC4000EX device, the
XC4028EX, can accommodate a 64-bit carry chain in a single column. Additionally, standard interconnect can be
used to route a carry signal in the downward direction.
Figure 13 on page 22 shows an XC4000E CLB with dedicated fast carry logic. The carry logic in the XC4000EX is
similar, except that COUT exits at the top only, and the signal CINDOWN does not exist. As shown in Figure 13, the
carry logic shares operand and control inputs with the function generators. The carry outputs connect to the function
generators, where they are combined with the operands to
form the sums.
Figure 14 and Figure 15 on page 23 show the details of the
carry logic for the XC4000E and the XC4000EX respectively. These diagrams show the contents of the box
labeled “CARRY LOGIC” in Figure 13. As shown, the
XC4000EX carry logic eliminated a multiplexer to reduce
delay on the pass-through carry chain. Additionally, the
multiplexer on the G4 path now has a memory-programmable input, which permits G4 to directly connect to COUT.
G4 thus becomes an additional high-speed initialization
path for carry-in.
September 18, 1996 (Version 1.04)
X6687
Figure 11: Available XC4000E Carry Propagation
Paths
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
X6610
Figure 12: Available XC4000EX Carry Propagation
Paths (dotted lines use general interconnect)
4-21
XC4000 Series Field Programmable Gate Arrays
CARRY
LOGIC
C OUT
D IN
C IN DOWN
G
Y
H
G
CARRY
G4
G3
G
DIN
G2
S/R
H
D
G
F
Q
YQ
Q
XQ
G1
EC
COUT0
H
H1
DIN
S/R
H
F
CARRY
D
G
F
EC
F4
F3
F
F2
H
F1
X
F
CIN UP
C OUT
K
S/R
EC
X6699
Figure 13: Fast Carry Logic in XC4000E CLB (shaded area not present in XC4000EX)
4-22
September 18, 1996 (Version 1.04)
C OUT
M
G1
M
1
0
1
G2
0
I
G4
G3
C OUT0
TO
FUNCTION
GENERATORS
M
F2
M
1
0
F1
M
0
1
0
M
F4
1
3
M
1
F3
M
0
M
M
M
1
0
C IN UP
X2000
C IN DOWN
Figure 14: Detail of XC4000E Dedicated Carry Logic
COUT
M
G1
M
1
0
M
1
G2
0
G4
G3
COUT0
TO
FUNCTION
GENERATORS
M
F2
M
1
0
F1
M
M
0
M
F3
M
1
0
F4
1
3
1
M
0
C IN UP
M
X6701
Figure 15: Detail of XC4000EX Dedicated Carry Logic (shaded areas show differences from XC4000E carry logic)
September 18, 1996 (Version 1.04)
4-23
XC4000 Series Field Programmable Gate Arrays
Input/Output Blocks (IOBs)
User-configurable input/output blocks (IOBs) provide the
interface between external package pins and the internal
logic. Each IOB controls one package pin and can be configured for input, output, or bidirectional signals.
Figure 16 shows a simplified block diagram of the
XC4000E IOB. A more complete diagram of the XC4000E
IOB can be found in Figure 42 on page 51, in the “Boundary
Scan” section. Figure 42 includes the boundary scan logic
in the IOB.
Figure 17 shows a simplified block diagram of the
XC4000EX IOB. The XC4000EX IOB contains some special features not included in the XC4000E IOB. These features are highlighted in Figure 17, and discussed
throughout this section. When XC4000EX special features
are discussed, they are clearly identified in the text. Any
feature not so identified is present in both XC4000E and
XC4000EX devices.
IOB Input Signals
Two paths, labeled I1 and I2 in Figure 16 and Figure 17,
bring input signals into the array. Inputs also connect to an
input register that can be programmed as either an edgetriggered flip-flop or a level-sensitive latch.
The choice is made by placing the appropriate library symbol. For example, IFD is the basic input flip-flop (rising
edge triggered), and ILD is the basic input latch (transparent-High). Variations with inverted clocks are available, and
some combinations of latches and flip-flops can be implemented in a single IOB, as described in the XACT Libraries
Guide.
The inputs can be globally configured for either TTL (1.2V,
default) or CMOS thresholds, using an option in the MakeBits program. There is a slight hysteresis of about 300mV.
The output levels are also configurable; the two global
adjustments of input threshold and output level are independent.
Inputs of the low-voltage devices must be configured as
CMOS at all times. They can be driven by the outputs of all
5-Volt XC4000-Series devices, provided that the 5-Volt outputs are in TTL mode. They can also be driven by any TTL
output that does not exceed 3.7 V. 5-Volt XC3000-family
device outputs, for example, are TTL-compatible, but since
the output voltage can exceed 3.7 V, they cannot be used to
drive an XC4000L or XC4000XL input.
Table 10: Supported Sources for XC4000-Series Device
Inputs
XC4000-Series Inputs
3.3 V,
5 V,
5 V,
CMOS
TTL
CMOS
Source
Any device, Vcc = 3.3 V,
CMOS outputs
XC4000-Series, Vcc = 5 V,
TTL outputs
Any device, Vcc = 5 V,
TTL outputs (Voh ≤ 3.7 V)
Any device, Vcc = 5 V,
CMOS outputs
√
√
√
√
√
√
Danger1
√
Unreliable
Data
√
1. Acceptable for XC4000XL if the designated 5-Volt
supply pad (VTT) is tied to 5V.
Registered Inputs
The I1 and I2 signals that exit the block can each carry
either the direct or registered input signal.
The input and output storage elements in each IOB have a
common clock enable input, which, through configuration,
can be activated individually for the input or output flip-flop,
or both. This clock enable operates exactly like the EC pin
on the XC4000-Series CLB. It cannot be inverted within
the IOB.
The storage element behavior is shown in Table 11.
Table 11: Input Register Functionality
(active rising edge is shown)
Mode
Clock
Power-Up or
GSR
Flip-Flop
X
Clock
Enable
X
__/
0
1
0
X
1*
X
1*
1*
0
Latch
Both
Legend:
X
__/
SR
0*
1*
D
Q
X
SR
D
X
X
D
X
D
Q
Q
D
Q
Don’t care
Rising edge
Set or Reset value. Reset is default.
Input is Low or unconnected (default value)
Input is High or unconnected (default value)
The inputs of XC4000-Series 5-Volt devices can be driven
by the outputs of any 3.3-Volt device, if the 5-Volt inputs are
in TTL mode.
Supported sources for XC4000-Series device inputs are
shown in Table 10.
4-24
September 18, 1996 (Version 1.04)
Passive
Pull-Up/
Pull-Down
Slew Rate
Control
T
Flip-Flop
D
Q
Out
Output
Buffer
CE
Pad
Output
Clock
I1
I2
Clock
Enable
Input
Buffer
FlipFlop/
Latch
Q
D
Delay
CE
Input
Clock
X6704
Figure 16: Simplified Block Diagram of XC4000E IOB
Slew Rate
Control
T
Passive
Pull-Up/
Pull-Down
Output MUX
0
1
Flip-Flop
Out
D
Q
Output
Buffer
CE
Pad
Input
Buffer
Output Clock
I1
I2
Clock Enable
Flip-Flop/
Latch
Q
CE
Delay
D
Fast
Capture
Latch
Delay
Q D
Latch
G
Input Clock
X5984
Figure 17: Simplified Block Diagram of XC4000EX IOB (shaded areas indicate differences from XC4000E)
September 18, 1996 (Version 1.04)
4-25
XC4000 Series Field Programmable Gate Arrays
Optional Delay Guarantees Zero Hold Time
The data input to the register can optionally be delayed by
several nanoseconds. With the delay enabled, the setup
time of the input flip-flop is increased so that normal clock
routing does not result in a positive hold-time requirement.
A positive hold time requirement can lead to unreliable,
temperature- or processing-dependent operation.
The input flip-flop setup time is defined between the data
measured at the device I/O pin and the clock input at the
IOB (not at the clock pin). Any routing delay from the
device clock pin to the clock input of the IOB must, therefore, be subtracted from this setup time to arrive at the real
setup time requirement relative to the device pins. A short
specified setup time might, therefore, result in a negative
setup time at the device pins, i.e., a positive hold-time
requirement.
When a delay is inserted on the data line, more clock delay
can be tolerated without causing a positive hold-time
requirement. Sufficient delay eliminates the possibility of a
data hold-time requirement at the external pin. The maximum delay is therefore inserted as the default.
The XC4000E IOB has a one-tap delay element: either the
delay is inserted (default), or it is not. The delay guarantees
a zero hold time with respect to clocks routed through any
of the XC4000E global clock buffers. (See “Global Nets
and Buffers (XC4000E only)” on page 41 for a description
of the global clock buffers in the XC4000E.) For a shorter
input register setup time, with non-zero hold, attach a
NODELAY attribute or property to the flip-flop.
The XC4000EX IOB has a two-tap delay element, with
choices of a full delay, a partial delay, or no delay. The
attributes or properties used to select the desired delay are
shown in Table 12. The choices are no added attribute,
MEDDELAY, and NODELAY. The default setting, with no
added attribute, ensures no hold time with respect to any of
the XC4000EX clock buffers, including the Global LowSkew buffers. MEDDELAY ensures no hold time with
respect to the Global Early and FastCLK buffers. Inputs
with NODELAY may have a positive hold time with respect
to all clock buffers, including the FastCLK buffers. For a
description of each of these buffers, see “Global Nets and
Buffers (XC4000EX only)” on page 43.
Additional Input Latch for Fast Capture (XC4000EX
only)
The XC4000EX IOB has an additional optional latch on the
input. This latch, as shown in Figure 17, is clocked by the
output clock — the clock used for the output flip-flop —
rather than the input clock. Therefore, two different clocks
can be used to clock the two input storage elements. This
additional latch allows the very fast capture of input data,
which is then synchronized to the internal clock by the IOB
flip-flop or latch.
To use this Fast Capture technique, drive the output clock
pin (the Fast Capture latching signal) from the output of one
of the Global Early or FastCLK buffers supplied in the
XC4000EX. The second storage element should be
clocked by a Global Low-Skew buffer, to synchronize the
incoming data to the internal logic. (See Figure 18.) These
special buffers are described in “Global Nets and Buffers
(XC4000EX only)” on page 43.
The Fast Capture latch is designed primarily for use with a
Global Early buffer. For Fast Capture, a single clock signal
is routed through both a Global Early buffer and a Global
Low-Skew buffer. (The two buffers share an input pad.)
The Fast Capture latch is clocked by the Global Early
buffer, and the standard IOB flip-flop or latch is clocked by
the Global Low-Skew buffer. This mode is the safest way to
use the Fast Capture latch, because the clock buffers on
both storage elements are driven by the same pad. There
is no external skew between clock pads to create potential
problems.
Alternatively, a FastCLK buffer can be used to minimize the
setup time of device inputs, if a positive hold time is acceptable. Use the FastCLK buffer to clock the Fast Capture
latch, and a slower clock buffer to clock the standard IOB
flip-flop or latch. Either the Global Early buffer or the Global
Low-Skew buffer can be used for the second storage ele-
ILFFX
IPAD
NODELAY
When to Use
Zero Hold with respect to Global LowSkew Buffer, Global Early Buffer, or
FastCLK Buffer
Zero Hold with respect to Global Early
Buffer or FastCLK Buffer
Short Setup, positive Hold time
Q
to internal
logic
Q
to internal
logic
GF
BUFGE
CE
C
IPAD
BUFGLS
Table 12: XC4000EX IOB Input Delay Element
Value
full delay
(default, no
attribute added)
MEDDELAY
D
ILFFX
IPAD
D
GF
IPAD
BUFFCLK
CE
C
NODELAY
IPAD
BUFGLS
X6705
Figure 18: Examples Using XC4000EX Fast
Capture Latch
4-26
September 18, 1996 (Version 1.04)
ment, but whichever one is used should be the same clock
as the related internal logic. Since the FastCLK pads are
different from the Global Early and Global Low-Skew pads,
care must be taken to ensure that skew external to the
device does not create internal timing difficulties.
To place the Fast Capture latch in a design, use one of the
special library symbols, ILFFX or ILFLX. ILFFX is a transparent-Low Fast Capture latch followed by an active-High
input flip-flop. ILFLX is a transparent-Low Fast Capture
latch followed by a transparent-High input latch. Any of the
clock inputs can be inverted before driving the library element, and the inverter is absorbed into the IOB. If a single
BUFG output is used to drive both clock inputs, the software automatically runs the clock through both a Global
Low-Skew buffer and a Global Early buffer, and clocks the
Fast Capture latch appropriately.
Figure 17 on page 25 also shows a two-tap delay on the
input. By default, if the Fast Capture latch is used, the Xilinx
software assumes a Global Early buffer is driving the clock,
and selects MEDDELAY to ensure a zero hold time. This
default can be overridden to remove the delay, if FastClk is
used, by attaching a NODELAY attribute or property to the
ILFFX or ILFLX latch. Select the desired delay based on
the discussion in the previous subsection.
IOB Output Signals
Output signals can be optionally inverted within the IOB,
and can pass directly to the pad or be stored in an edgetriggered flip-flop. The functionality of this flip-flop is shown
in Table 13.
An active-High 3-state signal can be used to place the output buffer in a high-impedance state, implementing 3-state
outputs or bidirectional I/O. Under configuration control,
the output (OUT) and output 3-state (T) signals can be
inverted. The polarity of these signals is independently
configured for each IOB.
The 4-mA maximum output current specification of many
FPGAs often forces the user to add external buffers, which
are especially cumbersome on bidirectional I/O lines. The
XC4000E and XC4000EX devices solve many of these
problems by providing a guaranteed output sink current of
12 mA. Two adjacent outputs can be interconnected externally to sink up to 24 mA. (XC4000L and XC4000XL outputs can sink up to 4 mA, and two adjacent XC4000L and
XC4000XL outputs can sink up to 8 mA.) The XC4000E
and XC4000EX FPGAs can thus directly drive buses on a
printed circuit board.
Table 13: Output Flip-Flop Functionality (active rising
edge is shown)
Mode
Power-Up
or GSR
Flip-Flop
Legend:
X
__/
SR
0*
1*
Z
Clock
X
Clock
Enable
X
T
0*
D
X
Q
SR
X
__/
X
0
0
1*
X
X
0*
0*
1
0*
X
D
X
X
Q
D
Z
Q
Don’t care
Rising edge
Set or Reset value. Reset is default.
Input is Low or unconnected (default value)
Input is High or unconnected (default value)
3-state
By default, the output pull-up structure is configured as a
TTL-like totem-pole. The High driver is an n-channel pullup transistor, pulling to a voltage one transistor threshold
below Vcc. Alternatively, the outputs can be globally configured as CMOS drivers, with p-channel pull-up transistors
pulling to Vcc. This MakeBits option applies to all outputs
on the device. It is not individually programmable.
Outputs of low-voltage devices must be configured as
CMOS at all times. They can drive the inputs of any 5-Volt
device with TTL-compatible thresholds.
Any XC4000-Series 5-Volt device with its outputs configured in TTL mode can drive the inputs of any typical 3.3Volt device. (For a detailed discussion of how to interface
between 5 V and 3.3 V devices, see the 3V Products section of The Programmable Logic Data Book.)
Supported destinations for XC4000-Series device outputs
are shown in Table 14.
Table 14: Supported Destinations for XC4000-Series
Outputs
XC4000-Series
Outputs
Destination
3.3 V,
5 V,
5 V,
CMOS TTL CMOS
Any typical device, Vcc = 3.3 V,
√
√
some1
CMOS-threshold inputs
Any device, Vcc = 5 V,
√
√
√
TTL-threshold inputs
Unreliable
√
Any device, Vcc = 5 V,
Data
CMOS-threshold inputs
1. Only if destination device has 5-V tolerant inputs
September 18, 1996 (Version 1.04)
4-27
XC4000 Series Field Programmable Gate Arrays
OPAD
OBUFT
X6702
Figure 19: Open-Drain Output
An output can be configured as open-drain (open-collector)
by placing an OBUFT symbol in a schematic or HDL code,
then tying the 3-state pin (T) to the output signal, and the
input pin (I) to Ground. (See Figure 19.)
Output Slew Rate
The slew rate of each output buffer is, by default, reduced,
to minimize power bus transients when switching non-critical signals. For critical signals, attach a FAST attribute or
property to the output buffer or flip-flop.
For XC4000E devices, maximum total capacitive load for
simultaneous fast mode switching in the same direction is
200 pF for all package pins between each Power/Ground
pin pair. For XC4000EX devices, additional internal Power/
Ground pin pairs are connected to special Power and
Ground planes within the packages, to reduce ground
bounce. Therefore, the maximum total capacitive load is
300 pF between each external Power/Ground pin pair.
Maximum loading may vary for the low-voltage devices.
For slew-rate limited outputs this total is two times larger for
each device type: 400 pF for XC4000E devices and 600 pF
for XC4000EX devices. This maximum capacitive load
should not be exceeded, as it can result in ground bounce
of greater than 1.5 V amplitude and more than 5 ns duration. This level of ground bounce may cause undesired
transient behavior on an output, or in the internal logic.
This restriction is common to all high-speed digital ICs, and
is not particular to Xilinx or the XC4000 Series.
XC4000-Series devices have a feature called “Soft Startup,” designed to reduce ground bounce when all outputs
are turned on simultaneously at the end of configuration.
When the configuration process is finished and the device
starts up, the first activation of the outputs is automatically
slew-rate limited. Immediately following the initial activation
of the I/O, the slew rate of the individual outputs is determined by the individual configuration option for each IOB.
Global Three-State
A separate Global 3-State line (not shown in Figure 16 or
Figure 17) forces all FPGA outputs to the high-impedance
state, unless boundary scan is enabled and is executing an
EXTEST instruction. This global net (GTS) does not compete with other routing resources; it uses a dedicated distribution network.
4-28
GTS can be driven from any user-programmable pin as a
global 3-state input. To use this global net, place an input
pad and input buffer in the schematic or HDL code, driving
the GTS pin of the STARTUP symbol. A specific pin location can be assigned to this input using a LOC attribute or
property, just as with any other user-programmable pad.
An inverter can optionally be inserted after the input buffer
to invert the sense of the Global 3-State signal. Using GTS
is similar to GSR. See Figure 2 on page 13 for details.
Alternatively, GTS can be driven from any internal node.
Output Multiplexer/2-Input Function Generator
(XC4000EX only)
As shown in Figure 17 on page 25, the output path in the
XC4000EX IOB contains an additional multiplexer not available in the XC4000E IOB. The multiplexer can also be configured as a 2-input function generator, implementing a
pass-gate, AND-gate, OR-gate, or XOR-gate, with 0, 1, or 2
inverted inputs. The logic used to implement these functions is shown in the upper gray area of Figure 17.
When configured as a multiplexer, this feature allows two
output signals to time-share the same output pad; effectively doubling the number of device outputs without requiring a larger, more expensive package.
When the MUX is configured as a 2-input function generator, logic can be implemented within the IOB itself. Combined with either a FastCLK or Global Early buffer, this
arrangement allows very high-speed gating of a single signal. For example, a wide decoder can be implemented in
CLBs, and its output gated with a Read or Write Strobe
driven by a FastCLK buffer, as shown in Figure 20. The
critical-path pin-to-pin delay of this circuit is less than 6
nanoseconds. (This value may not be achievable in
XC4000XL devices.)
As shown in Figure 17, the IOB input pins Out, Output
Clock, and Clock Enable have different delays and different
flexibilities regarding polarity. Additionally, Output Clock
sources are more limited than the other inputs. Therefore,
the Xilinx software does not move logic into the IOB function generators unless explicitly directed to do so.
IPAD
BUFFCLK
F
from
internal
logic
OAND2
OPAD
FAST
X6698
Figure 20: Fast Pin-to-Pin Path in XC4000E
September 18, 1996 (Version 1.04)
D0
are independent, except that in the XC4000EX, the Fast
Capture latch shares an IOB input with the output clock pin.
OMUX2
O
F
Early Clock for IOBs (XC4000EX only)
D1
OAND2
X6598
S0
X6599
Figure 21: Output AND and MUX Symbols in
XC4000EX IOB
The user can specify that the IOB function generator be
used, by placing special library symbols beginning with the
letter “O.” For example, a 2-input AND-gate in the IOB function generator is called OAND2. Use the symbol input pin
labelled “F” for the signal on the critical path. This signal is
placed on the OK pin — the IOB input with the shortest
delay to the function generator. Two examples are shown in
Figure 21.
Other IOB Options
There are a number of other programmable options in the
XC4000-Series IOB.
Pull-up and Pull-down Resistors
Programmable pull-up and pull-down resistors are useful
for tying unused pins to Vcc or Ground to minimize power
consumption and reduce noise sensitivity. The configurable
pull-up resistor is a p-channel transistor that pulls to Vcc.
The configurable pull-down resistor is an n-channel transistor that pulls to Ground.
The value of these resistors is 50 kΩ − 100 kΩ. This high
value makes them unsuitable as wired-AND pull-up resistors.
The pull-up resistors for most user-programmable IOBs are
active during the configuration process. See Table 24 on
page 78 for a list of pins with pull-ups active before and during configuration.
After configuration, voltage levels of unused pads, bonded
or unbonded, must be valid logic levels, to reduce noise
sensitivity and avoid excess current. Therefore, by default,
unused pads are configured with the internal pull-up resistor active. Alternatively, they can be individually configured
with the pull-down resistor, or as a driven output, or to be
driven by an external source. To activate the internal pullup, attach the PULLUP library component to the net
attached to the pad. To activate the internal pull-down,
attach the PULLDOWN library component to the net
attached to the pad.
Independent Clocks
Separate clock signals are provided for the input and output
flip-flops. The clock can be independently inverted for each
flip-flop within the IOB, generating either falling-edge or rising-edge triggered flip-flops. The clock inputs for each IOB
September 18, 1996 (Version 1.04)
Special early clocks are available for IOBs. These clocks
are sourced by the same sources as the Global Low-Skew
buffers, but are separately buffered. They have fewer loads
and therefore less delay. The early clock can drive either
the IOB output clock or the IOB input clock, or both. The
early clock allows fast capture of input data, and fast clockto-output on output data. The Global Early buffers that
drive these clocks are described in “Global Nets and Buffers (XC4000EX only)” on page 43.
Fast Clock for IOBs (XC4000EX only)
Very fast clocks driven by FastCLK buffers are also available for IOBs. These clocks are sourced by semi-dedicated
pads—the pads can be used as general I/O if not used to
drive FastCLK buffers. There are two FastCLK buffers on
the left edge, and two on the right edge of the device. They
provide the fastest method of reaching the IOB clock pins.
The FastCLK buffer can drive either the IOB output clock or
the IOB input clock, or both. These buffers allow the fastest
possible setup times and clock-to-output times. The FastCLK buffers are described in “Global Nets and Buffers
(XC4000EX only)” on page 43.
Global Set/Reset
As with the CLB registers, the Global Set/Reset signal
(GSR) can be used to set or clear the input and output registers, depending on the value of the INIT attribute or property. The two flip-flops can be individually configured to set
or clear on reset and after configuration. Other than the
global GSR net, no user-controlled set/reset signal is available to the I/O flip-flops. The choice of set or clear applies
to both the initial state of the flip-flop and the response to
the Global Set/Reset pulse. See “Global Set/Reset” on
page 13 for a description of how to use GSR.
JTAG Support
Embedded logic attached to the IOBs contains test structures compatible with IEEE Standard 1149.1 for boundary
scan testing, permitting easy chip and board-level testing.
More information is provided in “Boundary Scan” on
page 50.
Three-State Buffers
A pair of 3-state buffers is associated with each CLB in the
array. (See Figure 27 on page 34.) These 3-state buffers
can be used to drive signals onto the nearest horizontal
longlines above and below the CLB. They can therefore be
used to implement multiplexed or bidirectional buses on the
horizontal longlines, saving logic resources. Programmable pull-up resistors attached to these longlines help to
implement a wide wired-AND function.
4-29
XC4000 Series Field Programmable Gate Arrays
The buffer enable is an active-High 3-state (i.e. an activeLow enable), as shown in Table 15.
WAND4, WAND8, and WAND16 are also available. See
the XACT Libraries Guide for further information.
Another 3-state buffer with similar access is located near
each I/O block along the right and left edges of the array.
(See Figure 33 on page 39.)
The T pin is internally tied to the I pin. Connect the input to
the I pin and the output to the O pin. Connect the outputs of
all the WAND1s together and attach a PULLUP symbol.
The horizontal longlines driven by the 3-state buffers have a
weak keeper at each end. This circuit prevents undefined
floating levels. However, it is overridden by any driver, even
a pull-up resistor.
Wired OR-AND
The buffer can be configured as a Wired OR-AND. A High
level on either input turns off the output. Use the
WOR2AND library symbol, which is essentially an opendrain 2-input OR gate. The two input pins are functionally
equivalent. Attach the two inputs to the I0 and I1 pins and
tie the output to the O pin. Tie the outputs of all the
WOR2ANDs together and attach a PULLUP symbol.
Special longlines running along the perimeter of the array
can be used to wire-AND signals coming from nearby IOBs
or from internal longlines. These longlines form the wide
edge decoders discussed in “Wide Edge Decoders” on
page 31.
Three-State Buffer Examples
Three-State Buffer Modes
Figure 22 shows how to use the 3-state buffers to implement a wired-AND function. When all the buffer inputs are
High, the pull-up resistor(s) provide the High output.
The 3-state buffers can be configured in three modes:
•
•
•
Standard 3-state buffer
Wired-AND with input on the I pin
Wired OR-AND
Figure 23 shows how to use the 3-state buffers to implement a multiplexer. The selection is accomplished by the
buffer 3-state signal.
Standard 3-State Buffer
Pay particular attention to the polarity of the T pin when
using these buffers in a design. Active-High 3-state (T) is
identical to an active-Low output enable, as shown in
Table 15.
All three pins are used. Place the library element BUFT.
Connect the input to the I pin and the output to the O pin.
The T pin is an active-High 3-state (i.e. an active-Low
enable). Tie the T pin to Ground to implement a standard
buffer.
Table 15: Three-State Buffer Functionality
IN
X
IN
Wired-AND with Input on the I Pin
The buffer can be used as a Wired-AND. Use the WAND1
library symbol, which is essentially an open-drain buffer.
Z=D
D
A
D
WAND1
A
●D
B
● (D
C
T
1
0
P
U
L
L
+D ) ● (D +D )
D
E
F
D
C
D
B
WAND1
OUT
Z
IN
U
P
D
E
D
D
F
W0R2AND
W0R2AND
X6465
Figure 22: Open-Drain Buffers Implement a Wired-AND Function
Z = DA • A + D B • B + D C • C + D N • N
~100 kΩ
DA
DB
BUFT
A
DC
BUFT
B
DN
BUFT
C
BUFT
N
X6466
"Weak Keeper"
Figure 23: 3-State Buffers Implement a Multiplexer
4-30
September 18, 1996 (Version 1.04)
Wide Edge Decoders
INTERCONNECT
Dedicated decoder circuitry boosts the performance of
wide decoding functions. When the address or data field is
wider than the function generator inputs, FPGAs need
multi-level decoding and are thus slower than PALs.
XC4000-Series CLBs have nine inputs. Any decoder of up
to nine inputs is, therefore, compact and fast. However,
there is also a need for much wider decoders, especially for
address decoding in large microprocessor systems.
IOB
IOB
.I1
A
.I1
C
B
(
An XC4000-Series FPGA has four programmable decoders located on each edge of the device. The inputs to each
decoder are any of the IOB I1 signals on that edge plus one
local interconnect per CLB row or column. Each row or column of CLBs provides up to three variables or their compliments., as shown in Figure 24. Each decoder generates a
High output (resistor pull-up) when the AND condition of
the selected inputs, or their complements, is true. This is
analogous to a product term in typical PAL devices.
Each of these wired-AND gates is capable of accepting up
to 42 inputs on the XC4005E and 72 on the XC4013E.
There are up to 96 inputs for each decoder on the
XC4028EX and 132 on the XC4052EX. The decoders may
also be split in two when a larger number of narrower
decoders are required, for a maximum of 32 decoders per
device.
The decoder outputs can drive CLB inputs, so they can be
combined with other logic to form a PAL-like AND/OR structure. The decoder outputs can also be routed directly to the
chip outputs. For fastest speed, the output should be on
the same chip edge as the decoder. Very large PALs can
be emulated by ORing the decoder outputs in a CLB. This
decoding feature covers what has long been considered a
weakness of older FPGAs. Users often resorted to external PALs for simple but fast decoding functions. Now, the
dedicated decoders in the XC4000-Series device can
implement these functions fast and efficiently.
To use the wide edge decoders, place one or more of the
WAND library symbols (WAND1, WAND4, WAND8,
WAND16). Attach a DECODE attribute or property to each
WAND symbol. Tie the outputs together and attach a PULLUP symbol. Location attributes or properties such as L
(left edge) or TR (right half of top edge) should also be used
to ensure the correct placement of the decoder inputs.
C) .....
(A • B • C) .....
(A • B • C) .....
(A • B • C) .....
X2627
Figure 24: XC4000-Series Edge Decoding Example
On-Chip Oscillator
XC4000-Series devices include an internal oscillator. This
oscillator is used to clock the power-on time-out, for configuration memory clearing, and as the source of CCLK in
Master configuration modes. The oscillator runs at a nominal 8 MHz frequency that varies with process, Vcc, and
temperature. The output frequency falls between 4 and 10
MHz. (The oscillator operates more slowly at lower voltages. The output frequency may be reduced by as much
as 10% for low-voltage devices.)
The oscillator output is optionally available after configuration. Any two of four resynchronized taps of a built-in
divider are also available. These taps are at the fourth,
ninth, fourteenth and nineteenth bits of the divider. Therefore, if the primary oscillator output is running at the nominal 8 MHz, the user has access to an 8 MHz clock, plus any
two of 500 kHz, 16kHz, 490Hz and 15Hz (up to 10% lower
for low-voltage devices). These frequencies can vary by as
much as -50% or +25%.
These signals can be accessed by placing the OSC4
library element in a schematic or in HDL code (see
Figure 25).
The oscillator is automatically disabled after configuration if
the OSC4 symbol is not used in the design.
OSC4
F8M
F500K
F16K
F490
F15
X6703
Figure 25: XC4000-Series Oscillator Symbol
September 18, 1996 (Version 1.04)
4-31
XC4000 Series Field Programmable Gate Arrays
Programmable Interconnect
CLB Routing Connections
All internal connections are composed of metal segments
with programmable switching points and switching matrices
to implement the desired routing. A structured, hierarchical
matrix of routing resources is provided to achieve efficient
automated routing.
A high-level diagram of the routing resources associated
with one CLB is shown in Figure 26. The shaded arrows
represent routing present only in XC4000EX devices.
The XC4000E and XC4000EX share a basic interconnect
structure. XC4000EX devices, however, have additional
routing not available in the XC4000E. The extra routing
resources allow high utilization in high-capacity devices.
All XC4000EX-specific routing resources are clearly identified throughout this section. Any resources not identified
as XC4000EX-specific are present in all XC4000-Series
devices.
This section describes the varied routing resources available in XC4000-Series devices. The implementation software automatically assigns the appropriate resources
based on the density and timing requirements of the
design.
Interconnect Overview
There are several types of interconnect.
•
•
•
CLB routing is associated with each row and column of
the CLB array.
IOB routing forms a ring (called a VersaRing) around
the outside of the CLB array. It connects the I/O with
the internal logic blocks.
Global routing consists of dedicated networks primarily
designed to distribute clocks throughout the device with
minimum delay and skew. Global routing can also be
used for other high-fanout signals.
Five interconnect types are distinguished by the relative
length of their segments: single-length lines, double-length
lines, quad and octal lines (XC4000EX only), and longlines.
In the XC4000EX, direct connects allow fast data flow
between adjacent CLBs, and between IOBs and CLBs.
Extra routing is included in the IOB pad ring. The
XC4000EX also includes a ring of octal interconnect lines
near the IOBs to improve pin-swapping and routing to
locked pins.
Table 16 shows how much routing of each type is available
in XC4000E and XC4000EX CLB arrays. Clearly, very
large designs, or designs with a great deal of interconnect,
will route more easily in the XC4000EX. Smaller XC4000E
designs, typically requiring significantly less interconnect,
do not require the additional routing.
Figure 27 on page 34 is a detailed diagram of both the
XC4000E and the XC4000EX CLB, with associated routing. The shaded square is the programmable switch
matrix, present in both the XC4000E and the XC4000EX.
The L-shaped shaded area is present only in XC4000EX
devices. As shown in the figure, the XC4000EX block is
essentially an XC4000E block with additional routing.
CLB inputs and outputs are distributed on all four sides,
providing maximum routing flexibility. In general, the entire
architecture is symmetrical and regular. It is well suited to
established placement and routing algorithms. Inputs, outputs, and function generators can freely swap positions
within a CLB to avoid routing congestion during the placement and routing operation.
Table 16: Routing per CLB in XC4000-Series Devices
Singles
Doubles
Quads
Longlines
Direct
Connects
Globals
Carry Logic
Total
XC4000E
XC4000EX
Vertical Horizontal Vertical Horizontal
8
8
8
8
4
4
4
4
0
0
12
12
6
6
10
6
0
0
2
2
4
2
24
0
0
18
8
1
45
0
0
32
XC4000E devices include two types of global buffers, while
XC4000EX devices have three different types. These global buffers have different properties, and are intended for
different purposes. They are discussed in detail later in this
section.
4-32
September 18, 1996 (Version 1.04)
Quad
Single
Double
Long
CLB
Direct
Connect
Long
Quad
Long
Global
Clock
Long
Double Single Global
Clock
Carry Direct
Chain Connect
x5994
Figure 26: High-Level Routing Diagram of XC4000-Series CLB (shaded arrows indicate XC4000EX only)
September 18, 1996 (Version 1.04)
4-33
XC4000 Series Field Programmable Gate Arrays
QUAD
DOUBLE
SINGLE
DOUBLE
LONG
F4 C4 G4
YQ
DIRECT
Y
G1
C1
F1
CLB
G3
FEEDBACK
C3
F3
K
X
XQ
F2 C2 G2
LONG
LO
U
AD
G
D
LO
G
N
Q
LO
G
L
U
BL
E
O
SI
U
N
G
LE
G
LO
D
O
N
BA
N
G
BL
E
D
IR
LO
BA
L
EC
T
FE
ED
BA
C
K
Common to XC4000E and XC4000EX
XC4000EX only
Programmable Switch Matrix
Figure 27: Detail of Programmable Interconnect Associated with XC4000-Series CLB
4-34
September 18, 1996 (Version 1.04)
Single-Length Lines
e
ou
D
gl
in
S
bl
es
e
bl
ou
D
Single-length lines provide the greatest interconnect flexibility and offer fast routing between adjacent blocks. There
are eight vertical and eight horizontal single-length lines
associated with each CLB. These lines connect the switching matrices that are located in every row and a column of
CLBs.
Double
Singles
Six Pass Transistors
Per Switch Matrix
Interconnect Point
Double
X6600
Figure 28: Programmable Switch Matrix (PSM)
Programmable Switch Matrices
The horizontal and vertical single- and double-length lines
intersect at a box called a programmable switch matrix
(PSM). Each switch matrix consists of programmable pass
transistors used to establish connections between the lines
(see Figure 28).
For example, a single-length signal entering on the right
side of the switch matrix can be routed to a single-length
line on the top, left, or bottom sides, or any combination
thereof, if multiple branches are required. Similarly, a double-length signal can be routed to a double-length line on
any or all of the other three edges of the programmable
switch matrix.
CLB
Single-length lines are connected by way of the programmable switch matrices, as shown in Figure 29. Routing
connectivity is shown in Figure 27.
Single-length lines incur a delay whenever they go through
a switching matrix. Therefore, they are not suitable for routing signals for long distances. They are normally used to
conduct signals within a localized area and to provide the
branching for nets with fanout greater than one.
Double-Length Lines
The double-length lines consist of a grid of metal segments,
each twice as long as the single-length lines: they run past
two CLBs before entering a switch matrix. Double-length
lines are grouped in pairs with the switch matrices staggered, so that each line goes through a switch matrix at
every other row or column of CLBs (see Figure 29).
There are four vertical and four horizontal double-length
lines associated with each CLB. These lines provide faster
signal routing over intermediate distances, while retaining
routing flexibility. Double-length lines are connected by way
of the programmable switch matrices. Routing connectivity
is shown in Figure 27.
CLB
CLB
Doubles
PSM
Singles
PSM
Doubles
CLB
CLB
PSM
CLB
CLB
PSM
CLB
CLB
X6601
Figure 29: Single- and Double-Length Lines, with Programmable Switch Matrices (PSMs)
September 18, 1996 (Version 1.04)
4-35
XC4000 Series Field Programmable Gate Arrays
Quad Lines (XC4000EX only)
XC4000EX devices also include twelve vertical and twelve
horizontal quad lines per CLB row and column. Quad lines
are four times as long as the single-length lines. They are
interconnected via buffered switch matrices (shown as diamonds in Figure 27 on page 34). Quad lines run past four
CLBs before entering a buffered switch matrix. They are
grouped in fours, with the buffered switch matrices staggered, so that each line goes through a buffered switch
matrix at every fourth CLB location in that row or column.
(See Figure 30.)
The buffered switch matrixes have four pins, one on each
edge. All of the pins are bidirectional. Any pin can drive
any or all of the other pins.
Each buffered switch matrix contains one buffer and six
pass transistors. It resembles the programmable switch
matrix shown in Figure 28, with the addition of a programmable buffer. There can be up to two independent inputs
and up to two independent outputs. Only one of the independent inputs can be buffered.
The place and route software automatically uses the timing
requirements of the design to determine whether or not a
quad line signal should be buffered. A heavily loaded signal is typically buffered, while a lightly loaded one is not.
One scenario is to alternate buffers and pass transistors.
This allows both vertical and horizontal quad lines to be
buffered at alternating buffered switch matrices.
Due to the buffered switch matrices, quad lines are very
fast. They provide the fastest available method of routing
heavily loaded signals for long distances across the device.
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
CLB
X6602
Figure 30: Quad Lines (XC4000EX only)
4-36
September 18, 1996 (Version 1.04)
Longlines
Direct Interconnect (XC4000EX only)
Longlines form a grid of metal interconnect segments that
run the entire length or width of the array. Longlines are
intended for high fan-out, time-critical signal nets, or nets
that are distributed over long distances. In XC4000EX
devices, quad lines are preferred for critical nets, because
the buffered switch matrices make them faster for high fanout nets.
The XC4000EX offers two direct, efficient and fast connections between adjacent CLBs. These nets facilitate a data
flow from the left to the right side of the device, or from the
top to the bottom, as shown in Figure 31. Signals routed on
the direct interconnect exhibit minimum interconnect propagation delay and use no general routing resources.
IOB
CLB
CLB
IOB
~
~ ~
~
~
~ ~
~
IOB
IOB
CLB
IOB
IOB
IOB
IOB
IOB
~ ~
~
~
CLB
CLB
IOB
IOB
IOB
IOB
IOB
IOB
IOB
September 18, 1996 (Version 1.04)
CLB
~ ~
~
~
Routing connectivity of the longlines is shown in Figure 27
on page 34.
IOB
IOB
Each XC4000EX longline not driven by TBUFs has a buffered programmable splitter switch at the 1/4, 1/2, and 3/4
points of the array. Due to the buffering, XC4000EX longline performance does not deteriorate with the larger array
sizes. If the longline is split, the resulting partial longlines
are independent.
IOB
Each XC4000E longline has a programmable splitter switch
at its center, as does each XC4000EX longline driven by
TBUFs. This switch can separate the line into two independent routing channels, each running half the width or height
of the array.
The place and route software uses direct interconnect
whenever possible, to maximize routing resources and minimize interconnect delays.
IOB
Each horizontal longline driven by TBUFs has either two
(XC4000E) or eight (XC4000EX) pull-up resistors. To activate these resistors, attach a PULLUP symbol to the longline net. The software automatically activates the appropriate number of pull-ups. There is also a weak keeper at
each end of these two horizontal longlines. This circuit prevents undefined floating levels. However, it is overridden by
any driver, even a pull-up resistor.
The direct interconnect is also present between CLBs and
adjacent IOBs. Each IOB on the left and top device edges
has a direct path to the nearest CLB. Each CLB on the
right and bottom edges of the array has a direct path to the
nearest two IOBs, since there are two IOBs for each row or
column of CLBs.
~
~ ~
~
Two horizontal longlines per CLB can be driven by 3-state
or open-drain drivers (TBUFs). They can therefore implement unidirectional or bidirectional buses, wide multiplexers, or wired-AND functions. (See “Three-State Buffers” on
page 29 for more details.)
X6603
Figure 31: XC4000EX Direct Interconnect
4-37
XC4000 Series Field Programmable Gate Arrays
I/O Routing
XC4000-Series devices have additional routing around the
IOB ring. This routing is called a VersaRing. The
VersaRing facilitates pin-swapping and redesign without
affecting board layout. Included are eight double-length
lines spanning two CLBs (four IOBs), and four longlines.
Global lines and Wide Edge Decoder lines are provided.
XC4000EX devices also include eight octal lines.
Figure 33 is a detailed diagram of the XC4000E and
XC4000EX VersaRing. The area shown includes two IOBs.
There are two IOBs per CLB row or column, therefore this
diagram corresponds to the CLB routing diagram shown in
Figure 27 on page 34. The shaded areas represent routing
and routing connections present only in XC4000EX
devices.
A high-level diagram of the VersaRing is shown in
Figure 32. The shaded arrows represent routing present
only in XC4000EX devices.
IOB
WED
Quad
WED
Single
Double
INTERCONNECT
Long
Direct
Connect
Long
IOB
WED
Direct
Connect
Edge Double Long Global Octal
Decode
Clock
X5995
Figure 32: High-Level Routing Diagram of XC4000-Series VersaRing (Left Edge)
WED = Wide Edge Decoder, IOB = I/O Block (shaded arrows indicate XC4000EX only)
4-38
September 18, 1996 (Version 1.04)
QUAD
T
O
DOUBLE
SINGLE
C
L
B
DOUBLE
LONG
DECODER
IOB
I1
IK
OK
DECODER
T
I2
CE
O
DIRECT
A
R
R
A
Y
DECODER
IOB
O
T
OK
CE
IK
I1
I2
LONG
E
L
TA
C
O
L
BA
LO
G
G
E
BL
U
O
E
G OD
ED EC
D
N
LO
D
Common to XC4000E and XC4000EX
XC4000EX only
Figure 33: Detail of Programmable Interconnect Associated with XC4000-Series IOB (Left Edge)
September 18, 1996 (Version 1.04)
4-39
XC4000 Series Field Programmable Gate Arrays
Octal I/O Routing (XC4000EX only)
Between the XC4000EX CLB array and the pad ring, eight
interconnect tracks provide for versatility in pin assignment
and fixed pinout flexibility. (See Figure 34.)
These routing tracks are called octals, because they can be
broken every eight CLBs (sixteen IOBs) by a programmable buffer that also functions as a splitter switch. The buffers are staggered, so each line goes through a buffer at
every eighth CLB location around the device edge.
The octal lines bend around the corners of the device. The
lines cross at the corners in such a way that the segment
IOB
most recently buffered before the turn has the farthest distance to travel before the next buffer, as shown in
Figure 34.
IOB inputs and outputs interface with the octal lines via the
single-length interconnect lines. Single-length lines are
also used for communication between the octals and double-length lines, quads, and longlines within the CLB array.
Segmentation into buffered octals was found to be optimal
for distributing signals over long distances around the
device.
IOB
IOB
IOB
Segment with nearest buffer
connects to segment with furthest buffer
IOB
IOB
IOB
IOB
X6607
Figure 34: XC4000EX Octal I/O Routing
4-40
September 18, 1996 (Version 1.04)
Global Nets and Buffers
Both the XC4000E and the XC4000EX have dedicated global networks. These networks are designed to distribute
clocks and other high fanout control signals throughout the
devices with minimal skew. The global buffers are
described in detail in the following sections. The text
descriptions and diagrams are summarized in Table 17.
The table shows which CLB and IOB clock pins can be
sourced by which global buffers.
In both XC4000E and XC4000EX devices, placement of a
library symbol called BUFG results in the software choosing the appropriate clock buffer, based on the timing
requirements of the design. The detailed information in
these sections is included only for reference.
Global Nets and Buffers (XC4000E only)
Four vertical longlines in each CLB column are driven
exclusively by special global buffers. These longlines are
in addition to the vertical longlines used for standard interconnect. The four global lines can be driven by either of
two types of global buffers. The clock pins of every CLB
and IOB can also be sourced from local interconnect.
Two different types of clock buffers are available in the
XC4000E:
•
•
Primary Global Buffers (BUFGP)
Secondary Global Buffers (BUFGS)
Four Primary Global buffers offer the shortest delay and
negligible skew. Four Secondary Global buffers have
slightly longer delay and slightly more skew due to potentially heavier loading, but offer greater flexibility when used
to drive non-clock CLB inputs.
The Primary Global buffers must be driven by the semidedicated pads. The Secondary Global buffers can be
sourced by either semi-dedicated pads or internal nets.
Each CLB column has four dedicated vertical Global lines.
Each of these lines can be accessed by one particular Primary Global buffer, or by any of the Secondary Global buffers, as shown in Figure 35. Each corner of the device has
one Primary buffer and one Secondary buffer.
IOBs along the left and right edges have four vertical global
longlines. Top and bottom IOBs can be clocked from the
global lines in the adjacent CLB column.
A global buffer should be specified for all timing-sensitive
global signal distribution. To use a global buffer, place a
BUFGP (primary buffer), BUFGS (secondary buffer), or
BUFG (either primary or secondary buffer) element in a
schematic or in HDL code. If desired, attach a LOC
attribute or property to direct placement to the designated
location. For example, attach a LOC=L attribute or property
to a BUFGS symbol to direct that a buffer be placed in one
of the two Secondary Global buffers on the left edge of the
device, or a LOC=BL to indicate the Secondary Global
buffer on the bottom edge of the device, on the left.
Table 17: Clock Pin Access
XC4000E
All CLBs in Quadrant
All CLBs in Device
IOBs on Adjacent Vertical
Half Edge
IOBs on Adjacent Vertical
Full Edge
IOBs on Adjacent Horizontal
Half Edge (Direct)
IOBs on Adjacent Horizontal
Half Edge (through CLB globals)
IOBs on Adjacent Horizontal
Full Edge (through CLB globals)
BUFGP
BUFGS
BUFGLS
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
XC4000EX
L&R
T&B
BUFGE
BUFGE
√
√
√
√
BUFFCL
K
Local
Interconnect
√
√
√
√
√
√
√
√
√
√
√
√
L = Left, R = Right, T = Top, B = Bottom
September 18, 1996 (Version 1.04)
4-41
XC4000 Series Field Programmable Gate Arrays
IOB
locals
locals
locals
BUFGS
IOB
locals
IOB
IOB
BUFGP
PGCK1
SGCK4
PGCK4
SGCK1
4
4
BUFGS
BUFGP
4
locals
4
locals
CLB
CLB
IOB
IOB
locals
Any BUFGS
X4
locals
X4
Any BUFGS
X4
One BUFGP
per Global Line
IOB
locals
X4
locals
One BUFGP
per Global Line
CLB
locals
CLB
IOB
locals
BUFGP
BUFGS
SGCK3
IOB
IOB
IOB
PGCK3
locals
locals
locals
BUFGP
locals
PGCK2
SGCK2
BUFGS
IOB
X6604
Figure 35: XC4000E Global Net Distribution
IOB
BUFGLS
GCK1
IOB
IOB
IOB
BUFGLS
GCK8
GCK7
BUFFCLK
CLB
X4
FCLK1
BUFGLS 8
locals
locals
BUFGE
locals
locals
BUFGLS
BUFFCLK
X8 8
locals
8
locals
8
CLB CLOCKS
(PER COLUMN)
CLB CLOCKS
(PER COLUMN)
IOB
IOB
CLOCKS
locals
locals
locals
IOB
CLOCKS
IOB
locals
locals
8
locals
8 BUFGLS
locals
BUFGLS 8
8 BUFGLS
X8
X8
BUFFCLK
FCLK3
X8
BUFFCLK
locals
CLB
locals
locals
CLB
locals
X4
IOB
8
4
BUFGLS 8
FCLK2
BUFGE
GCK2
IOB
CLOCKS
CLB CLOCKS
(PER COLUMN)
CLB CLOCKS
(PER COLUMN)
8
BUFGE
FCLK4
8 BUFGLS
8
IOB
CLOCKS
locals
X8
BUFGLS
4
IOB
BUFGLS
BUFGE
CLB
X8
locals
BUFGLS 8
locals
BUFGLS
BUFGE
BUFGLS
BUFGE
GCK4
GCK3
BUFGLS
GCK6
BUFGE
BUFGE
IOB
IOB
IOB
IOB
BUFGLS
GCK5
X6694
Figure 36: XC4000EX Global Net Distribution
4-42
September 18, 1996 (Version 1.04)
Global Nets and Buffers (XC4000EX only)
Eight vertical longlines in each CLB column are driven by
special global buffers. These longlines are in addition to
the vertical longlines used for standard interconnect. The
global lines are broken in the center of the array, to allow
faster distribution and to minimize skew across the whole
array. Each half-column global line has its own buffered
multiplexer, as shown in Figure 36. The top and bottom global lines cannot be connected across the center of the
device, as this connection might introduce unacceptable
skew. The top and bottom halves of the global lines must
be separately driven — although they can be driven by the
same global buffer.
The eight global lines in each CLB column can be driven by
either of two types of global buffers. They can also be
driven by internal logic, because they can be accessed by
single, double, and quad lines at the top, bottom, half, and
quarter points. Consequently, the number of different
clocks that can be used simultaneously in an XC4000EX
device is very large.
There are four global lines feeding the IOBs at the left edge
of the device. IOBs along the right edge have eight global
lines. There is a single global line along the top and bottom
edges with access to the IOBs. All IOB global lines are broken at the center. They cannot be connected across the
center of the device, as this connection might introduce
unacceptable skew.
IOB global lines can be driven from any of three types of
global buffers, or from local interconnect. Alternatively, top
and bottom IOBs can be clocked from the global lines in the
adjacent CLB column.
Three different types of clock buffers are available in the
XC4000EX:
•
•
•
Global Low-Skew Buffers (BUFGLS)
Global Early Buffers (BUFGE)
FastCLK Buffers (BUFFCLK)
Global Low-Skew Buffers are the standard clock buffers.
They should be used for most internal clocking, whenever a
large portion of the device must be driven.
Global Early Buffers are designed to provide a faster clock
access, but CLB access is limited to one-fourth of the
device. They also facilitate a faster I/O interface.
FastCLK buffers are specifically designed to provide the
fastest possible I/O clock. They have only the standard
input access to CLBs, through local interconnect.
Figure 36 is a conceptual diagram of the global net structure in the XC4000EX.
Global Early buffers and Global Low-Skew buffers share a
single pad. Therefore, the same IPAD symbol can drive
one buffer of each type, in parallel. This configuration is
particularly useful when using the Fast Capture latches, as
described in “IOB Input Signals” on page 24. Paired Global
September 18, 1996 (Version 1.04)
Early and Global Low-Skew buffers share a common input;
they cannot be driven by two different signals.
Choosing an XC4000EX Clock Buffer
The clocking structure of the XC4000EX provides a large
variety of features. However, it can be simple to use, without understanding all the details. The software automatically handles clocks, along with all other routing, when the
appropriate clock buffer is placed in the design. In fact, if a
buffer symbol called BUFG is placed, rather than a specific
type of buffer, the software even chooses the buffer most
appropriate for the design. The detailed information in this
section is provided for those users who want a finer level of
control over their designs.
If fine control is desired, use the following summary and
Table 17 on page 41 to choose an appropriate clock buffer.
•
•
•
•
The simplest thing to do is to use a Global Low-Skew
buffer.
If a faster clock path is needed, try a BUFG. The
software will first try to use a Global Low-Skew Buffer. If
timing requirements are not met, a faster buffer will
automatically be used.
If a single quadrant of the chip is sufficient for the
clocked logic, and the timing requires a faster clock than
the Global Low-Skew buffer, use a Global Early buffer.
In special cases, where both external and internal
timing have been carefully studied, a FastCLK buffer
can be used, for the fastest possible I/O clock path.
Global Low-Skew Buffers
Each corner of the XC4000EX device has two Global LowSkew buffers. Any of the eight Global Low-Skew buffers
can drive any of the eight vertical Global lines in a column
of CLBs. In addition, any of the buffers can drive any of the
four vertical lines accessing the IOBs on the left edge of the
device, and any of the eight vertical lines accessing the
IOBs on the right edge of the device. (See Figure 37 on
page 44.)
IOBs at the top and bottom edges of the device are
accessed through the vertical Global lines in the CLB array,
as in the XC4000E. Any Global Low-Skew buffer can,
therefore, access every IOB and CLB in the device.
The Global Low-Skew buffers can be driven by either semidedicated pads or internal logic.
To use a Global Low-Skew buffer, place a BUFGLS element in a schematic or in HDL code. If desired, attach a
LOC attribute or property to direct placement to the designated location. For example, attach a LOC=T attribute or
property to direct that a BUFGLS be placed in one of the
two Global Low-Skew buffers on the top edge of the device,
or a LOC=TR to indicate the Global Low-Skew buffer on the
top edge of the device, on the right.
4-43
XC4000 Series Field Programmable Gate Arrays
8
7
IOB
8
IOB
1
6
I
O
B
2
IOB
1
6
CLB
I
O
B
I
O
B
CLB
CLB
I
O
B
CLB
CLB
I
O
B
I
O
B
CLB
CLB
I
O
B
IOB
IOB
IOB
IOB
CLB
I
O
B
7
IOB
3
5
2
3
4
5
4
X6751
X6753
Figure 37: Any BUFGLS (GCK1 - GCK8) Can
Drive Any or All Clock Inputs on the Device
Figure 38: Left and Right BUFGEs Can Drive Any or
All Clock Inputs in Same Quadrant or Edge (GCK1 is
shown. GCK2, GCK5 and GCK6 are similar.)
Global Early Buffers
Each corner of the XC4000EX device has two Global Early
buffers. The primary purpose of the Global Early buffers is
to provide an earlier clock access than the potentially
heavily-loaded Global Low-Skew buffers. A clock source
applied to both buffers will result in the Global Early clock
edge occurring several nanoseconds earlier than the Global Low-Skew buffer clock edge, due to the lighter loading.
Global Early buffers also facilitate the fast capture of device
inputs, using the Fast Capture latches described in “IOB
Input Signals” on page 24. For Fast Capture, take a single
clock signal, and route it through both a Global Early buffer
and a Global Low-Skew buffer. (The two buffers share an
input pad.) Use the Global Early buffer to clock the Fast
Capture latch, and the Global Low-Skew buffer to clock the
normal input flip-flop or latch, as shown in Figure 18 on
page 26.
The Global Early buffers can also be used to provide a fast
Clock-to-Out on device output pins. However, an early
clock in the output flip-flop IOB must be taken into consideration when calculating the internal clock speed for the
design.
The left-side Global Early buffers can each drive two of the
four vertical lines accessing the IOBs on the entire left edge
of the device. The right-side Global Early buffers can each
drive two of the eight vertical lines accessing the IOBs on
the entire right edge of the device. (See Figure 38.)
Each left and right Global Early buffer can also drive half of
the IOBs along either the top or bottom edge of the device,
using a dedicated line that can only be accessed through
the Global Early buffers.
The top and bottom Global Early buffers can drive half of
the IOBs along either the left or right edge of the device, as
shown in Figure 39. They can only access the top and bottom IOBs via the CLB global lines.
8
4-44
IOB
1
The Global Early buffers at the left and right edges of the
chip have slightly different capabilities than the ones at the
top and bottom. Refer to Figure 38, Figure 39, and
Figure 36 on page 42 while reading the following explanation.
Each Global Early buffer can access the eight vertical Global lines for all CLBs in the quadrant. Therefore, only onefourth of the CLB clock pins can be accessed. This restriction is in large part responsible for the faster speed of the
buffers, relative to the Global Low-Skew buffers.
7
IOB
6
I
O
B
CLB
CLB
I
O
B
I
O
B
CLB
CLB
I
O
B
IOB
IOB
2
3
5
4
X6747
Figure 39: Top and Bottom BUFGEs Can Drive Any
or All Clock Inputs in Same Quadrant (GCK8 is
shown. GCK3, GCK4 and GCK7 are similar.)
September 18, 1996 (Version 1.04)
The Global Early buffers can be driven by either semi-dedicated pads or internal logic. They share pads with the Global Low-Skew buffers, so a single net can drive both global
buffers, as described above.
To use a Global Early buffer, place a BUFGE element in a
schematic or in HDL code. If desired, attach a LOC
attribute or property to direct placement to the designated
location. For example, attach a LOC=T attribute or property to direct that a BUFGE be placed in one of the two Global Early buffers on the top edge of the device, or a
LOC=TR to indicate the Global Early buffer on the top edge
of the device, on the right.
IOB
IOB
1
I
O
B
CLB
CLB
I
O
B
4
2
I
O
B
CLB
CLB
I
O
B
3
IOB
IOB
FastCLK Buffers
The fastest way to bring a clock into the XC4000EX device
is through a FastCLK buffer. Two FastCLK buffers are
present on the left edge, and two on the right edge, of the
XC4000EX die. There are no FastCLK buffers on the top or
bottom edges.
One purpose of the FastCLK buffers is to create a very fast
pin-to-pin path by using the IOB 2-input function generator
in conjunction with the FastCLK. Drive the F input of the
IOB function generator with the FastCLK buffer output, as
described in “IOB Output Signals” on page 27.
Alternatively, a FastCLK buffer can be used to minimize the
setup time of device inputs, if a positive hold time is acceptable. Use the FastCLK buffer to clock the Fast Capture
latch, and a slower clock buffer to clock the standard IOB
flip-flop or latch. Either the Global Early buffer or the Global
Low-Skew buffer can be used for the second storage element, but whichever one is used should be the same clock
as the related internal logic. Since the FastCLK pads are
different from the Global Early and Global Low-Skew pads,
care must be taken to ensure that skew external to the
device does not create internal timing difficulties.
The FastCLK buffers can also be used to provide a fast
Clock-to-Out on device output pins. However, a fast clock
in the output flip-flop IOB must be taken into consideration
when calculating the internal clock speed for the design.
September 18, 1996 (Version 1.04)
X6745
Figure 40: Each BUFFCLK Can Drive Any or
All Clock Inputs in Same Half-Edge (FCLK1 is shown.
FCLK2, FCLK3 and FCLK4 are similar.)
The FastCLK buffers are limited to accessing IOBs on onehalf of the die edge only, as shown in Figure 40 and
Figure 36 on page 42. They can each drive two of the four
vertical lines accessing the IOBs on the left edge of the
device, or two of the eight vertical lines accessing the IOBs
on the right edge of the device. They can only access the
CLB array through single- and double-length lines.
The FastCLK buffers must be driven by the semi-dedicated
IOBs. They are not accessible from internal nets. Other
than the FastCLK feature, these IOBs are identical to all
other IOBs.
To use a FastCLK buffer, place a BUFFCLK element in a
schematic or in HDL code. If desired, attach a LOC
attribute or property to direct placement to the designated
location. For example, attach a LOC=LB attribute or property to direct that a BUFFCLK be placed on the left edge of
the device at the bottom, or use LOC=L to indicate either of
the buffers on the left edge.
The input to the BUFFCLK symbol must be driven by a
input pad symbol, such as IPAD, or by an input flip-flop or
latch, such as INFF, ILD, ILFFX, or ILFLX.
4-45
XC4000 Series Field Programmable Gate Arrays
Power Distribution
Pin Descriptions
Power for the FPGA is distributed through a grid to achieve
high noise immunity and isolation between logic and I/O.
Inside the FPGA, a dedicated Vcc and Ground ring surrounding the logic array provides power to the I/O drivers,
as shown in Figure 41. An independent matrix of Vcc and
Ground lines supplies the interior logic of the device.
There are three types of pins in the XC4000-Series
devices:
This power distribution grid provides a stable supply and
ground for all internal logic, providing the external package
power pins are all connected and appropriately decoupled.
Typically, a 0.1 µF capacitor connected near the Vcc and
Ground pins of the package will provide adequate decoupling.
Output buffers capable of driving/sinking the specified 12
mA (XC4000E) or 24 mA (XC4000EX) loads under specified worst-case conditions may be capable of driving/sinking up to 10 times as much current under best case
conditions.
Noise can be reduced by minimizing external load capacitance and reducing simultaneous output transitions in the
same direction. It may also be beneficial to locate heavily
loaded output buffers near the Ground pads. The I/O Block
output buffers have a slew-rate limited mode (default) which
should be used where output rise and fall times are not
speed-critical.
GND
•
•
•
Permanently dedicated pins
User I/O pins that can have special functions
Unrestricted user-programmable I/O pins.
Before and during configuration, all outputs not used for the
configuration process are 3-stated with a 50 kΩ - 100 kΩ
pull-up resistor.
After configuration, if an IOB is unused it is configured as
an input with a 50 kΩ - 100 kΩ pull-up resistor.
XC4000-Series devices have no dedicated Reset input.
Any user I/O can be configured to drive the Global Set/
Reset net, GSR. See “Global Set/Reset” on page 13 for
more information on GSR.
XC4000-Series devices have no Powerdown control input,
as the XC3000 and XC2000 families do. The XC3000/
XC2000 Powerdown control also 3-stated all of the device I/
O pins. For XC4000-Series devices, use the global 3-state
net, GTS, instead. This net 3-states all outputs, but does
not place the device in low-power mode. See “IOB Output
Signals” on page 27 for more information on GTS.
Device pins for XC4000-Series devices are described in
Table 18. Pin functions during configuration for each of the
seven configuration modes are summarized in Table 24 on
page 78, in the “Configuration Timing” section.
Ground and
Vcc Ring for
I/O Drivers
Vcc
Vcc
Logic
Power Grid
GND
X5422
Figure 41: XC4000-Series Power Distribution
4-46
September 18, 1996 (Version 1.04)
Table 18: Pin Descriptions
I/O
I/O
During
After
Pin Name
Config. Config.
Permanently Dedicated Pins
Pin Description
Eight or more (depending on package) connections to the nominal +5 V supply voltage
(+3.3 V for low-voltage devices). All must be connected, and each must be decoupled
with a 0.01 - 0.1 µF capacitor to Ground.
Eight or more (depending on package type) connections to Ground. All must be conGND
I
I
nected.
During configuration, Configuration Clock (CCLK) is an output in Master modes or Asynchronous Peripheral mode, but is an input in Slave mode, Synchronous Peripheral
mode, and Express mode. After configuration, CCLK has a weak pull-up resistor and
CCLK
I or O
I
can be selected as the Readback Clock. There is no CCLK High time restriction on
XC4000-Series devices, except during Readback. See “Violating the Maximum High
and Low Time Specification for the Readback Clock” on page 65 for an explanation of
this exception.
DONE is a bidirectional signal with an optional internal pull-up resistor. As an output, it
indicates the completion of the configuration process. As an input, a Low level on
DONE can be configured to delay the global logic initialization and the enabling of outDONE
I/O
O
puts.
The optional pull-up resistor is selected as an option in MakeBits, the XACTstep program that creates the configuration bitstream. The resistor is included by default.
PROGRAM is an active Low input that forces the FPGA to clear its configuration memory. It is used to initiate a configuration cycle. When PROGRAM goes High, the FPGA
finishes the current clear cycle and executes another complete clear cycle, before it
PROGRAM
I
I
goes into a WAIT state and releases INIT.
The PROGRAM pin has a permanent weak pull-up, so it need not be externally pulled
up to Vcc.
User I/O Pins That Can Have Special Functions
During Peripheral mode configuration, this pin indicates when it is appropriate to write
another byte of data into the FPGA. The same status is also available on D7 in AsynRDY/BUSY
O
I/O
chronous Peripheral mode, if a read operation is performed when the device is selected.
After configuration, RDY/BUSY is a user-programmable I/O pin.
RDY/BUSY is pulled High with a high-impedance pull-up prior to INIT going High.
During Master Parallel configuration, each change on the A0-A17 outputs (A0 - A21 for
XC4000EX) is preceded by a rising edge on RCLK, a redundant output signal. RCLK
O
I/O
RCLK
is useful for clocked PROMs. It is rarely used during configuration. After configuration,
RCLK is a user-programmable I/O pin.
As Mode inputs, these pins are sampled after INIT goes High to determine the configuration mode to be used. After configuration, M0 and M2 can be used as inputs, and M1
can be used as a 3-state output. These three pins have no associated input or output
registers.
I (M0), During configuration, these pins have weak pull-up resistors. For the most popular conM0, M1, M2
I
O (M1), figuration mode, Slave Serial, the mode pins can thus be left unconnected. The three
I (M2) mode inputs can be individually configured with or without weak pull-up or pull-down resistors. A pull-down resistor value of 4.7 kΩ is recommended.
These pins can only be used as inputs or outputs when called out by special schematic
definitions. To use these pins, place the library components MD0, MD1, and MD2 instead of the usual pad symbols. Input or output buffers must still be used.
VCC
I
I
September 18, 1996 (Version 1.04)
4-47
XC4000 Series Field Programmable Gate Arrays
Table 18: Pin Descriptions (Continued)
Pin Name
I/O
I/O
During
After
Config. Config.
TDO
O
O
TDI, TCK,
TMS
I
I/O
or I
(JTAG)
HDC
O
I/O
LDC
O
I/O
INIT
I/O
I/O
PGCK1 PGCK4
(XC4000E
only)
Weak
Pull-up
I or I/O
SGCK1 SGCK4
(XC4000E
only)
Weak
Pull-up
I or I/O
GCK1 GCK8
(XC4000EX
only)
Weak
Pull-up
I or I/O
FCLK1 FCLK4
(XC4000EX
only)
Weak
Pull-up
I or I/O
4-48
Pin Description
If boundary scan is used, this pin is the Test Data Output. If boundary scan is not used,
this pin is a 3-state output without a register, after configuration is completed.
This pin can be user output only when called out by special schematic definitions. To
use this pin, place the library component TDO instead of the usual pad symbol. An output buffer must still be used.
If boundary scan is used, these pins are Test Data In, Test Clock, and Test Mode Select
inputs respectively. They come directly from the pads, bypassing the IOBs. These pins
can also be used as inputs to the CLB logic after configuration is completed.
If the BSCAN symbol is not placed in the design, all boundary scan functions are inhibited once configuration is completed, and these pins become user-programmable I/O.
In this case, they must be called out by special schematic definitions. To use these pins,
place the library components TDI, TCK, and TMS instead of the usual pad symbols. Input or output buffers must still be used.
High During Configuration (HDC) is driven High until the I/O go active. It is available as
a control output indicating that configuration is not yet completed. After configuration,
HDC is a user-programmable I/O pin.
Low During Configuration (LDC) is driven Low until the I/O go active. It is available as a
control output indicating that configuration is not yet completed. After configuration,
LDC is a user-programmable I/O pin.
Before and during configuration, INIT is a bidirectional signal. A 1 kΩ - 10 kΩ external
pull-up resistor is recommended.
As an active-Low open-drain output, INIT is held Low during the power stabilization and
internal clearing of the configuration memory. As an active-Low input, it can be used
to hold the FPGA in the internal WAIT state before the start of configuration. Master
mode devices stay in a WAIT state an additional 30 to 300 µs after INIT has gone High.
During configuration, a Low on this output indicates that a configuration data error has
occurred. After the I/O go active, INIT is a user-programmable I/O pin.
Four Primary Global inputs each drive a dedicated internal global net with short delay
and minimal skew. If not used to drive a global buffer, any of these pins is a user-programmable I/O.
The PGCK1-PGCK4 pins drive the four Primary Global Buffers. Any input pad symbol
connected directly to the input of a BUFGP symbol is automatically placed on one of
these pins.
Four Secondary Global inputs each drive a dedicated internal global net with short delay
and minimal skew. These internal global nets can also be driven from internal logic. If
not used to drive a global net, any of these pins is a user-programmable I/O pin.
The SGCK1-SGCK4 pins provide the shortest path to the four Secondary Global Buffers. Any input pad symbol connected directly to the input of a BUFGS symbol is automatically placed on one of these pins.
Eight inputs can each drive a Global Low-Skew buffer. In addition, each can drive a Global Early buffer. Each pair of global buffers can also be driven from internal logic, but
must share an input signal. If not used to drive a global buffer, any of these pins is a
user-programmable I/O.
Any input pad symbol connected directly to the input of a BUFGLS or BUFGE symbol
is automatically placed on one of these pins.
Four FCLK inputs can each drive a FastCLK buffer. The FastCLK buffers cannot be
driven from internal logic. If not used to drive a global buffer, any of these pins is a userprogrammable I/O.
Any input pad symbol connected directly to the input of a BUFFCLK symbol is automatically placed on one of these pins.
September 18, 1996 (Version 1.04)
Table 18: Pin Descriptions (Continued)
Pin Name
I/O
I/O
During
After
Config. Config.
Pin Description
These four inputs are used in Asynchronous Peripheral mode. The chip is selected
when CS0 is Low and CS1 is High. While the chip is selected, a Low on Write Strobe
(WS) loads the data present on the D0 - D7 inputs into the internal data buffer. A Low
CS0, CS1,
on Read Strobe (RS) changes D7 into a status output — High if Ready, Low if Busy —
I
I/O
WS, RS
and drives D0 - D6 High.
In Express mode, CS1 is used as a serial-enable signal for daisy-chaining.
WS and RS should be mutually exclusive, but if both are Low simultaneously, the Write
Strobe overrides. After configuration, these are user-programmable I/O pins.
During Master Parallel configuration, these 18 output pins address the configuration
A0 - A17
O
I/O
EPROM. After configuration, they are user-programmable I/O pins.
A18 - A21
During Master Parallel configuration with an XC4000EX master, these 4 output pins add
(XC4000EX
O
I/O
4 more bits to address the configuration EPROM. After configuration, they are user-proonly)
grammable I/O pins.
During Master Parallel and Peripheral configuration, these eight input pins receive conD0 - D7
I
I/O
figuration data. After configuration, they are user-programmable I/O pins.
During Slave Serial or Master Serial configuration, DIN is the serial configuration data
DIN
I
I/O
input receiving data on the rising edge of CCLK. During Parallel configuration, DIN is
the D0 input. After configuration, DIN is a user-programmable I/O pin.
During configuration in any mode but Express mode, DOUT is the serial configuration
data output that can drive the DIN of daisy-chained slave FPGAs. DOUT data changes
on the falling edge of CCLK, one-and-a-half CCLK periods after it was received at the
DOUT
O
I/O
DIN input.
In Express mode, DOUT is the status output that can drive the CS1 of daisy-chained
FPGAs, to enable and disable downstream devices.
After configuration, DOUT is a user-programmable I/O pin.
Unrestricted User-Programmable I/O Pins
These pins can be configured to be input and/or output after configuration is completed.
Weak
I/O
I/O
Before configuration is completed, these pins have an internal high-value pull-up resisPull-up
tor (50 kΩ - 100 kΩ) that defines the logic level as High.
September 18, 1996 (Version 1.04)
4-49
XC4000 Series Field Programmable Gate Arrays
Boundary Scan
Data Registers
The ‘bed of nails’ has been the traditional method of testing
electronic assemblies. This approach has become less
appropriate, due to closer pin spacing and more sophisticated assembly methods like surface-mount technology
and multi-layer boards. The IEEE Boundary Scan Standard 1149.1 was developed to facilitate board-level testing
of electronic assemblies. Design and test engineers can
imbed a standard test logic structure in their device to
achieve high fault coverage for I/O and internal logic. This
structure is easily implemented with a four-pin interface on
any boundary scan-compatible IC. IEEE 1149.1-compatible devices may be serial daisy-chained together, connected in parallel, or a combination of the two.
The primary data register is the boundary scan register.
For each IOB pin in the FPGA, bonded or not, it includes
three bits for In, Out and 3-State Control. Non-IOB pins
have appropriate partial bit population for In or Out only.
PROGRAM, CCLK and DONE are not included in the
boundary scan register. Each EXTEST CAPTURE-DR
state captures all In, Out, and 3-state pins.
The XC4000 Series implements IEEE 1149.1-compatible
BYPASS, PRELOAD/SAMPLE and EXTEST boundary
scan instructions. When the boundary scan configuration
option is selected, three normal user I/O pins become dedicated inputs for these functions. Another user output pin
becomes the dedicated boundary scan output. The details
of how to enable this circuitry are covered later in this section.
By exercising these input signals, the user can serially load
commands and data into these devices to control the driving of their outputs and to examine their inputs. This
method is an improvement over bed-of-nails testing. It
avoids the need to over-drive device outputs, and it reduces
the user interface to four pins. An optional fifth pin, a reset
for the control logic, is described in the standard but is not
implemented in Xilinx devices.
The dedicated on-chip logic implementing the IEEE 1149.1
functions includes a 16-state state machine, an instruction
register and a number of data registers. The functional
details can be found in the IEEE 1149.1 specification and
are also discussed in the Xilinx application note XAPP 017:
"Boundary Scan in XC4000 Devices."
Figure 42 shows a simplified block diagram of the
XC4000E Input/Output Block with boundary scan implemented. XC4000EX boundary scan logic is identical.
Figure 43 on page 52 is a diagram of the XC4000-Series
boundary scan logic. It includes three bits of Data Register
per IOB, the IEEE 1149.1 Test Access Port controller, and
the Instruction Register with decodes.
XC4000-Series devices can also be configured through the
boundary scan logic. See “Configuration Through the
Boundary Scan Pins” on page 64.
4-50
The data register also includes the following non-pin bits:
TDO.T, and TDO.O, which are always bits 0 and 1 of the
data register, respectively, and BSCANT.UPD, which is
always the last bit of the data register. These three boundary scan bits are special-purpose Xilinx test signals.
The other standard data register is the single flip-flop
BYPASS register. It synchronizes data being passed
through the FPGA to the next downstream boundary scan
device.
The FPGA provides two additional data registers that can
be specified using the BSCAN macro. The FPGA provides
two user pins (BSCAN.SEL1 and BSCAN.SEL2) which are
the decodes of two user instructions. For these instructions, two corresponding pins (BSCAN.TDO1 and
BSCAN.TDO2) allow user scan data to be shifted out on
TDO. The data register clock (BSCAN.DRCK) is available
for control of test logic which the user may wish to implement with CLBs. The NAND of TCK and RUN-TEST-IDLE
is also provided (BSCAN.IDLE).
Instruction Set
The XC4000-Series boundary scan instruction set also
includes instructions to configure the device and read back
the configuration data. The instruction set is coded as
shown in Table 19.
Table 19: Boundary Scan Instructions
Instruction
I2
I1
I0
0
0
0
0
0
1
0
1
0
0
1
1
1
0
0
1
1
1
0
1
1
1
0
1
Test
TDO Source
Selected
EXTEST
DR
SAMPLE/
DR
PRELOAD
USER 1
BSCAN.
TDO1
USER 2
BSCAN.
TDO2
READBACK Readback
Data
CONFIGURE
DOUT
Reserved
—
BYPASS
Bypass
Register
I/O Data
Source
DR
Pin/Logic
User Logic
User Logic
Pin/Logic
Disabled
—
—
September 18, 1996 (Version 1.04)
EXTEST
TS INV
M
SLEW
RATE
PULL
DOWN
PULL
UP
TS/OE
3-State TS
Boundary
Scan
VCC
TS - capture
TS - update
OUTPUT
INVERT
OUTPUT
M
sd
D
Ouput Data O
Q
EC
M
INVERT
PAD
M
Ouput Clock OK
rd
M
OUT
SEL
S/R
M
Clock Enable
Boundary
Scan
O - capture
Q - capture
O - update
M
I - capture
Boundary
Scan
Input Data 1 I1
I - update
M M
sd
Q
D
M M
Input Data 2 I2
EC
DELAY
M
INVERT
QL
M
FLIP-FLOP/LATCH
Input Clock IK
rd
M
S/R
INPUT
GLOBAL
S/R
X5792
Figure 42: Block Diagram of XC4000E IOB with Boundary Scan (some details not shown).
XC4000EX Boundary Scan Logic is Identical.
September 18, 1996 (Version 1.04)
4-51
XC4000 Series Field Programmable Gate Arrays
DATA IN
1
0
sd
D
Q
D
Q
LE
1
0
IOB.Q
IOB.T
0
1
0
IOB
IOB
IOB
IOB
sd
D
Q
D
Q
1
LE
IOB
IOB
IOB
IOB
IOB
1
sd
D
Q
D
Q
0
LE
IOB
IOB
IOB
IOB
1
IOB.I
0
1
IOB
IOB
IOB
IOB
IOB
IOB
BYPASS
REGISTER
0
sd
D
Q
D
Q
LE
1
0
IOB.Q
IOB.T
0
M TDO
U
X
INSTRUCTION REGISTER
TDI
1
0
sd
D
Q
D
Q
1
LE
M
U
TDO X
TDI
INSTRUCTION REGISTER
1
BYPASS
REGISTER
IOB
IOB
IOB
IOB
IOB
IOB
IOB
IOB
IOB
IOB
IOB
IOB
IOB
IOB
0
sd
D
Q
D
Q
LE
1
IOB.I
0
1
0
sd
D
Q
D
Q
LE
0
1
IOB.O
DATAOUT
IOB
IOB
IOB
IOB
IOB
SHIFT/
CAPTURE
UPDATE
EXTEST
CLOCK DATA
REGISTER
X1523
Figure 43: XC4000-Series Boundary Scan Logic
4-52
September 18, 1996 (Version 1.04)
Bit Sequence
The bit sequence within each IOB is: In, Out, 3-State. The
input-only M0 and M2 mode pins contribute only the In bit
to the boundary scan I/O data register, while the outputonly M1 pin contributes all three bits.
The first two bits in the I/O data register are TDO.T and
TDO.O, which can be used for the capture of internal signals. The final bit is BSCANT.UPD, which can be used to
drive an internal net. These locations are primarily used by
Xilinx for internal testing.
From a cavity-up view of the chip (as shown in XDE or
Epic), starting in the upper right chip corner, the boundary
scan data-register bits are ordered as shown in Figure 44.
The device-specific pinout tables for the XC4000 Series
include the boundary scan locations for each IOB pin.
BSDL (Boundary Scan Description Language) files for
XC4000-Series devices are available on the Xilinx BBS.
Including Boundary Scan in a Schematic
If boundary scan is only to be used during configuration, no
special schematic elements need be included in the schematic or HDL code. In this case, the special boundary scan
pins TDI, TMS, TCK and TDO can be used for user functions after configuration.
Even if the boundary scan symbol is used in a schematic,
the input pins TMS, TCK, and TDI can still be used as
inputs to be routed to internal logic. Care must be taken not
to force the chip into an undesired boundary scan state by
inadvertently applying boundary scan input patterns to
these pins. The simplest way to prevent this is to keep TMS
High, and then apply whatever signal is desired to TDI and
TCK.
Avoiding Inadvertent Boundary Scan
Activation
If TMS or TCK is used as user I/O, care must be taken to
ensure that at least one of these pins is held constant during configuration. In some applications, a situation may
occur where TMS or TCK is driven during configuration.
This may cause the device to go into boundary scan mode
and disrupt the configuration process.
To prevent activation of boundary scan during configuration, do either of the following:
•
•
TMS: Tie High to put the Test Access Port controller
in a benign RESET state
TCK: Tie High or Low—don't toggle this clock input.
For more information regarding boundary scan, refer to the
Xilinx Application Note XAPP 017.001, “Boundary Scan in
XC4000E Devices.“
To indicate that boundary scan remain enabled after configuration, place the BSCAN library symbol and connect the
TDI, TMS, TCK and TDO pad symbols to the appropriate
pins, as shown in Figure 45.
Optional
Bit 0 ( TDO end)
Bit 1
Bit 2
TDO.T
TDO.O
To User
Logic
IBUF
BSCAN
Top-edge IOBs (Right to Left)
TDI
TDI
Left-edge IOBs (Top to Bottom)
TMS
TMS
MD1.T
MD1.O
MD1.I
MD0.I
MD2.I
TCK
TCK
IDLE
TDO1
SEL1
TDO2
SEL2
From
User Logic
TDO
TDO
DRCK
To User
Logic
X2675
Bottom-edge IOBs (Left to Right)
Figure 45: Boundary Scan Schematic Example
Right-edge IOBs (Bottom to Top)
(TDI end)
B SCANT.UPD
X6075
Figure 44:
Boundary Scan Bit Sequence
September 18, 1996 (Version 1.04)
4-53
XC4000 Series Field Programmable Gate Arrays
Configuration
Table 20: Configuration Modes
Configuration is the process of loading design-specific programming data into one or more FPGAs to define the functional operation of the internal blocks and their
interconnections. This is somewhat like loading the command registers of a programmable peripheral chip.
XC4000-Series devices use several hundred bits of configuration data per CLB and its associated interconnects.
Each configuration bit defines the state of a static memory
cell that controls either a function look-up table bit, a multiplexer input, or an interconnect pass transistor. The XACTstep development system translates the design into a
netlist file. It automatically partitions, places and routes the
logic and generates the configuration data in PROM format.
Mode
Master Serial
Slave Serial
Master
Parallel Up
M2
0
1
1
M1
0
1
0
M0
0
1
0
CCLK
output
input
output
Master
Parallel Down
1
1
0
output
Peripheral
Synchronous*
Peripheral
Asynchronous
Express
(XC4000EX
only)
Reserved
0
1
1
input
Data
Bit-Serial
Bit-Serial
Byte-Wide,
increment
from 00000
Byte-Wide,
decrement
from 3FFFF
Byte-Wide
1
0
1
output
Byte-Wide
0
1
0
input
Byte-Wide
0
0
1
—
—
Special Purpose Pins
Three configuration mode pins (M2, M1, M0) are sampled
prior to configuration to determine the configuration mode.
After configuration, these pins can be used as auxiliary
connections. M2 and M0 can be used as inputs, and M1
can be used as an output. The XACTstep development
system does not use these resources unless they are
explicitly specified in the design entry. This is done by placing a special pad symbol called MD2, MD1, or MD0 instead
of the input or output pad symbol.
In XC4000-Series devices, the mode pins have weak pullup resistors during configuration. With all three mode pins
High, Slave Serial mode is selected, which is the most popular configuration mode. Therefore, for the most common
configuration mode, the mode pins can be left unconnected. (Note, however, that the internal pull-up resistor
value can be as high as 100 kΩ.) After configuration, these
pins can individually have weak pull-up or pull-down resistors, as specified in the design. A pull-down resistor value
of 4.7 kΩ is recommended.
These pins are located in the lower left chip corner and are
near the readback nets. This location allows convenient
routing if compatibility with the XC2000 and XC3000 family
conventions of M0/RT, M1/RD is desired.
Configuration Modes
XC4000E devices have six configuration modes.
XC4000EX devices have the same six modes, plus an
additional configuration mode. These modes are selected
by a 3-bit input code applied to the M2, M1, and M0 inputs.
There are three self-loading Master modes, two Peripheral
modes, and a Serial Slave mode, which is used primarily
for daisy-chained devices. The seventh mode, called
Express mode, is an additional slave mode that allows
high-speed parallel configuration of the high-capacity
XC4000EX devices. The coding for mode selection is
shown in Table 20.
4-54
Note:
* Peripheral Synchronous can be considered bytewide Slave Parallel
A detailed description of each configuration mode, with timing information, is included later in this data sheet. During
configuration, some of the I/O pins are used temporarily for
the configuration process. All pins used during configuration are shown in Table 24 on page 78.
Master Modes
The three Master modes use an internal oscillator to generate a Configuration Clock (CCLK) for driving potential slave
devices. They also generate address and timing for external PROM(s) containing the configuration data.
Master Parallel (Up or Down) modes generate the CCLK
signal and PROM addresses and receive byte parallel data.
The data is internally serialized into the FPGA data-frame
format. The up and down selection generates starting
addresses at either zero or 3FFFF, for compatibility with different microprocessor addressing conventions. The Master
Serial mode generates CCLK and receives the configuration data in serial form from a Xilinx serial-configuration
PROM.
CCLK speed is selectable as either 1 MHz (default) or 8
MHz (up to 10% lower for low-voltage devices). Configuration always starts at the default slow frequency, then can
switch to the higher frequency during the first frame. Frequency tolerance is -50% to +25%.
Peripheral Modes
The two Peripheral modes accept byte-wide data from a
bus. A RDY/BUSY status is available as a handshake signal. In Asynchronous Peripheral mode, the internal oscillator generates a CCLK burst signal that serializes the bytewide data. CCLK can also drive slave devices. In the syn-
September 18, 1996 (Version 1.04)
chronous mode, an externally supplied clock input to CCLK
serializes the data.
Slave Serial Mode
In Slave Serial mode, the FPGA receives serial configuration data on the rising edge of CCLK and, after loading its
configuration, passes additional data out, resynchronized
on the next falling edge of CCLK.
Multiple slave devices with identical configurations can be
wired with parallel DIN inputs. In this way, multiple devices
can be configured simultaneously.
Serial Daisy Chain
Multiple devices with different configurations can be connected together in a “daisy chain,” and a single combined
bitstream used to configure the chain of slave devices.
To configure a daisy chain of devices, wire the CCLK pins
of all devices in parallel, as shown in Figure 55 on page 68.
Connect the DOUT of each device to the DIN of the next.
The lead or master FPGA and following slaves each
passes resynchronized configuration data coming from a
single source. The header data, including the length count,
is passed through and is captured by each FPGA when it
recognizes the 0010 preamble. Following the length-count
data, each FPGA outputs a High on DOUT until it has
received its required number of data frames.
After an FPGA has received its configuration data, it
passes on any additional frame start bits and configuration
data on DOUT. When the total number of configuration
clocks applied after memory initialization equals the value
of the 24-bit length count, the FPGAs begin the start-up
sequence and become operational together. FPGA I/O are
normally released two CCLK cycles after the last configuration bit is received. Figure 49 on page 61 shows the startup timing for an XC4000-Series device.
The daisy-chained bitstream is not simply a concatenation
of the individual bitstreams. The MakePROM program
must be used to combine the bitstreams for a daisychained configuration.
CCLK pulses until it reaches its finish point F. The different
families generate or require different numbers of additional
CCLK pulses until they reach F. Not reaching F means that
the device does not really finish its configuration, although
DONE may have gone High, the outputs became active,
and the internal reset was released. For the XC4000Series device, not reaching F means that readback cannot
be initiated and most boundary scan instructions cannot be
used.
The user has some control over the relative timing of these
events and can, therefore, make sure that they occur at the
proper time and the finish point F is reached. Timing is controlled using MakeBits options.
XC3000 Master with an XC4000-Series Slave
Some designers want to use an inexpensive lead device in
peripheral mode and have the more precious I/O pins of the
XC4000-Series devices all available for user I/O. Figure 46
provides a solution for that case.
This solution requires one CLB, one IOB and pin, and an
internal oscillator with a frequency of up to 5 MHz as a
clock source. The XC3000 master device must be configured with late Internal Reset, which is the default option.
One CLB and one IOB in the lead XC3000-family device
are used to generate the additional CCLK pulse required by
the XC4000-Series devices. When the lead device
removes the internal RESET signal, the 2-bit shift register
responds to its clock input and generates an active Low
output signal for the duration of the subsequent clock
period. An external connection between this output and
CCLK thus creates the extra CCLK pulse.
Express Mode (XC4000EX only)
Express mode is similar to Slave Serial mode, except the
data is presented in parallel format, and is clocked into the
target device a byte at a time rather than a bit at a time. The
data is loaded in parallel into eight different columns: it is
not internally serialized. Eight bits of configuration data are
loaded with every CCLK cycle, therefore this configuration
Multi-Family Daisy Chain
All Xilinx FPGAs of the XC2000, XC3000, and XC4000
Series use a compatible bitstream format and can, therefore, be connected in a daisy chain in an arbitrary
sequence. There is, however, one limitation. The lead
device must belong to the highest family in the chain. If the
chain contains XC4000-Series devices, the master normally cannot be an XC2000 or XC3000 device.
The reason for this rule is shown in Figure 49 on page 61.
Since all devices in the chain store the same length count
value and generate or receive one common sequence of
CCLK pulses, they all recognize length-count match on the
same CCLK edge, as indicated on the left edge of
Figure 49. The master device then generates additional
September 18, 1996 (Version 1.04)
OE/T
Reset
0
0
1
0
1
1
0
1
0
1
. etc .
.
.
Output
Connected
to CCLK
Active Low Output
Active High Output
X5223
Figure 46: CCLK Generation for XC3000 Master
Driving an XC4000-Series Slave
4-55
XC4000 Series Field Programmable Gate Arrays
mode runs at eight times the data rate of the other six
modes. A length count is not used in Express mode.
Express mode must be specified as an option to the MakeBits program, which generates the bitstream. The Express
mode bitstream is not compatible with the other six configuration modes.
Multiple slave devices with identical configurations can be
wired with parallel D0-D7 inputs. In this way, multiple
devices can be configured simultaneously.
Pseudo Daisy Chain
Multiple devices with different configurations can be connected together in a pseudo daisy chain, provided that all of
the devices are in Express mode. A single combined bitstream is used to configure the chain of Express mode
devices, but the input data bus must drive D0-D7 of each
device. Tie High the CS1 pin of the first device to be configured. Connect the DOUT pin of each FPGA to the CS1
pin of the next device in the chain. The D0-D7 inputs are
wired to each device in parallel. The DONE pins are wired
together, with one or more internal DONE pull-ups activated. Alternatively, a 4.7 kΩ external resistor can be used,
if desired. (See Figure 63 on page 76.)
The requirement that all DONE pins in a daisy chain be
wired together applies only to Express mode, and only if all
devices in the chain are to become active simultaneously.
All XC4000EX devices in Express mode are synchronized
to the DONE pin. User I/O for each device become active
after the DONE pin for that device goes High. (The exact
timing is determined by MakeBits options.) Since the
DONE pin is open-drain and does not drive a High value,
tying the DONE pins of all devices together prevents all
devices in the chain from going High until the last device in
the chain has completed its configuration cycle.
Because only XC4000EX and XC5200 devices support
Express mode, only these devices can be used to form an
Express mode daisy chain. XC5200 devices used in a
combined daisy chain with XC4000EX devices should be
configured as synchronized to DONE (MakeBits option
CCLK_SYNC or UCLK_SYNC), and their DONE pins wired
together with those of the XC4000EX devices.
Setting CCLK Frequency
For Master modes, CCLK can be generated in either of two
frequencies. In the default slow mode, the frequency
ranges from 0.5 MHz to 1.25 MHz (up to 10% lower for lowvoltage devices). In fast CCLK mode, the frequency ranges
from 4 MHz to 10 MHz (up to 10% lower for low-voltage
devices). The frequency is selected by an option when running MakeBits, the bitstream generation software tool. If an
XC4000-Series Master is driving an XC3000- or XC2000family slave, slow CCLK mode must be used. Slow mode is
the default.
4-56
Data Stream Format
The data stream (“bitstream”) format is identical for all configuration modes, with the exception of Express mode. In
Express mode, the device becomes active when DONE
goes High, therefore no length count is required. Additionally, CRC error checking is not supported in Express mode.
The data stream formats are shown in Table 21. Express
mode data is shown with D0 at the left and D7 at the right.
For all other modes, bit-serial data is read from left to right,
and byte-parallel data is effectively assembled from this
serial bitstream, with the first bit in each byte assigned to
D0.
The configuration data stream begins with a string of eight
ones, a preamble code, followed by a 24-bit length count
and a separator field of ones (or 24 fill bits, in Express
mode). This header is followed by the actual configuration
data in frames. The length and number of frames depends
on the device type (see Table 22 and Table 23). Each
frame begins with a start field and ends with an error check.
In all modes except Express mode, a postamble code is
required to signal the end of data for a single device. In all
cases, additional start-up bytes of data are required to provide four clocks for the startup sequence at the end of configuration. Long daisy chains require additional startup
bytes to shift the last data through the chain. All startup
bytes are don’t-cares; these bytes are not included in bitstreams created by the Xilinx software.
Table 21: XC4000-Series Data Stream Formats
Data Type
Fill Byte
Preamble Code
Length Count
Fill Bits
Start Field
Data Frame
CRC or Constant
Field Check
Extend Write Cycle
Postamble
Start-Up Bytes
Express Mode
(D0-D7)
11111111b
11110010b
FFFFFFh
—
11010010b
DATA(n-1:0)
11010010b
FFFFFFFFFFh
—
xxxxxxxxh
All Other
Modes (D0...)
11111111b
0010b
COUNT(23:0)
1111b
0b
DATA(n-1:0)
xxxx (CRC)
or 0110b
—
01111111b
xxh
LEGEND:
Unshaded
Once per bitstream
Light
Dark
Once per data frame
Once per device
September 18, 1996 (Version 1.04)
Table 22: XC4000E Program Data
Device
Max Logic Gates
CLBs
(Row x Col.)
IOBs
Flip-Flops
Horizontal
Longlines
TBUFs per
Longline
Bits per Frame
Frames
Program Data
PROM Size
(bits)
XC4003E XC4005E/L XC4006E
3,000
5,000
6,000
100
196
256
(10 x 10)
(14 x 14)
(16 x 16)
80
112
128
360
616
768
20
28
32
XC4008E XC4010E/L XC4013E/L XC4020E
8,000
10,000
13,000
20,000
324
400
576
784
(18 x 18)
(20 x 20)
(24 x 24)
(28 x 28)
144
160
192
224
936
1,120
1,536
2,016
36
40
48
56
XC4025E
25,000
1,024
(32 x 32)
256
2,560
64
12
16
18
20
22
26
30
34
126
428
53,936
53,984
166
572
94,960
95,008
186
644
119,792
119,840
206
716
147,504
147,552
226
788
178,096
178,144
266
932
247,920
247,968
306
1,076
329,264
329,312
346
1,220
422,128
422,176
Notes: 1. Bits per Frame = (10 x number of rows) + 7 for the top + 13 for the bottom + 1 + 1 start bit + 4 error check bits
Number of Frames = (36 x number of columns) + 26 for the left edge + 41 for the right edge + 1
Program Data = (Bits per Frame x Number of Frames) + 8 postamble bits
PROM Size = Program Data + 40
2. The user can add more “one” bits as leading dummy bits in the header, or, if CRC = off, as trailing dummy bits at the end of
any frame, following the four error check bits. However, the Length Count value must be adjusted for all such extra “one”
bits, even for extra leading ones at the beginning of the header.
The MakeBits software creates the configuration bitstream.
In Express mode, only non-CRC error checking is supported. In all other modes, MakeBits allows a selection of
CRC or non-CRC error checking. The non-CRC error
checking tests for a designated end-of-frame field for each
frame. For CRC error checking, MakeBits calculates a running CRC and inserts a unique four-bit partial check at the
end of each frame. The 11-bit CRC check of the last frame
of an FPGA includes the last seven data bits.
Detection of an error results in the suspension of data loading and the pulling down of the INIT pin. In Master modes,
CCLK and address signals continue to operate externally.
The user must detect INIT and initialize a new configuration
by pulsing the PROGRAM pin Low or cycling Vcc.
Cyclic Redundancy Check (CRC) for
Configuration and Readback
The Cyclic Redundancy Check is a method of error detection in data transmission applications. Generally, the transmitting system performs a calculation on the serial
bitstream. The result of this calculation is tagged onto the
data stream as additional check bits. The receiving system
September 18, 1996 (Version 1.04)
performs an identical calculation on the bitstream and compares the result with the received checksum.
Each data frame of the configuration bitstream has four
error bits at the end, as shown in Table 21. If a frame data
error is detected during the loading of the FPGA, the configuration process with a potentially corrupted bitstream is
terminated. The FPGA pulls the INIT pin Low and goes into
a Wait state.
During Readback, 11 bits of the 16-bit checksum are added
to the end of the Readback data stream. The checksum is
computed using the CRC-16 CCITT polynomial, as shown
in Figure 47. The checksum consists of the 11 most significant bits of the 16-bit code. A change in the checksum
indicates a change in the Readback bitstream. A comparison to a previous checksum is meaningful only if the readback data is independent of the current device state. CLB
outputs should not be included (Read Capture MakeBits
option not used), and if RAM is present, the RAM content
must be unchanged.
Statistically, one error out of 2048 might go undetected.
4-57
XC4000 Series Field Programmable Gate Arrays
Table 23: XC4000EX Program Data
Device
Max Logic Gates
CLBs
(Row x Col.)
IOBs
Flip-Flops
Horizontal Longlines
TBUFs per Longline
Bits per Frame
Frames
Program Data
PROM Size (bits)
XC4028EX/XL
28,000
1,024
(32 x 32)
256
2,560
192
34
421
1587
668,127
668,167
XC4036EX/XL
36,000
1,296
(36 x 36)
288
3,168
216
38
469
1775
832,483
832,523
XC4044EX/XL
44,000
1,600
(40 x 40)
320
3,840
240
42
517
1963
1,014,879
1,014,919
XC4052XL
52,000
1,936
(44 x 44)
352
4,576
264
46
565
2151
1,215,323
1,215,363
XC4062XL
62,000
2,304
(48 x 48)
384
5,376
288
50
613
2,339
1,433,807
1,433,847
Notes: 1. Bits per Frame = (12 x number of rows) + 8 for the top + 16 for the bottom + 8 + 1 start bit + 4 error check bits
Number of Frames = (47 x number of columns) + 27 for the left edge + 52 for the right edge + 4
Program Data = (Bits per Frame x Number of Frames) + 8 postamble bits
PROM Size = Program Data + 40
2. The user can add more “one” bits as leading dummy bits in the header, or, if CRC = off, as trailing dummy bits at the end of
any frame, following the four error check bits. However, the Length Count value must be adjusted for all such extra “one”
bits, even for extra leading ones at the beginning of the header.
3. Express mode bitfiles are slightly larger (see Table 21).
X2
X15
X16
0 1
2
3
4
5
6
7
8
9 10 11 12 13 14
15
SERIAL DATA IN
Polynomial: X16 + X15 + X2 + 1
1
1
1
1
0 15 14 13 12 11 10 9
START BIT
1
LAST DATA FRAME
8
7
6
5
CRC – CHECKSUM
Readback Data Stream
X1789
Figure 47: Circuit for Generating CRC-16
4-58
September 18, 1996 (Version 1.04)
Configuration Sequence
Configuration Memory Clear
Initialization
Configuration
Start-Up
No
Yes
Test M0 Generate
One Time-Out Pulse
of 16 or 64 ms
PROGRAM
= Low
The full process is illustrated in Figure 48.
Yes
Configuration Memory Clear
When power is first applied or is reapplied to an FPGA, an
internal circuit forces initialization of the configuration logic.
When Vcc reaches an operational level, and the circuit
passes the write and read test of a sample pair of configuration bits, a time delay is started. This time delay is nominally 16 ms, and up to 10% longer in the low-voltage
devices. The delay is four times as long when in Master
Modes (M0 Low), to allow ample time for all slaves to reach
a stable Vcc. When all INIT pins are tied together, as recommended, the longest delay takes precedence. Therefore, devices with different time delays can easily be mixed
and matched in a daisy chain.
This delay is applied only on power-up. It is not applied
when reconfiguring an FPGA by pulsing the PROGRAM pin
Low. During this time delay, or as long as the PROGRAM
input is asserted, the configuration logic is held in a Configuration Memory Clear state. The configuration-memory
frames are consecutively initialized, using the internal oscillator.
At the end of each complete pass through the frame
addressing, the power-on time-out delay circuitry and the
level of the PROGRAM pin are tested. If neither is
asserted, the logic initiates one additional clearing of the
configuration frames and then tests the INIT input.
Keep Clearing
Configuration Memory
EXTEST*
SAMPLE/PRELOAD
Completely Clear
BYPASS
Configuration Memory
CONFIGURE*
Once More
(* if PROGRAM = High)
INIT
High? if
Master
Yes
No
Master Waits 50 to 250 µs
Before Sampling Mode Lines
Sample
Mode Lines
Master CCLK
Goes Active
Load One
Configuration
Data Frame
Frame
Error
Yes
Pull INIT Low
and Stop
No
SAMPLE/PRELOAD
BYPASS
Configuration
memory
Full
No
Yes
Initialization
Pass
Configuration
Data to DOUT
During initialization and configuration, user pins HDC, LDC,
INIT and DONE provide status outputs for the system interface. The outputs LDC, INIT and DONE are held Low and
HDC is held High starting at the initial application of power.
The open drain INIT pin is released after the final initialization pass through the frame addresses. There is a deliberate delay of 50 to 250 µs (up to 10% longer for low-voltage
devices) before a Master-mode device recognizes an inactive INIT. Two internal clocks after the INIT pin is recognized
as High, the FPGA samples the three mode lines to determine the configuration mode. The appropriate interface
lines become active and the configuration preamble and
data can be loaded.
~1.3 µs per Frame
LDC Output = L, HDC Output = H
•
•
•
•
VCC
>3.5 V
Boundary Scan
Instructions
Available:
CCLK
Count Equals
Length
Count
No
Yes
Start-Up
Sequence
F
EXTEST
SAMPLE PRELOAD
BYPASS
USER 1
USER 2
CONFIGURE
READBACK
Operational
I/O Active
There are four major steps in the XC4000-Series power-up
configuration sequence.
If Boundary Scan
is Selected
X6076
Figure 48: Power-up Configuration Sequence
September 18, 1996 (Version 1.04)
4-59
XC4000 Series Field Programmable Gate Arrays
Configuration
The 0010 preamble code, included for all modes except
Express mode, indicates that the following 24 bits represent the length count. The length count is the total number
of configuration clocks needed to load the complete configuration data. (Four additional configuration clocks are
required to complete the configuration process, as discussed below.) After the preamble and the length count
have been passed through to all devices in the daisy chain,
DOUT is held High to prevent frame start bits from reaching
any daisy-chained devices. In Express mode, the length
count bits are ignored, and DOUT is held Low, to disable
the next device in the pseudo daisy chain.
A specific configuration bit, early in the first frame of a master device, controls the configuration-clock rate and can
increase it by a factor of eight. Therefore, if a fast configuration clock is selected by the bitstream, the slower clock
rate is used until this configuration bit is detected.
Each frame has a start field followed by the frame-configuration data bits and a frame error field. If a frame data error
is detected, the FPGA halts loading, and signals the error
by pulling the open-drain INIT pin Low. After all configuration frames have been loaded into an FPGA, DOUT again
follows the input data so that the remaining data is passed
on to the next device. In Express mode, when the first
device is fully programmed, DOUT goes High to enable the
next device in the chain.
Delaying Configuration After Power-Up
There are two methods of delaying configuration after
power-up: put a logic Low on the PROGRAM input, or pull
the bidirectional INIT pin Low, using an open-collector
(open-drain) driver. (See Figure 48 on page 59.)
A Low on the PROGRAM input is the more radical
approach, and is recommended when the power-supply
rise time is excessive or poorly defined. As long as PROGRAM is Low, the FPGA keeps clearing its configuration
memory. When PROGRAM goes High, the configuration
memory is cleared one more time, followed by the beginning of configuration, provided the INIT input is not externally held Low. Note that a Low on the PROGRAM input
automatically forces a Low on the INIT output. The
XC4000-Series PROGRAM pin has a permanent weak
pull-up.
Using an open-collector or open-drain driver to hold INIT
Low before the beginning of configuration causes the
FPGA to wait after completing the configuration memory
clear operation. When INIT is no longer held Low externally, the device determines its configuration mode by capturing its mode pins, and is ready to start the configuration
process. A master device waits up to an additional 250 µs
4-60
to make sure that any slaves in the optional daisy chain
have seen that INIT is High.
Start-Up
Start-up is the transition from the configuration process to
the intended user operation. This transition involves a
change from one clock source to another, and a change
from interfacing parallel or serial configuration data where
most outputs are 3-stated, to normal operation with I/O pins
active in the user-system. Start-up must make sure that the
user-logic ‘wakes up’ gracefully, that the outputs become
active without causing contention with the configuration signals, and that the internal flip-flops are released from the
global Reset or Set at the right time.
Figure 49 describes start-up timing for the three Xilinx families in detail. Express mode configuration always uses
either CCLK_SYNC or UCLK_SYNC timing, the other configuration modes can use any of the four timing sequences.
To access the internal start-up signals, place the STARTUP
library symbol.
Start-up Timing
Different FPGA families have different start-up sequences.
The XC2000 family goes through a fixed sequence. DONE
goes High and the internal global Reset is de-activated one
CCLK period after the I/O become active.
The XC3000A family offers some flexibility. DONE can be
programmed to go High one CCLK period before or after
the I/O become active. Independent of DONE, the internal
global Reset is de-activated one CCLK period before or
after the I/O become active.
The XC4000 Series offers additional flexibility. The three
events — DONE going High, the internal Set/Reset being
de-activated, and the user I/O going active — can all occur
in any arbitrary sequence. Each of them can occur one
CCLK period before or after, or simultaneous with, any of
the others. This relative timing is selected by means of software options in MakeBits, the bitstream generation software.
The default option, and the most practical one, is for DONE
to go High first, disconnecting the configuration data source
and avoiding any contention when the I/Os become active
one clock later. Reset/Set is then released another clock
period later to make sure that user-operation starts from
stable internal conditions. This is the most common
sequence, shown with heavy lines in Figure 49, but the
designer can modify it to meet particular requirements.
Normally, the start-up sequence is controlled by the internal
device oscillator output (CCLK), which is asynchronous to
the system clock.
September 18, 1996 (Version 1.04)
Length Count Match
CCLK Period
CCLK
F
DONE
XC2000
I/O
Global Reset
F = Finished, no more
configuration clocks needed
Daisy-chain lead device
must have latest F
F
XC3000
DONE
I/O
Heavy lines describe
default timing
Global Reset
F
DONE
C1
XC4000E/EX
C2
C3
C4
C2
C3
C4
C2
C3
C4
I/O
CCLK_NOSYNC
GSR Active
DONE IN
F
DONE
C1, C2 or C3
XC4000E/EX
I/O
CCLK_SYNC
Di
Di+1
GSR Active
Di
Di+1
F
DONE
C1
XC4000E/EX
U2
U3
U4
U2
U3
U4
U2
U3
U4
I/O
UCLK_NOSYNC
GSR Active
DONE IN
F
DONE
C1
XC4000E/EX
U2
I/O
UCLK_SYNC
Di
Di+1
Di+2
Di+1
Di+2
GSR Active
Synchronization
Uncertainty
Di
UCLK Period
X6700
Figure 49: Start-up Timing
September 18, 1996 (Version 1.04)
4-61
XC4000 Series Field Programmable Gate Arrays
The XC4000 Series offers another start-up clocking option,
UCLK_NOSYNC. The three events described above need
not be triggered by CCLK. They can, as a configuration
option, be triggered by a user clock. This means that the
device can wake up in synchronism with the user system.
When the UCLK_SYNC option is enabled, the user can
externally hold the open-drain DONE output Low, and thus
stall all further progress in the start-up sequence until
DONE is released and has gone High. This option can be
used to force synchronization of several FPGAs to a common user clock, or to guarantee that all devices are successfully configured before any I/Os go active.
If either of these two options is selected, and no user clock
is specified in the design or attached to the device, the chip
could reach a point where the configuration of the device is
complete and the Done pin is asserted, but the outputs do
not become active. The solution is either to recreate the bitstream specifying the start-up clock as CCLK, or to supply
the appropriate user clock.
Start-up Sequence
The Start-up sequence begins when the configuration
memory is full, and the total number of configuration clocks
received since INIT went High equals the loaded value of
the length count.
The next rising clock edge sets a flip-flop Q0, shown in
Figure 50. Q0 is the leading bit of a 5-bit shift register. The
outputs of this register can be programmed to control three
events.
•
•
•
The release of the open-drain DONE output
The change of configuration-related pins to the user
function, activating all IOBs.
The termination of the global Set/Reset initialization of
all CLB and IOB storage elements.
The DONE pin can also be wire-ANDed with DONE pins of
other FPGAs or with other external signals, and can then
be used as input to bit Q3 of the start-up register. This is
called “Start-up Timing Synchronous to Done In” and is
selected by the CCLK_SYNC and UCLK_SYNC MakeBits
options.
When DONE is not used as an input, the operation is called
“Start-up Timing Not Synchronous to DONE In,” and is
selected by the CCLK_NOSYNC and UCLK_NOSYNC
MakeBits options.
As a configuration option, the start-up control register
beyond Q0 can be clocked either by subsequent CCLK
pulses or from an on-chip user net called STARTUP.CLK.
These signals can be accessed by placing the STARTUP
library symbol.
4-62
Start-up from CCLK
If CCLK is used to drive the start-up, Q0 through Q3 provide the timing. Heavy lines in Figure 49 show the default
timing, which is compatible with XC2000 and XC3000
devices using early DONE and late Reset. The thin lines
indicate all other possible timing options.
Start-up from a User Clock (STARTUP.CLK)
When, instead of CCLK, a user-supplied start-up clock is
selected, Q1 is used to bridge the unknown phase relationship between CCLK and the user clock. This arbitration
causes an unavoidable one-cycle uncertainty in the timing
of the rest of the start-up sequence.
DONE Goes High to Signal End of Configuration
In all configuration modes except Express mode, XC4000Series devices read the expected length count from the bitstream and store it in an internal register. The length count
varies according to the number of devices and the composition of the daisy chain. Each device also counts the number of CCLKs during configuration.
Two conditions have to be met in order for the DONE pin to
go high:
•
•
the chip's internal memory must be full, and
the configuration length count must be met, exactly.
This is important because the counter that determines
when the length count is met begins with the very first
CCLK, not the first one after the preamble.
Therefore, if a stray bit is inserted before the preamble, or
the data source is not ready at the time of the first CCLK,
the internal counter that holds the number of CCLKs will be
one ahead of the actual number of data bits read. At the
end of configuration, the configuration memory will be full,
but the number of bits in the internal counter will not match
the expected length count.
As a consequence, a Master mode device will continue to
send out CCLKs until the internal counter turns over to
zero, and then reaches the correct length count a second
time. This will take several seconds [224 ∗ CCLK period] —
which is sometimes interpreted as the device not configuring at all.
If it is not possible to have the data ready at the time of the
first CCLK, the problem can be avoided by increasing the
number in the length count by the appropriate value. The
XACT User Guide includes detailed information about manually altering the length count.
In Express mode, there is no length count. The DONE pin
for each device goes High when the device has received its
quota of configuration data. Wiring the DONE pins of several devices together delays start-up of all devices until all
are fully configured.
September 18, 1996 (Version 1.04)
Q3
STARTUP
Q1/Q4
DONE
IN
Q2
*
IOBs OPERATIONAL PER CONFIGURATION
*
GLOBAL SET/RESET OF
ALL CLB AND IOB FLIP-FLOP
1
0
GSR ENABLE
GSR INVERT
STARTUP.GSR
CONTROLLED BY STARTUP SYMBOL
IN THE USER SCHEMATIC (SEE
LIBRARIES GUIDE)
STARTUP.GTS
GTS INVERT
GTS ENABLE
0
GLOBAL 3-STATE OF ALL IOBs
1
Q
S
R
*
DONE
1
1
0
0
Q0
FULL
LENGTH COUNT
Q1
Q2
Q3
" FINISHED "
ENABLES BOUNDARY
SCAN, READBACK AND
CONTROLS THE OSCILLATOR
Q4
1
S
Q
D
Q
D
Q
0
D
Q
D
Q
M
K
K
K
*
K
K
CLEAR MEMORY
CCLK
0
STARTUP.CLK
USER NET
1
M
*
*
CONFIGURATION BIT OPTIONS SELECTED BY USER IN "MAKEBITS"
X1528
Figure 50: Start-up Logic
Note that DONE is an open-drain output and does not go
High unless an internal pull-up is activated or an external
pull-up is attached. The internal pull-up is activated as the
default by MakeBits, the bitstream generation software.
Release of User I/O After DONE Goes High
By default, the user I/O are released one CCLK cycle after
the DONE pin goes High. If CCLK is not clocked after
DONE goes High, the outputs remain in their initial state —
3-stated, with a 50 kΩ - 100 kΩ pull-up. The delay from
DONE High to active user I/O is controlled by a MakeBits
option.
September 18, 1996 (Version 1.04)
Release of Global Set/Reset After DONE Goes
High
By default, Global Set/Reset (GSR) is released two CCLK
cycles after the DONE pin goes High. If CCLK is not
clocked twice after DONE goes High, all flip-flops are held
in their initial set or reset state. The delay from DONE High
to GSR inactive is controlled by a MakeBits option.
Configuration Complete After DONE Goes High
Three full CCLK cycles are required after the DONE pin
goes High, as shown in Figure 49 on page 61. If CCLK is
not clocked three times after DONE goes High, readback
cannot be initiated and most boundary scan instructions
cannot be used.
4-63
XC4000 Series Field Programmable Gate Arrays
Configuration Through the Boundary Scan
Pins
XC4000-Series devices can be configured through the
boundary scan pins. The basic procedure is as follows:
•
Power up the FPGA with INIT held Low (or drive the
PROGRAM pin Low for more than 300 ns followed by a
High while holding INIT Low). Holding INIT Low allows
enough time to issue the CONFIG command to the
FPGA. The pin can be used as I/O after configuration if
a resistor is used to hold INIT Low.
• Issue the CONFIG command to the TMS input
• Wait for INIT to go High
• Sequence the boundary scan Test Access Port to the
SHIFT-DR state
• Toggle TCK to clock data into TDI pin.
The user must account for all TCK clock cycles after INIT
goes High, as all of these cycles affect the Length Count
compare.
For more detailed information, refer to the Xilinx application
note XAPP017, “Boundary Scan in XC4000 Devices.” This
application note also applies to XC4000E and XC4000EX
devices.
Readback
The user can read back the content of configuration memory and the level of certain internal nodes without interfering with the normal operation of the device.
Readback not only reports the downloaded configuration
bits, but can also include the present state of the device,
represented by the content of all flip-flops and latches in
CLBs and IOBs, as well as the content of function generators used as RAMs.
Note that in XC4000-Series devices, configuration data is
not inverted with respect to configuration as it is in XC2000
and XC3000 families.
Readback of Express mode bitstreams results in data that
does not resemble the original bitstream, because the bitstream format differs from other modes.
XC4000-Series Readback does not use any dedicated
pins, but uses four internal nets (RDBK.TRIG, RDBK.DATA,
RDBK.RIP and RDBK.CLK) that can be routed to any IOB.
To access the internal Readback signals, place the READBACK library symbol and attach the appropriate pad symbols, as shown in Figure 51.
After Readback has been initiated by a Low-to-High transition on RDBK.TRIG, the RDBK.RIP (Read In Progress)
output goes High on the next rising edge of RDBK.CLK.
Subsequent rising edges of this clock shift out Readback
data on the RDBK.DATA net.
Readback data does not include the preamble, but starts
with five dummy bits (all High) followed by the Start bit
(Low) of the first frame. The first two data bits of the first
frame are always High.
Each frame ends with four error check bits. They are read
back as High. The last seven bits of the last frame are also
read back as High. An additional Start bit (Low) and an
11-bit Cyclic Redundancy Check (CRC) signature follow,
before RDBK.RIP returns Low.
IF UNCONNECTED,
DEFAULT IS CCLK
DATA
CLK
MD0
READ_TRIGGER
TRIG
IBUF
READBACK
RIP
READ_DATA
MD1
OBUF
X1786
Figure 51: Readback Schematic Example
4-64
September 18, 1996 (Version 1.04)
Readback Options
Readback options are: Read Capture, Read Abort, and
Clock Select. They are set with MakeBits, the bitstream
generation software.
I/O
I/O
PROGRAMMABLE
INTERCONNECT
When the Read Capture option is not selected, the values
of the capture bits reflect the configuration data originally
written to those memory locations.
If the RAM capability of the CLBs is used, RAM data are
available in readback, since they directly overwrite the F
and G function-table configuration of the CLB.
RDBK.TRIG is located in the lower-left corner of the device,
as shown in Figure 52.
Read Abort
When the Read Abort option is selected, a High-to-Low
transition on RDBK.TRIG terminates the readback operation and prepares the logic to accept another trigger.
After an aborted readback, additional clocks (up to one
readback clock per configuration frame) may be required to
re-initialize the control logic. The status of readback is indicated by the output control net RDBK.RIP. RDBK.RIP is
High whenever a readback is in progress.
rdbk
I
When the Read Capture option is selected, the readback
data stream includes sampled values of CLB and IOB signals. The rising edge of RDBK.TRIG latches the inverted
values of the four CLB outputs, the IOB output flip-flops and
the input signals I1 and I2. Note that while the bits describing configuration (interconnect, function generators, and
RAM content) are not inverted, the CLB and IOB output signals are inverted.
TRIG
DATA
RIP
Read Capture
I/O
I/O
I/O
rdclk
X1787
Figure 52: READBACK Symbol in Graphical Editor
Violating the Maximum High and Low Time
Specification for the Readback Clock
The readback clock has a maximum High and Low time
specification. In some cases, this specification cannot be
met. For example, if a processor is controlling readback, an
interrupt may force it to stop in the middle of a readback.
This necessitates stopping the clock, and thus violating the
specification.
The specification is mandatory only on clocking data at the
end of a frame prior to the next start bit. The transfer mechanism will load the data to a shift register during the last six
clock cycles of the frame, prior to the start bit of the following frame. This loading process is dynamic, and is the
source of the maximum High and Low time requirements.
Clock Select
Therefore, the specification only applies to the six clock
cycles prior to and including any start bit, including the
clocks before the first start bit in the readback data stream.
At other times, the frame data is already in the register and
the register is not dynamic. Thus, it can be shifted out just
like a regular shift register.
CCLK is the default clock. However, the user can insert
another clock on RDBK.CLK. Readback control and data
are clocked on rising edges of RDBK.CLK. If readback
must be inhibited for security reasons, the readback control
nets are simply not connected.
The user must precisely calculate the location of the readback data relative to the frame. The system must keep
track of the position within a data frame, and disable interrupts before frame boundaries. Frame lengths and data
formats are listed in Table 21, Table 22 and Table 23.
RDBK.CLK is located in the lower right chip corner, as
shown in Figure 52.
Readback with the XChecker Cable
The XChecker Universal Download/Readback Cable and
Logic Probe uses the readback feature for bitstream verification. It can also display selected internal signals on the
PC or workstation screen, functioning as a low-cost in-circuit emulator.
September 18, 1996 (Version 1.04)
4-65
XC4000 Series Field Programmable Gate Arrays
Configuration Timing
The seven configuration modes are discussed in detail in
this section. Timing specifications are included.
Master Serial Mode
In Master Serial mode, the CCLK output of the lead FPGA
drives a Xilinx Serial PROM that feeds the FPGA DIN input.
Each rising edge of the CCLK output increments the Serial
PROM internal address counter. The next data bit is put on
the SPROM data output, connected to the FPGA DIN pin.
The lead FPGA accepts this data on the subsequent rising
CCLK edge.
The lead FPGA then presents the preamble data—and all
data that overflows the lead device—on its DOUT pin.
There is an internal pipeline delay of 1.5 CCLK periods,
which means that DOUT changes on the falling CCLK
edge, and the next FPGA in the daisy chain accepts data
on the subsequent rising CCLK edge.
In MakeBits, the user can specify Fast ConfigRate, which,
starting several bits into the first frame, increases the CCLK
frequency by a factor of eight. The value increases from
between 0.5 and 1.25 MHz, to a value between 4 and 10
MHz. (For low-voltage devices, the frequency can be up to
10% lower.) Be sure that the serial PROM and slaves are
fast enough to support this data rate. XC2000, XC3000/A,
and XC3100A devices do not support the Fast ConfigRate
option.
The SPROM CE input can be driven from either LDC or
DONE. Using LDC avoids potential contention on the DIN
pin, if this pin is configured as user-I/O, but LDC is then
restricted to be a permanently High user output after configuration. Using DONE can also avoid contention on DIN,
provided the early DONE option is invoked.
Figure 53 shows a full master/slave system. The leftmost
device is in Master Serial mode.
Master Serial mode is selected by a on the mode
pins (M2, M1, M0).
NOTE:
M2, M1, M0 can be shorted
to VCC if not used as I/O
NOTE:
M2, M1, M0 can be shorted
to Ground if not used as I/O
VCC
N/C
4.7 KΩ
4.7 KΩ
4.7 KΩ
4.7 KΩ
4.7 KΩ
4.7 KΩ
M0 M1
M2
N/C
DOUT
XC4000E/EX
MASTER
SERIAL
DIN
M0 M1
M2
DOUT
CCLK
VCC
XC1700D
4.7 KΩ
CCLK
M0 M1
M2
CLK
DIN
DATA
PROGRAM
LDC
CE
DONE
INIT
RESET/OE
+5 V
DIN
DOUT
CCLK
XC4000E/EX,
XC5200
XC3100A
SLAVE
SLAVE
VPP
CEO
PWRDN
PROGRAM
DONE
RESET
INIT
D/P
INIT
(Low Reset Option Used)
PROGRAM
X6608
Figure 53: Master Serial Mode Circuit Diagram
4-66
September 18, 1996 (Version 1.04)
CCLK
(Output)
2 TCKDS
1
Serial Data In
Serial DOUT
(Output)
TDSCK
n
n–3
n+1
n–2
n+2
n–1
n
X3223
CCLK
Description
DIN setup
DIN hold
1
2
Symbol
TDSCK
TCKDS
Min
20
0
Max
Units
ns
ns
Notes: 1. At power-up, Vcc must rise from 2.0 V to Vcc min in less than 25 ms, otherwise delay configuration by pulling PROGRAM
Low until Vcc is valid.
2. Master Serial mode timing is based on testing in slave mode.
Figure 54: Master Serial Mode Programming Switching Characteristics
September 18, 1996 (Version 1.04)
4-67
XC4000 Series Field Programmable Gate Arrays
Slave Serial Mode
In Slave Serial mode, an external signal drives the CCLK
input of the FPGA. The serial configuration bitstream must
be available at the DIN input of the lead FPGA a short setup
time before each rising CCLK edge.
The lead FPGA then presents the preamble data—and all
data that overflows the lead device—on its DOUT pin.
There is an internal delay of 0.5 CCLK periods, which
means that DOUT changes on the falling CCLK edge, and
the next FPGA in the daisy chain accepts data on the subsequent rising CCLK edge.
Figure 55 shows a full master/slave system. An XC4000Series device in Slave Serial mode should be connected as
shown in the third device from the left.
Slave Serial mode is selected by a on the mode pins
(M2, M1, M0). Slave Serial is the default mode if the mode
pins are left unconnected, as they have weak pull-up resistors during configuration.
NOTE:
M2, M1, M0 can be shorted
to VCC if not used as I/O
NOTE:
M2, M1, M0 can be shorted
to Ground if not used as I/O
VCC
N/C
4.7 KΩ
4.7 KΩ
4.7 KΩ
4.7 KΩ
4.7 KΩ
4.7 KΩ
M0 M1
M2
N/C
DOUT
XC4000E/EX
MASTER
SERIAL
M0 M1
M2
DOUT
CCLK
VCC
XC1700D
4.7 KΩ
CCLK
M0 M1
M2
DIN
CLK
DIN
DATA
PROGRAM
LDC
CE
DONE
INIT
RESET/OE
+5 V
DOUT
CCLK
XC4000E/EX,
XC5200
XC3100A
SLAVE
SLAVE
VPP
CEO
PWRDN
DIN
PROGRAM
DONE
RESET
INIT
D/P
INIT
(Low Reset Option Used)
PROGRAM
X6608
Figure 55: Slave Serial Mode Circuit Diagram
4-68
September 18, 1996 (Version 1.04)
DIN
Bit n
1 TDCC
Bit n + 1
2 TCCD
5 TCCL
CCLK
4 TCCH
DOUT
(Output)
3 TCCO
Bit n - 1
Bit n
X5379
CCLK
Note:
Description
DIN setup
DIN hold
DIN to DOUT
High time
Low time
Frequency
1
2
3
4
5
Symbol
TDCC
TCCD
TCCO
TCCH
TCCL
FCC
Min
20
0
Max
30
45
45
10
Units
ns
ns
ns
ns
ns
MHz
Configuration must be delayed until the INIT pins of all daisy-chained FPGAs are High.
Figure 56: Slave Serial Mode Programming Switching Characteristics
September 18, 1996 (Version 1.04)
4-69
XC4000 Series Field Programmable Gate Arrays
Master Parallel Modes
In the two Master Parallel modes, the lead FPGA directly
addresses an industry-standard byte-wide EPROM, and
accepts eight data bits just before incrementing or decrementing the address outputs.
The eight data bits are serialized in the lead FPGA, which
then presents the preamble data—and all data that overflows the lead device—on its DOUT pin. There is an internal delay of 1.5 CCLK periods, after the rising CCLK edge
that accepts a byte of data (and also changes the EPROM
address) until the falling CCLK edge that makes the LSB
(D0) of this byte appear at DOUT. This means that DOUT
changes on the falling CCLK edge, and the next FPGA in
the daisy chain accepts data on the subsequent rising
CCLK edge.
4.7KΩ
M0
The PROM address pins can be incremented or decremented, depending on the mode pin settings. This option
allows the FPGA to share the PROM with a wide variety of
microprocessors and microcontrollers. Some processors
must boot from the bottom of memory (all zeros) while others must boot from the top. The FPGA is flexible and can
load its configuration bitstream from either end of the memory.
Master Parallel Up mode is selected by a on the
mode pins (M2, M1, M0). The EPROM addresses start at
00000 and increment.
Master Parallel Down mode is selected by a on the
mode pins. The EPROM addresses start at 3FFFF and
decrement.
TO DIN OF OPTIONAL
DAISY-CHAINED FPGAS
HIGH
or
LOW
N/C
M1
M2
N/C
TO CCLK OF OPTIONAL
DAISY-CHAINED FPGAS
CCLK
NOTE:M0 can be shorted
to Ground if not used
as I/O.
DOUT
VCC
4.7KΩ
INIT
A17
...
A16
...
A15
...
A14
...
A13
...
M0
M1
DIN
EPROM
(8K x 8)
(OR LARGER)
A12
A11
A11
A10
A10
PROGRAM
A9
A9
D7
A8
A8
D6
A7
A7
D7
D5
A6
A6
D6
D4
A5
A5
D5
D3
A4
A4
D4
D2
A3
A3
D3
D1
A2
A2
D2
D0
A1
A1
D1
A0
A0
D0
DONE
OE
DOUT
CCLK
USER CONTROL OF HIGHER
ORDER PROM ADDRESS BITS
CAN BE USED TO SELECT BETWEEN
ALTERNATIVE CONFIGURATIONS
A12
M2
XC4000E/EX
SLAVE
PROGRAM
DONE
INIT
CE
DATA BUS
8
PROGRAM
X6697
Figure 57: Master Parallel Mode Circuit Diagram
4-70
September 18, 1996 (Version 1.04)
A0-A17
(output)
Address for Byte n
Address for Byte n + 1
1 TRAC
D0-D7
Byte
3 TRCD
2 TDRC
RCLK
(output)
7 CCLKs
CCLK
CCLK
(output)
DOUT
(output)
D6
D7
Byte n - 1
RCLK
Description
Delay to Address valid
Data setup time
Data hold time
1
2
3
Symbol
TRAC
TDRC
TRCD
Min
0
60
0
X6078
Max
200
Units
ns
ns
ns
Notes: 1. At power-up, Vcc must rise from 2.0 V to Vcc min in less than 25 ms, otherwise delay configuration by pulling PROGRAM
Low until Vcc is valid.
2. The first Data byte is loaded and CCLK starts at the end of the first RCLK active cycle (rising edge).
This timing diagram shows that the EPROM requirements are extremely relaxed. EPROM access time can be longer than
500 ns. EPROM data output has no hold-time requirements.
Figure 58: Master Parallel Mode Programming Switching Characteristics
September 18, 1996 (Version 1.04)
4-71
XC4000 Series Field Programmable Gate Arrays
Synchronous Peripheral Mode
Synchronous Peripheral mode can also be considered
Slave Parallel mode. An external signal drives the CCLK
input(s) of the FPGA(s). The first byte of parallel configuration data must be available at the Data inputs of the lead
FPGA a short setup time before the rising CCLK edge.
Subsequent data bytes are clocked in on every eighth consecutive rising CCLK edge.
The same CCLK edge that accepts data, also causes the
RDY/BUSY output to go High for one CCLK period. The pin
name is a misnomer. In Synchronous Peripheral mode it is
really an ACKNOWLEDGE signal. Synchronous operation
does not require this response, but it is a meaningful signal
for test purposes. Note that RDY/BUSY is pulled High with
a high-impedance pullup prior to INIT going High.
The lead FPGA serializes the data and presents the preamble data (and all data that overflows the lead device) on
its DOUT pin. There is an internal delay of 1.5 CCLK periods, which means that DOUT changes on the falling CCLK
edge, and the next FPGA in the daisy chain accepts data
on the subsequent rising CCLK edge.
In order to complete the serial shift operation, 10 additional
CCLK rising edges are required after the last data byte has
been loaded, plus one more CCLK cycle for each daisychained device.
Synchronous Peripheral mode is selected by a on
the mode pins (M2, M1, M0).
NOTE:
M2 can be shorted to Ground
if not used as I/O
N/C
M0 M1
N/C
4.7 kΩ
M2
M0 M1
CCLK
CLOCK
OPTIONAL
DAISY-CHAINED
FPGAs
8
DATA BUS
D0-7
DOUT
VCC
4.7 kΩ
M2
CCLK
XC4000E/EX
SYNCHRONOUS
PERIPHERAL
DIN
DOUT
XC4000E/EX
SLAVE
RDY/BUSY
CONTROL
SIGNALS
INIT
DONE
INIT
DONE
4.7 kΩ
PROGRAM
PROGRAM
PROGRAM
X5996
Figure 59: Synchronous Peripheral Mode Circuit Diagram
4-72
September 18, 1996 (Version 1.04)
CCLK
INIT
BYTE
0
BYTE
1
BYTE 0 OUT
DOUT
0
1
2
3
4
BYTE 1 OUT
5
6
7
0
1
RDY/BUSY
X6096
CCLK
Description
INIT (High) setup time
D0 - D7 setup time
D0 - D7 hold time
CCLK High time
CCLK Low time
CCLK Frequency
Symbol
TIC
TDC
TCD
TCCH
TCCL
FCC
Min
5
60
0
50
60
Max
8
Units
µs
ns
ns
ns
ns
MHz
Notes: 1. Peripheral Synchronous mode can be considered Slave Parallel mode. An external CCLK provides timing, clocking in the
first data byte on the second rising edge of CCLK after INIT goes High. Subsequent data bytes are clocked in on every
eighth consecutive rising edge of CCLK.
2. The RDY/BUSY line goes High for one CCLK period after data has been clocked in, although synchronous operation does
not require such a response.
3. The pin name RDY/BUSY is a misnomer. In Synchronous Peripheral mode this is really an ACKNOWLEDGE signal.
4. Note that data starts to shift out serially on the DOUT pin 0.5 CCLK periods after it was loaded in parallel. Therefore,
additional CCLK pulses are clearly required after the last byte has been loaded.
Figure 60: Synchronous Peripheral Mode Programming Switching Characteristics
September 18, 1996 (Version 1.04)
4-73
XC4000 Series Field Programmable Gate Arrays
Asynchronous Peripheral Mode
The READY/BUSY handshake can be ignored if the delay
from any one Write to the end of the next Write is guaranteed to be longer than 10 CCLK periods.
Write to FPGA
Asynchronous Peripheral mode uses the trailing edge of
the logic AND condition of WS and CS0 being Low and RS
and CS1 being High to accept byte-wide data from a microprocessor bus. In the lead FPGA, this data is loaded into a
double-buffered UART-like parallel-to-serial converter and
is serially shifted into the internal logic.
Status Read
The logic AND condition of the CS0, CS1and RS inputs
puts the device status on the Data bus.
•
•
•
The lead FPGA presents the preamble data (and all data
that overflows the lead device) on its DOUT pin. The RDY/
BUSY output from the lead FPGA acts as a handshake signal to the microprocessor. RDY/BUSY goes Low when a
byte has been received, and goes High again when the
byte-wide input buffer has transferred its information into
the shift register, and the buffer is ready to receive new
data. A new write may be started immediately, as soon as
the RDY/BUSY output has gone Low, acknowledging
receipt of the previous data. Write may not be terminated
until RDY/BUSY is High again for one CCLK period. Note
that RDY/BUSY is pulled High with a high-impedance pullup prior to INIT going High.
D7 High indicates Ready
D7 Low indicates Busy
D0 through D6 go unconditionally High
It is mandatory that the whole start-up sequence be started
and completed by one byte-wide input. Otherwise, the pins
used as Write Strobe or Chip Enable might become active
outputs and interfere with the final byte transfer. If this
transfer does not occur, the start-up sequence is not completed all the way to the finish (point F in Figure 49 on page
61).
In this case, at worst, the internal reset is not released. At
best, Readback and Boundary Scan are inhibited. The
length-count value, as generated by MakeBits and
MakePROM, ensures that these problems never occur.
Although RDY/BUSY is brought out as a separate signal,
microprocessors can more easily read this information on
one of the data lines. For this purpose, D7 represents the
RDY/BUSY status when RS is Low, WS is High, and the
two chip select lines are both active.
The length of the BUSY signal depends on the activity in
the UART. If the shift register was empty when the new
byte was received, the BUSY signal lasts for only two CCLK
periods. If the shift register was still full when the new byte
was received, the BUSY signal can be as long as nine
CCLK periods.
Asynchronous Peripheral mode is selected by a on
the mode pins (M2, M1, M0).
Note that after the last byte has been entered, only seven of
its bits are shifted out. CCLK remains High with DOUT
equal to bit 6 (the next-to-last bit) of the last byte entered.
N/C
N/C
N/C
4.7 kΩ
M0
8
DATA
BUS
D0–7
M1
CCLK
DOUT
4.7 kΩ
...
4.7 kΩ
M1
M2
CCLK
OPTIONAL
DAISY-CHAINED
FPGAs
VCC
ADDRESS
BUS
M0
M2
ADDRESS
DECODE
LOGIC
DOUT
DIN
CS0
XC4000E/EX
ASYNCHRONOUS
PERIPHERAL
XC4000E/EX
SLAVE
CS1
RS
WS
CONTROL
SIGNALS
RDY/BUSY
REPROGRAM
INIT
INIT
DONE
DONE
PROGRAM
PROGRAM
4.7 kΩ
X6696
Figure 61:
4-74
Asynchronous Peripheral Mode Circuit Diagram
September 18, 1996 (Version 1.04)
Write to LCA
Read Status
RS, CS0
WS/CS0
RS, CS1
WS, CS1
1
TCA
2
3
TDC
TCD
4
7
READY
BUSY
D0-D7
D7
CCLK
TWTRB 4
6
TBUSY
RDY/BUSY
Previous Byte D6
DOUT
D7
D0
D1
D2
X6097
Write
RDY
Description
Effective Write time
(CS0, WS=Low; RS, CS1=High)
DIN setup time
DIN hold time
RDY/BUSY delay after end of
Write or Read
RDY/BUSY active after beginning
of Read
RDY/BUSY Low output (Note 4)
1
2
3
4
Symbol
TCA
TDC
TCD
TWTRB
Min
100
60
0
7
6
TBUSY
Max
2
Units
ns
60
ns
ns
ns
60
ns
9
CCLK
periods
Notes: 1. Configuration must be delayed until the INIT pins of all daisy-chained FPGAs are High.
2. The time from the end of WS to CCLK cycle for the new byte of data depends on the completion of previous byte processing
and the phase of the internal timing generator for CCLK.
3. CCLK and DOUT timing is tested in slave mode.
4. TBUSY indicates that the double-buffered parallel-to-serial converter is not yet ready to receive new data. The shortest TBUSY
occurs when a byte is loaded into an empty parallel-to-serial converter. The longest TBUSY occurs when a new word is
loaded into the input register before the second-level buffer has started shifting out data.
This timing diagram shows very relaxed requirements. Data need not be held beyond the rising edge of WS. RDY/BUSY will
go active within 60 ns after the end of WS. A new write may be asserted immediately after RDY/BUSY goes Low, but write
may not be terminated until RDY/BUSY has been High for one CCLK period.
Figure 62: Asynchronous Peripheral Mode Programming Switching Characteristics
September 18, 1996 (Version 1.04)
4-75
XC4000 Series Field Programmable Gate Arrays
Express Mode (XC4000EX only)
nized as High, and remains Low until the device’s configuration memory is full. DOUT is then pulled High to signal
the next device in the chain to accept the configuration data
on the D0-D7 bus.
Express mode is similar to Slave Serial mode, except that
data is processed one byte per CCLK cycle instead of one
bit per CCLK cycle. An external source is used to drive
CCLK, while byte-wide data is loaded directly into the configuration data shift registers. A CCLK frequency of 1 MHz
is equivalent to a 8 MHz serial rate, because eight bits of
configuration data are loaded per CCLK cycle. Express
mode does not support CRC error checking, but does support constant-field error checking.
The DONE pins of all devices in the chain should be tied
together, with one or more active internal pull-ups. If a
large number of devices are included in the chain, deactivate some of the internal pull-ups, since the Low-driving
DONE pin of the last device in the chain must sink the current from all pull-ups in the chain. The DONE pull-up is
activated by default. It can be deactivated using a MakeBits option.
In Express mode, an external signal drives the CCLK input
of the FPGA device. The first byte of parallel configuration
data must be available at the D inputs of the FPGA a short
setup time before the second rising CCLK edge. Subsequent data bytes are clocked in on each consecutive rising
CCLK edge.
XC4000EX devices in Express mode are always synchronized to DONE. The device becomes active after DONE
goes High. DONE is an open-drain output. With the DONE
pins tied together, therefore, the external DONE signal
stays low until all devices are configured, then all devices in
the daisy chain become active simultaneously. If the DONE
pin of a device is left unconnected, the device becomes
active as soon as that device has been configured.
XC5200 devices in the chain should be configured as synchronized to DONE (MakeBits option CCLK_SYNC or
UCLK_SYNC), and their DONE pins wired together with
those of the XC4000EX devices.
Express mode is only supported by the XC4000EX and
XC5200 families. It may not be used, therefore, when an
XC4000EX or XC5200 device is daisy-chained with
devices from other Xilinx families.
If the first device is configured in Express mode, additional
devices may be daisy-chained only if every device in the
chain is also configured in Express mode. CCLK pins are
tied together and D0-D7 pins are tied together for all
devices along the chain. A status signal is passed from
DOUT to CS1 of successive devices along the chain. The
lead device in the chain has its CS1 input tied High (or floating, since there is an internal pullup). Frame data is
accepted only when CS1 is High and the device’s configuration memory is not already full. The status pin DOUT is
pulled Low two internal-oscillator cycles after INIT is recog-
Express mode must be specified as an option to the MakeBits program, which generates the bitstream. The Express
mode bitstream is not compatible with the other six configuration modes.
Express mode is selected by a on the mode pins
(M2, M1, M0).
VCC
NOTE:
M2, M1, M0 can be shorted
to Ground if not used as I/O
4.7KΩ
8
M0
M1
CS1
DATA BUS
8
M2
8
XC4000EX/
XC5200
4.7KΩ
PROGRAM
PROGRAM
INIT
INIT
CCLK
M1
CS1
DOUT
D0-D7
VCC
M0
M2
To Additional
Optional
Daisy-Chained
Devices
DOUT
D0-D7
Optional
Daisy-Chained
XC4000EX/
XC5200
PROGRAM
DONE
INIT
DONE
CCLK
To Additional
Optional
Daisy-Chained
Devices
CCLK
X6611
Figure 63: Express Mode Circuit Diagram
4-76
September 18, 1996 (Version 1.04)
CCLK
Description
INIT (High) setup time
D0 - D7 setup time
D0 - D7 hold time
CCLK High time
CCLK Low time
CCLK Frequency
Symbol
TIC
TDC
TCD
TCCH
TCCL
FCC
Min
0
-
Max
-
Units
µs
ns
ns
ns
ns
MHz
Preliminary
CCLK
1
TIC
INIT
TCD 3
2 T
DC
D0-D7
BYTE
0
BYTE
1
BYTE
2
BYTE
3
DOUT
FPGA Filled
Internal INIT
RDY/BUSY
CS1
X6710
Note: If not driven by the preceding DOUT, CS1 must remain High until the device is fully configured.
Figure 64: Express Mode Programming Switching Characteristics
September 18, 1996 (Version 1.04)
4-77
XC4000 Series Field Programmable Gate Arrays
Table 24: Pin Functions During Configuration
SLAVE
SERIAL
M2(HIGH) (I)
M1(HIGH) (I)
M0(HIGH) (I)
HDC (HIGH)
LDC (LOW)
INIT
DONE
PROGRAM (I)
CCLK (I)
DIN (I)
DOUT
TDI
TCK
TMS
TDO
CONFIGURATION MODE
SYNCH.
ASYNCH.
MASTER
MASTER
MASTER
PERIPHPERIPHPARALLEL
PARALLEL UP
EXPRESS
SERIAL
ERAL
ERAL
DOWN
M2(LOW) (I)
M2(LOW) (I)
M2(HIGH) (I)
M2(HIGH) (I)
M2(HIGH) (I)
M2(LOW) (I)
M1(LOW) (I)
M1(HIGH) (I)
M1(LOW) (I)
M1(HIGH) (I)
M1(LOW) (I)
M1(HIGH) (I)
M0(LOW) (I)
M0(HIGH) (I)
M0(HIGH) (I)
M0(LOW) (I)
M0(LOW) (I)
M0(HIGH) (I)
HDC (HIGH)
HDC (HIGH)
HDC (HIGH)
HDC (HIGH)
HDC (HIGH)
HDC (HIGH)
LDC (LOW)
LDC (LOW)
LDC (LOW)
LDC (LOW)
LDC (LOW)
LDC (LOW)
INIT
INIT
INIT
INIT
INIT
INIT
DONE
DONE
DONE
DONE
DONE
DONE
PROGRAM (I) PROGRAM (I) PROGRAM (I) PROGRAM (I) PROGRAM (I) PROGRAM (I)
CCLK (O)
CCLK (I)
CCLK (O)
CCLK (O)
CCLK (O)
CCLK (I)
RDY/BUSY (O) RDY/BUSY (O)
RCLK (O)
RCLK (O)
RS (I)
CS0 (I)
DATA 7 (I)
DATA 7 (I)
DATA 7 (I)
DATA 7 (I)
DATA 7 (I)
DATA 6 (I)
DATA 6 (I)
DATA 6 (I)
DATA 6 (I)
DATA 6 (I)
DATA 5 (I)
DATA 5 (I)
DATA 5 (I)
DATA 5 (I)
DATA 5 (I)
DATA 4 (I)
DATA 4 (I)
DATA 4 (I)
DATA 4 (I)
DATA 4 (I)
DATA 3 (I)
DATA 3 (I)
DATA 3 (I)
DATA 3 (I)
DATA 3 (I)
DATA 2 (I)
DATA 2 (I)
DATA 2 (I)
DATA 2 (I)
DATA 2 (I)
DATA 1 (I)
DATA 1 (I)
DATA 1 (I)
DATA 1 (I)
DATA 1 (I)
DIN (I)
DATA 0 (I)
DATA 0 (I)
DATA 0 (I)
DATA 0 (I)
DATA 0 (I)
DOUT
DOUT
DOUT
DOUT
DOUT
DOUT
TDI
TDI
TDI
TDI
TDI
TDI
TCK
TCK
TCK
TCK
TCK
TCK
TMS
TMS
TMS
TMS
TMS
TMS
TDO
TDO
TDO
TDO
TDO
TDO
WS (I)
A0
A0
A1
A1
CS1
A2
A2
A3
A3
A4
A4
A5
A5
A6
A6
A7
A7
A8
A8
A9
A9
A10
A10
A11
A11
A12
A12
A13
A13
A14
A14
A15
A15
A16
A16
A17
A17
A18*
A18*
A19*
A19*
A20*
A20*
A21*
A21*
USER
OPERATION
(I)
(O)
(I)
I/O
I/O
I/O
DONE
PROGRAM
CCLK (I)
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
SGCK4-GCK5-I/O
TDI-I/O
TCK-I/O
TMS-I/O
TDO-(O)
I/O
PGCK4-GCK6-I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
SGCK1-GCK7-I/O
PGCK1-GCK8-I/O
I/O
I/O
I/O
I/O
I/O
ALL OTHERS
* XC4000EX only
Notes 1. A shaded table cell represents a 50 kΩ - 100 kΩ pull-up before and during configuration.
2. (I) represents an input; (O) represents an output.
3. INIT is an open-drain output during configuration.
4-78
September 18, 1996 (Version 1.04)
Configuration Switching Characteristics
T POR
Vcc
RE-PROGRAM
>300 ns
PROGRAM
T PI
INIT
T ICCK
TCCLK
CCLK OUTPUT or INPUT