OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
LMX2471 3.6 GHz Delta-Sigma Fractional-N PLL with 1.7 GHz Integer-N PLL
Check for Samples: LMX2471
FEATURES
1
•
2
•
•
•
•
•
Low in-band phase noise and low fractional
spurs
12 bit or 22 bit selectable fractional modulus
Up to 4th order programmable delta-sigma
modulator
Enhanced Anti-Cycle Slip Fastlock Circuitry
– Fastlock
– Cycle slip reduction
– Integrated timeout counters
Digital lock detect output
Prescalers allow wide range of N values
– RF PLL: 16/17/20/21
•
•
•
•
•
– IF PLL: 8/9 or 16/17
Crystal Reference Frequency up to 110 MHz
On-chip crystal reference frequency doubler.
Phase Comparison Frequency up to 50 MHz
Hardware and software power-down control
Ultra low consumption: ICC = 5.6 mA (typical)
APPLICATIONS
•
•
•
•
Cellular Phones and Base Stations
Applications requiring fine frequency
resolution
Satellite and Cable TV Tuners
WLAN Standards
DESCRIPTION
The LMX2471 is a low power, high performance delta-sigma fractional-N PLL with an auxiliary integer-N PLL.
The device is fabricated using National Semiconductor’s advanced BiCMOS process.
With delta-sigma architecture, fractional spur compensation is achieved with noise shaping capability of the deltasigma modulator and the inherent low pass filtering of the PLL loop filter. Fractional spurs at lower frequencies
are pushed to higher frequencies outside the loop bandwidth. Unlike analog compensation, the digital feedback
techniques used in the LMX2471 are highly resistant to changes in temperature and variations in wafer
processing. With delta-sigma architecture, the ability to push close in spur and phase noise energy to higher
frequencies is a direct function of the modulator order. The higher the order, the more this energy can be spread
to higher frequencies. The LMX2471 has a programmable modulator up to order four, which allows the designer
to select the optimum modulator order to fit the phase noise, spur, and lock time requirements of the system.
Programming is fast and simple. Serial data is transferred into the LMX2471 via a three line MICROWIRE
interface (Data, Clock, Load Enable). Nominal supply voltage is 2.5 V. The LMX2471 features a typical current
consumption of 5.6 mA at 2.5 V. The LMX2471 is available in a 24 lead 3.5 X 4.5 X 0.6 mm package.
1
2
Please be aware that an important notice concerning availability, standard warranty, and use in critical applications of
Texas Instruments semiconductor products and disclaimers thereto appears at the end of this data sheet.
All trademarks are the property of their respective owners.
PRODUCTION DATA information is current as of publication date.
Products conform to specifications per the terms of the Texas
Instruments standard warranty. Production processing does not
necessarily include testing of all parameters.
Copyright © 2004, Texas Instruments Incorporated
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
Functional Block Diagram
IF
Fastlock
IF N Divider
B Counter
8/9
or
16/17
Prescaler
A Counter
FinIF
Phase
Comp
FLoutIF
Charge
Pump
CPoutIF
Ftest/LD
MUX
Ftest/LD
Charge
Pump
CPoutRF
RF Fastlock
FLoutRF
ENOSC
IF
LD
OSCin
IF R
Divider
OSCout*
VccRF
RF R
Divider
1X / 2X
RF LD
VddRF
VccIF
VddIF
RF N Divider
C Counter
16/17/20/21
B Counter
Prescaler
A Counter
FinRF
FinRF*
Phase
Comp
EN
CLK
DATA
LE
GND
GND
GND
GND
GND
6'
Compensation
MICROWIRE
Interface
Connection Diagram
FLoutRF
VddRF
OSCin
Figure 1. 24-Pin CSP (SLE) Package
24
23
22
2
20
FLoutIF
GND
3
19
CPoutIF
GND
4
18
GND
FinRF
5
17
FinIF
FinRF*
6
16
GND
VccRF
7
15
VccIF
EN
8
14
VddIF
ENOSC
9
13
Ftest/LD
10
11
12
LE
GND
DATA
1
CLK
CPoutRF
21
OSCout*
Pin Functions
Pin Descriptions
2
Pin #
Pin Name
I/O
1
CPoutRF
O
RF charge pump output.
Pin Description
2
GND
-
Ground
3
GND
-
RF Ground
4
GND
-
Ground for RF PLL digital circuitry.
5
FinRF
I
RF prescaler input. Small signal input from the VCO.
6
FinRF*
I
RF prescaler complimentary input. For single-ended operation, a bypass capacitor should be placed
as close as possible to this pin and be connected directly to the ground plane.
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
Pin Descriptions (continued)
Pin #
Pin Name
7
VccRF
I/O
Pin Description
8
EN
I
Chip enable input. High impedance CMOS input. When EN is high, the chip is powered up, otherwise
it is powered down.
RF PLL power supply voltage input. Must be equal to VccIF . May range from 2.35V to 2.75V.
Bypass capacitors should be placed as close as possible to this pin and be connected directly to the
ground plane.
9
ENOSC
I
This pin should be grounded for normal operation.
10
CLK
I
MICROWIRE Clock. High impedance CMOS Clock input. Data for the various counters is clocked
into the 24 bit shift register on the rising edge.
11
DATA
I
MICROWIRE Data. High impedance binary serial data input.
12
LE
13
Ftest/LD
O
Test frequency output / Lock Detect
14
VddIF
-
Digital power supply for IF PLL
15
VccIF
-
IF power supply voltage input. Must be equal to VccRF. Input may range from 2.35 V to 2.75 V.
Bypass capacitors should be placed as close as possible to this pin and be connected directly to the
ground plane.
16
GND
-
Ground for RF PLL digital circuitry.
17
FinIF
I
IF prescaler input. Small signal input from the VCO.
MICROWIRE Load Enable. High impedance CMOS input. Data stored in the shift registers is loaded
into the internal latches when LE goes HIGH
18
GND
-
Digital ground for IF PLL
19
CPoutIF
O
IF PLL charge pump output
20
FLoutIF
O
IF Fastlock Output. Also functions as Programmable TRI-STATE CMOS output.
21
OSCout*
I/O
Complementary reference input or oscillator output.
22
OSCin
I
Reference input
23
VddRF
-
Digital power supply for RF PLL
24
FLoutRF
O
RF Fastlock Output. Also functions as Programmable TRI-STATE CMOS output.
These devices have limited built-in ESD protection. The leads should be shorted together or the device placed in conductive foam
during storage or handling to prevent electrostatic damage to the MOS gates.
Absolute Maximum Ratings
(1) (2)
Parameter
Power Supply Voltage
Symbol
Value
Min
Typ
Max
Units
VCC
-0.3
3.0
V
VDD
VCC
VCC
V
Vi
-0.3
VCC + 0.3
V
Storage Temperature Range
Ts
-65
+150
°C
Lead Temperature (Solder 4 sec.)
TL
+260
°C
Voltage on any pin with GND =VSS = 0V
(1)
(2)
“Absolute Maximum Ratings” indicate limits beyond which damage to the device may occur. "Recommended Operating Conditions"
indicate conditions for which the device is intended to be functional, but do not guarantee specific performance limits. For guaranteed
specifications and test conditions, see the Electrical Characteristics. The guaranteed specifications apply only for the test conditions
listed. Note also that these maximum ratings imply that the voltage at all the power supply pins of VccRF, VccIF, VddRF, and VddIF are
the same. VCCwill be used to refer to the voltage at these pins.
This Device is a high performance RF integrated circuit with an ESD rating < 2 kV and is ESD sensitive. Handling and assembly of this
device should only be done at ESD-free workstations.
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
3
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
Recommended Operating Conditions
Parameter
Power Supply Voltage
Symbol
(1)
Operating Temperature
(1)
4
Value
Min
Typ
Max
Units
VCC
2.25
2.75
V
VDD
VCC
VCC
V
TA
-40
+85
°C
“Absolute Maximum Ratings” indicate limits beyond which damage to the device may occur. "Recommended Operating Conditions"
indicate conditions for which the device is intended to be functional, but do not guarantee specific performance limits. For guaranteed
specifications and test conditions, see the Electrical Characteristics. The guaranteed specifications apply only for the test conditions
listed. Note also that these maximum ratings imply that the voltage at all the power supply pins of VccRF, VccIF, VddRF, and VddIF are
the same. VCCwill be used to refer to the voltage at these pins.
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
Electrical Characteristics
(VCC = 2.5V; -40°C ≤ TA ≤ +85°C unless otherwise specified)
Symbol
Parameter
Conditions
Value
Min
Typ
Max
Units
Icc PARAMETERS
ICCRF
ICCIF
ICCTOTAL
ICCPD
Power Supply Current, RF
Synthesizer
IF PLL OFF
RF PLL ON
Charge Pump TRI-STATE
OSC=0
3.6
5.7
mA
Power Supply Current, IF
Synthesizer
IF PLL ON
RF PLL OFF
Charge Pump TRI-STATE
OSC=0
2.0
2.7
mA
Power Supply Current, Entire
Synthesizer
IF PLL ON
RF PLL ON
Charge Pump TRI-STATE
OSC=0
5.6
8.5
mA
Power Down Current
EN = ENOSC = 0V
CLK, DATA, LE = 0V
1
15
µA
RF SYNTHESIZER PARAMETERS
fFinRF
Operating Frequency
500
3600
MHz
pFinRF
Input Sensitivity
-15
0
dBm
fCOMP
Phase Detector Frequency
50
MHz
ICPoutRFSRCE
RF Charge Pump Source
Current
RF_CPG = 0
VCPoutRF=VCC /2
100
µA
RF_CPG = 1
VCPoutRF=VCC/2
200
µA
...
ICPoutRFSINK
RF Charge Pump Sink Current
...
µA
RF_CPG = 15
VCPoutRF=VCC/2
1600
µA
RF_CPG = 0
VCPoutRF=VCC/2
-100
µA
RF_CPG = 1
VCPoutRF=VCC/2
-200
µA
...
µA
-1600
µA
...
RF_CPG = 15
VCPoutRF=VCC/2
ICPoutRFTRI
RF Charge Pump TRI-STATE
Current Magnitude
0.4 ≤ VCPoutRF ≤ VCC -0.4
2
10
nA
ICPoutRF%MIS
RF CP Sink vs. CP Source
Mismatch
VCPoutRF = VCC/2
TA = 25°C
3
10
%
ICPoutRF%V
RF CP Current vs. CP Voltage
0.4 ≤ VCPoutRF ≤ VCC -0.4
TA = 25°C
5
15
%
ICPoutRF%TEMP
RF CP Current vs. Temperature
VCPoutRF = VCC/2
8
%
IF SYNTHESIZER PARAMETERS
fFinIF
Operating Frequency
250
1700
MHz
pFinIF
IF Input Sensitivity
-15
0
dBm
fCOMP
Phase Detector Frequency
10
MHz
ICPoutIFSRCE
IF Charge Pump Source Current
ICPoutIFSINK
IF Charge Pump Sink Current
IF_CPG = 0
VCPoutIF = VCC /2
1
mA
IF_CPG = 1
VCPoutIF = VCC/2
4
mA
IF_CPG = 0
VCPoutIF = VCC/2
-1
mA
IF_CPG = 1
VCPoutIF = VCC/2
-4
mA
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
5
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
Electrical Characteristics (continued)
(VCC = 2.5V; -40°C ≤ TA ≤ +85°C unless otherwise specified)
Symbol
Parameter
Conditions
Value
Min
Typ
Max
10
ICPoutIFTRI
IF Charge Pump TRI-STATE
Current Magnitude
0.4 ≤ VCPoutIF ≤ VCC RF -0.4
2
ICPoutIF%MIS
IF CP Sink vs. CP Source
Mismatch
VCPoutIF = VCC/2
TA = 25°C
3
ICPoutIF%V
IF CP Current vs. CP Voltage
0.4 ≤ VCPoutIF ≤ VCC -0.4
TA = 25°C
8
ICPoutIF%TEMP
IF CP Current vs. Temperature
VCPoutIF = VCC/2
8
Units
nA
%
15
%
%
OSCILLATOR PARAMETERS
fOSCin
Oscillator Operating Frequency
vOSCin
Oscillator Input Sensitivity
IOSCin
Oscillator Input Current
OSC2X = 0
5
110
MHz
OSC2X = 1
5
20
MHz
OSC=0
0.5
VCC
V
-100
100
µA
1.6
VCC
V
DIGITAL INTERFACE (DATA, CLK, LE, EN, ENRF, Ftest/LD, FLoutRF, FLoutIF)
VIH
High-Level Input Voltage
VIL
Low-Level Input Voltage
IIH
High-Level Input Current
VIH = VCC
IIL
Low-Level Input Current
VIL = 0 V
VOH
High-Level Output Voltage
IOH = -500 µA
VOL
Low-Level Output Voltage
IOL = 500 µA
0.4
V
-1.0
1.0
µA
-1.0
1.0
µA
VCC-0.4
V
0.4
V
MICROWIRE INTERFACE TIMING
TCS
Data to Clock Set Up Time
See Microwire Input Timing
50
ns
TCH
Data to Clock Hold Time
See Microwire Input Timing
10
ns
TCWH
Clock Pulse Width High
See Microwire Input Timing
50
ns
TCWL
Clock Pulse Width Low
See Microwire Input Timing
50
ns
TES
Clock to Load Enable Set Up
Time
See Microwire Input Timing
50
ns
TEW
Load Enable Pulse Width
See Microwire Input Timing
50
ns
RF Synthesizer Normalized
Phase Noise Contribution (1)
RF_CPG = 0
-200
dBc/Hz
RF_CPG = 3
-206
dBc/Hz
RF_CPG = 7
-208
dBc/Hz
RF_CPG = 15
-210
dBc/Hz
Applies to both low and high current
modes
OSC=0
-214
dBc/Hz
PHASE NOISE
LF1HzRF
LF1HzIF
IF Synthesizer Normalized
Phase Noise Contribution
(1)
(1)
6
Normalized Phase Noise Contribution is defined as: LN(f) = L(f) – 20log(N) – 10log(fCOMP) where L(f) is defined as the single side band
phase noise measured at an offset frequency, f, in a 1 Hz Bandwidth. The offset frequency, f, must be chosen sufficiently smaller than
the PLL loop bandwidth, yet large enough to avoid substantial phase noise contribution from the reference source. The offset chosen
was 4 KHz.
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
Figure 2. MICROWIRE INPUT TIMING DIAGRAM
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
7
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
: Sensitivity
RF N Counter Sensitivity
TA = 25°C
RF N Counter Sensitivity
VCC = 2.5 V
20
20
TA =85o C
VCC =2.75 V
10
10
-10
-20
VCC =2.25 V
VCC =2.75 V
-30
TA =-40o C
TA =25o C
0
INPUT POWER (dBm)
INPUT POWER (dBm)
VCC =2.25 V
VCC =2.5 V
0
-10
-20
TA =-40o C
TA =85o C
-30
-40
-40
TA =25o C
VCC =2.5 V
-50
-50
0
1000
2000
3000
4000
0
1000
2000
FREQUENCY (MHz)
IF N Counter Sensitivity
TA = 25°C
4000
1500
2000
IF N Counter Sensitivity
VCC = 2.5 V
20
20
TA = 25o C
VCC =2.75 V
10
10
TA = -40o C
TA = 85o C
VCC =2.25 V
VCC =2.5 V
0
INPUT POWER (dBm)
0
INPUT POWER (dBm)
3000
FREQUENCY (MHz)
-10
-20
VCC =2.75 V
-30
-10
-20
TA = 85o C
TA = 25o C
-30
-40
-40
VCC =2.25 V
VCC =2.5 V
TA = -40o C
-50
-50
0
500
1000
1500
2000
0
500
1000
FREQUENCY (MHz)
FREQUENCY (MHz)
OSCin Counter Sensitivity
OSC=0
TA = 25° C
OSCin Counter Sensitivity
OSC=0
VCC= 2.5 V
20
20
10
10
VCC =2.25 V, 2.5 V, and 2.75 V
TA = -40o C, 25o C, and 85o C
0
INPUT POWER (dBm)
INPUT POWER (dBm)
0
-10
-20
VCC =2.75 V
-30
VCC =2.5 V
-10
-20
TA = 85o C
-30
TA = 25o C
VCC =2.25 V
-40
-40
TA = -40o C
-50
0
50
100
FREQUENCY (MHz)
150
200
-50
0
50
100
150
200
FREQUENCY (MHz)
8
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
: Sensitivity (continued)
: FinRF Input Impedance
Marker 1:
500 MHz
Marker 2:
1000 MHz
Marker 3:
2000 MHz
1
2
Marker 4:
4000 MHz
3
Start 500 MHz
4
Stop 5000 MHz
FinRF Input Impedance (VCC=2.5 V, TA=25° C)
Frequency (MHz)
Real (Ohms)
Imaginary (Ohms)
500
389.9
-158.3
750
270.8
-186.8
1000
160.4
-172.3
1250
95.7
-144.7
1500
61.1
-120.0
1750
44.0
-104.5
2000
36.2
-95.6
2250
32.6
-90.4
2500
32.0
-86.3
2750
30.7
-80.5
3000
28.3
-74.6
3250
26.3
-65.2
3500
22.6
-57.4
3750
18.7
-49.9
4000
16.7
-44.1
: FinIF Input Impedance
Marker 1:
500 MHz
Marker 2:
1000 MHz
Marker 3:
2000 MHz
1
2
Marker 4:
2500 MHz
Start 500 MHz
Stop 2500 MHz
4
3
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
9
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
FinIF Input Impedance (VCC=2.5 V, TA=25°C)
Freqeuncy (MHz)
Real (Ohms)
Imaginary (Ohms)
500
377.4
-174.6
600
325.4
-186.6
700
272.7
-192.6
800
222.6
-191.1
900
178.2
-182.4
1000
143.0
-170.6
1100
113.8
-157.1
1200
92.6
-144.5
1300
76.5
-133.0
1400
64.1
-123.0
1500
55.2
-114.9
1600
48.5
-108.2
1700
43.3
-102.9
1800
39.4
-98.4
1900
36.4
-94.6
2000
34.5
-91.6
: OSCin Input Impedance
MAGNITUDE OF INPUT IMPEDANCE (:)
6000
5000
4000
3000
2000
1000
0
0
20
40
60
80
100
120
FREQUENCY (MHz)
OSCin Input Impedance (VCC=2.5 V, TA=25 °C)
10
Frequency (MHz)
Real (Ohms)
Imaginary (Ohms)
Magnitude (Ohms)
50
2200
-4700
5189
10
710
-2700
2792
20
229
-1500
1517
30
133
-988
997
40
93
-752
758
50
74
-606
611
60
62
-505
509
70
53
-435
438
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
OSCin Input Impedance (VCC=2.5 V, TA=25 °C)
Frequency (MHz)
Real (Ohms)
Imaginary (Ohms)
Magnitude (Ohms)
80
49
-382
385
90
45
-341
344
100
42
-309
312
110
40
-282
285
: Currents
Figure 3. Total Current Consumption
OSC=0
7.0
TA = 85o C
CURRENT CONSUMPTION (mA)
6.0
5.0
TA = 25o C
4.0
TA = -40o C
3.0
2.0
1.0
0
2.25
2.5
2.75
VOLTAGE (V)
Figure 4. Powerdown Current
EN = LOW
6.0
CURRENT CONSUMPTION (PA)
5.0
4.0
TA = 85o C
3.0
2.0
1.0
TA = -40 o C
TA = 25o C
0
2.25
2.5
2.75
VCC (V)
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
11
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
Figure 5. RF Charge Pump Current
VCC = 2.5 Volts
2000
1500
CHARGE PUMP CURRENT (PA)
RF_CPG = 15
1000
RF_CPG = 3
500
0
-500
RF_CPG = 0
-1000
RF_CPG = 1
-1500
-2000
0
0.5
1.0
1.5
2.0
2.5
2.0
2.5
CHARGE PUMP VOLTAGE (V)
Figure 6. IF Charge Pump Current
VCC = 2.5 Volts
5.0
4.0
CHARGE PUMP CURRENT (mA)
3.0
IF_CPG=1
2.0
1.0
IF_CPG=0
0
-1.0
-2.0
-3.0
-4.0
-5.0
0
0.5
1.0
1.5
CHARGE PUMP VOLTAGE (V)
12
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
Figure 7. Charge Pump Leakage
RF PLL
10
8
LEAKAGE CURRENT (nA)
6
TA = 85o C
4
2
0
TA = -40o C
-2
TA = 25o C
-4
-6
-8
-10
0.5
0
1.5
1.0
2.0
2.5
CHARGE PUMP VOLTAGE (V)
Figure 8. Charge Pump Leakage
IF PLL
10
8
6
LEAKAGE CURRENT (nA)
TA = 85o C
4
2
0
-2
TA = -40o C
-4
TA = 25o C
-6
-8
-10
0
0.5
1.0
1.5
2.0
2.5
CHARGE PUMP VOLTAGE (V)
The input impedance of the FinRF, FinIF, and OSCin pins does not change significantly with voltage or temperature.
The impedance of the FinRF and FinIF pins also does not change much when the PLL is powered up or down.
Typical performance characteristics do not imply any sort of guarantee. Guaranteed specifications are in the electrical
characteristics section.
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
13
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
Bench Test Setups
CHARGE PUMP CURRENT MEASUREMENT PROCEDURE
DC
Blocking
Capacitor
10 MHz
SMA Cable
Frequency
Input Pin
SMA Cable
CPout
Pin
Signal Generator
Semiconductor
Parameter
Analyzer
Device
Under
Test
OSCin
Pin
Evaluation Board
Power Supply
The above block diagram shows the test procedure for testing the RF and IF charge pumps. These tests include
absolute current level, mismatch, and leakage. In order to measure the charge pump currents, a signal is applied
to the high frequency input pins. The reason for this is to guarantee that the phase detector gets enough
transitions in order to be able to change states. If no signal is applied, it is possible that the charge pump current
reading will be low due to the fact that the duty cycle is not 100%. The OSCin Pin is tied to the supply. The
charge pump currents can be measured by simply programming the phase detector to the necessary polarity.
For instance, in order to measure the RF charge pump current, a 10 MHz signal is applied to the FinRF pin. The
source current can be measured by setting the RF PLL phase detector to a positive polarity, and the sink current
can be measured by setting the phase detector to a negative polarity. The IF PLL currents can be measured in a
similar way. Note that the magnitude of the RF and IF PLL charge pump currents are also controlled by the
RF_CPG and IF_CPG bits. Once the charge pump currents are known, the mismatch can be calculated as well.
In order to measure leakage currents, the charge pump current is set to a TRI-STATE mode by enabling the
counter reset bits. This is RF_RST for the RF PLL and IF_RST for the IF PLL. The table below shows a
summary of the various charge pump tests.
Current Test
RF_CPG
RF_CPP
RF_CPT
IF_CPG
IF_CPP
IF_CPT
RF Source
0 to 15
0
0
X
X
X
RF Sink
0 to 15
1
0
X
X
X
RF TRI-STATE
X
X
1
X
X
X
IF Source
X
X
X
0 to 1
0
0
IF Sink
X
X
X
0 to 1
1
0
IF TRI-STATE
X
X
X
X
X
1
14
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
SENSITIVITY MEASUREMENT PROCEDURE
SMA Cable
Signal Generator
Frequency
Input Pin
Matching
Network
DC
Blocking
Capacitor
SMA Cable
Device
Under
Test
Ftest/LD
Pin
Frequency Counter
Evaluation Board
Power Supply
Frequency Input Pin
DC Blocking
Capacitor
Corresponding
Counter
Default Counter
Value
MUX Value
OSCin
1000 pF
FinRF
47 pF
FinIF
100 pF
OSC
RF_R / 2
50
14
0
RF_N / 2
500
15
X
IF_N / 2
500
13
X
Sensitivity is defined as the power level limits beyond which the output of the counter being tested is off by 1 Hz
or more of its expected value. It is typically measured over frequency, voltage, and temperature. In order to test
sensitivity, the MUX[3:0] word is programmed to the appropriate value. The counter value is then programmed to
a fixed value and a frequency counter is set to monitor the frequency of this pin. The expected frequency at the
Ftest/LD pin should be the signal generator frequency divided by twice the corresponding counter value. The
factor of two comes in because the LMX2471 has a flip-flop which divides this frequency by two to make the duty
cycle 50% in order to make it easier to read with the frequency counter. The frequency counter input impedance
should be set to high impedance. In order to perform the measurement, the temperature, frequency, and voltage
is set to a fixed value and the power level of the signal is varied. The power level at the part is assumed to be 4
dB less than the signal generator power level. This accounts for 1 dB for cable losses and 3 dB for the pad. The
power level range where the frequency is correct at the Ftest/LD pin to within 1 Hz accuracy is recorded for the
sensitivity limits. The temperature, frequency, and voltage can be varied in order to produce a family of sensitivity
curves. Since this is an open-loop test, the charge pump is set to TRI-STATE and the unused side of the PLL
(RF or IF) is powered down when not being tested. For this part, there are actually four frequency input pins,
although there is only one frequency test pin (Ftest/LD). The conditions specific to each pin are show above. The
LMX2471 has a test bit that may be useful in debugging sensitivity problems at the FinRF pin. The location of
this bit is R6[22] and should always be set to 0 for normal operation. If this bit is set to 1, then the sensitivity is
degraded. When one suspects a sensitivity problem, try setting this bit to 1 and see what happens. If the problem
is unaffected, it is likely not to be a sensitivity problem at the FinRF pin.
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
15
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
INPUT IMPEDANCE MEASUREMENT PROCEDURE
Frequency
Input Pin
Network Analyzer
Device
Under
Test
Evaluation Board
Power Supply
The above block diagram shows the test procedure measuring the input impedance for the LMX2471. This
applies to the FinRF, FinIF, and OSCin pins and is measured in a 50 ohm environment. The basic test procedure
is to calibrate the network analyzer, ensure that the part is powered up, and then measure the input impedance.
The network analyzer can be calibrated by using either calibration standards or by soldering resistors directly to
the evaluation board. An open can be implemented by putting no resistor, a short can be implemented by using a
0 ohm resistor, and a load can be implemented by using two 100 ohm resistors in parallel. Note that no DC
blocking capacitor is used for this test procedure. This is done with the PLL removed from the PCB. This requires
the use of a clamp down fixture that may not always be generally available. If no clamp down fixture is available,
then this procedure can be done by calibrating up to the point where the DC blocking capacitor usually is, and
then adding a 0 ohm resistor back for the actual measurement. Once that the network analyzer is calibrated, it is
necessary to ensure that the PLL is powered up. This can be done by toggling the power down bits (RF_PD and
IF_PD) and observing that the current consumption indeed increases when the bit is disabled. Sometimes it may
be necessary to apply a signal to the OSCin pin in order to program the part. If this is necessary, disconnect the
signal once it is established that the part is powered up. It is useful to know the input impedance of the PLL for
the purposes of debugging RF problems and designing matching networks. Another use of knowing this
parameter is make the trace width on the PCB such that the input impedance of this trace matches the real part
of the input impedance of the PLL frequency of operation. In general, it is good practice to keep trace lengths
short and make designs that are relatively resistant to variations in the input impedance of the PLL.
Functional Description
GENERAL
The basic phase-lock-loop (PLL) configuration consists of a high-stability crystal reference oscillator, a frequency
synthesizer such as the National Semiconductor LMX2471, a voltage controlled oscillator (VCO), and a passive
loop filter. The frequency synthesizer includes a phase detector, current mode charge pump, as well as
programmable reference [R] and feedback [N] frequency dividers. The VCO frequency is established by dividing
the crystal reference signal down via the R counter to obtain a frequency that sets the comparison frequency.
This comparison frequency, fCOMP, is input of a phase/frequency detector and compared with another signal, fN,
the feedback signal, which was obtained by dividing the VCO frequency down by way of the N counter and
fractional circuitry. The phase/frequency detector's current source outputs a charge into the loop filter, which is
then converted into the VCO's control voltage. The function of the phase/frequency comparator is to adjust the
voltage presented to the VCO until the frequency and phase of the feedback signal match that of the reference
signal. When this ‘phase-locked’ condition exists, the VCO frequency will be N+F times that of the comparison
frequency, where N is the integer component of the divide ratio and F is the fractional component. Fractional
synthesis allows the phase detector frequency to be increased while maintaining the same frequency step size
for channel selection. The division value N is thereby reduced giving a lower phase noise referred to the phase
detector input, and the comparison frequency is increased allowing faster switching times.
16
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
PHASE DETECTOR OPERATING FREQUENCY
The maximum phase detector operating frequency for the LMX2471 is 50 MHz. However, this is not possible in
all circumstances due to illegal divide ratios of the N counter. The crystal reference frequency also limits the
phase detector frequency. There are trade-offs in choosing what phase detector frequency to operate at. If this
frequency is run higher, then phase noise will be lower, but lock time may be increased due to cycle slipping.
After this phase detector frequency gets sufficiently high, then there are diminishing returns for phase noise, and
raising the charge pump current has a greater impact on phase noise. This phase detector frequency also has an
impact on fractional spurs. In general, the spur performance is better at higher phase detector frequencies,
although this is application specific. The current consumption may also slightly increase with higher phase
detector frequencies.
OSCILLATOR
The LMX2471 provides maximum flexibility for choosing an oscillator reference. One possible method is to use a
single-ended TCXO to drive the OSCin pin. The part can also be configured to be driven differentially using the
OSCin and OSCout* pins. Note that the OSCin and OSCout* pins can not be used as an inverter for a crystal.
Selection between these two modes does have a noticeable impact on phase noise and sub-fractional spurs.
Regardless of which mode is used, the performance is generally best for higher oscillator power levels.
POWER DOWN AND POWER UP MODES
The power down state of the LMX2471 is controlled by many factors. The one factor that overrides all other
factors is the EN pin. If this pin is low, this guarantees the part will be powered down. Asserting a high logic level
on EN is necessary to power up the chip, however, there are other bits in the programming registers that can
override this and put the PLL back in a power down state. Provided that the voltage on the EN pin is high,
programming the RF_PD and IF_PD bits to zero guarantees that the part will be powered up. Programming
either one of these bits to one will power down the appropriate section of the synthesizer, provided that the
ATPU[1:0] ( Auto Power Up ) bits do not override this.
There are many different ways to power down this chip and many different things that can be powered down.
This section discusses how to power down the PLLs on the chip. There are two terms that need to be defined
first: synchronous power down and asynchronous power down. In the case of synchronous power down, the PLL
chip powers down after the charge pump turns off. This is best to prevent unwanted frequency glitches upon
power up. However, in certain cases where the charge pump is stuck on, such as the case when there is no
VCO signal applied, this type of power down will not reliably work and asynchronous power down is necessary.
In the case of asynchronous power down, the PLL powers down regardless of the status of the charge pump.
There are 4 factors that affect the power down state of the chip: the EN pin, the power down bit, the TRI-STATE
bit, and writing to the RF N counter with the RF_ATPU[1:0] bits enabled
EN Pin
ATPU[1:0] Bits Enabled
+
RF N Counter Written To
RF_PD Bit
RF_CPT Bit
PLL State
Low
X
X
X
Asynchronous Power
Down
High
Yes
X
X
PLL is active with charge
pump in the active state.
High
No
0
0
PLL is active with charge
pump in the active state.
High
No
0
1
PLL is active, but charge
pump is TRI-STATE.
High
No
1
0
Synchronous Power Down
High
No
1
1
Asynchronous Power
Down
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
17
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
DIGITAL LOCK DETECT OPERATION
The RF PLL digital lock detect circuitry compares the difference between the phase of the inputs of the phase
detector to a RC generated delay of 10 nS. To enter the locked state (Lock = HIGH) the phase error must be
less than the 10nS RC delay for 5 consecutive reference cycles. Once in lock (Lock = HIGH), the RC delay is
changed to approximately 20nS. To exit the locked state (Lock = LOW), the phase error must become greater
than the 20nS RC delay. When the PLL is in the power down mode, Lock is forced LOW. For the RF PLL, the
digital lock detect circuitry does not function reliably for comparison frequencies above 20 MHz.
The IF PLL digital lock detect circuitry works in a very similar way as the RF PLL digitial lock circuitry, except that
it uses a delay of less than 15 nS for 5 reference cycles to determine a locked condition and a delay of greater
than 30 nS to determine the IF PLL is unlocked. Note that if the MUX[3:0] word is set such as to view lock detect
for both PLLs, an unlocked (LOW) condition is shown whenever either one of the PLLs is determined to be out of
lock. A flow chart of the IF digital lock detect circuitry is shown below.
PCB LAYOUT CONSIDERATIONS
Power Supply Pins For these pins, it is recommended that these be filtered by taking a series 18 ohm resistor
and then placing two capacitors shunt to ground, thus creating a low pass filter. Although it makes sense to use
large capacitor values in theory, the ESR ( Equivalent Series Resistance ) is greater for larger capacitors. For
optimal filtering minimize the sum of the ESR and theoretical impedance of the capacitor. It is therefore
recommended to provide two capacitors of very different sizes for the best filtering. 0.1 µF and 100 pF are typical
values. The charge pump supply pins in particular are vuvulnerablenerable to power supply noise.
18
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
High Frequency Input Pins, FinRF and FinIF The signal path from the VCO to the PLL is the most sensitive
and challenging for board layout. It is generally recommended that the VCO output go through a resistive pad
and then through a DC blocking capacitor before it gets to these high frequency input pins. If the trace length is
sufficiently short ( < 1/10th of a wavelength ), then the pad may not be necessary, but a series resistor of about
39 ohms is still recommended to isolate the PLL from the VCO. The DC blocking capacitor should be chosen at
least to be 100 pF. It may turn out that the frequency in this trace is above the self-resonant frequency of the
capacitor, but since the input impedance of the PLL tends to be capacitive, it actually be a benefit to exceed the
self-resonant frequency. The pad and the DC blocking capacitor should be placed as close to the PLL as
possible
Complimentary High Frequency Pin, FinRF* These inputs may be used to drive the PLL differentially, but it is
very common to drive the PLL in a single ended fashion. A shunt capacitor should be placed at the FinRF* pin.
The value of this capacitor should be chosen such that the impedance, including the ESR of the capacitor, is as
close to an AC short as possible at the operating frequency of the PLL. 100 pF is a typical value.
FASTLOCK AND CYCLE SLIP REDUCTION
The LMX2471 has enhanced features for Fastlock and cycle slip operation. The next several sections discuss
the the benefits of using both of these features. There are four possible combinations that are possible, and
these are shown in the table below:
Keep Comparison
Frequency the Same
Charge Pump Current
Increase Charge Pump Current
Keep Charge Pump Current the Same
Decrease Charge Pump Current
Decrease Comparison
Frequency (CSR)
(RF Side Only)
Classical Fastlock
Allows the loop bandwidth to be increased.
This has a frequency glitch caused by
switching the charge pump currents, but
there is no frequency glitch caused by
switching from fractional to integer mode
CSR/Fastlock Combination
Engaging the CSR does decrease the loop
bandwidth during frequency acquisition, but
may be necessary to reduce cycle slipping.
By also increasing the charge pump current,
this can compensate for the reduce loop
bandwidth due to the CSR
Operation with No Fastlock
This mode represents using no Fastlock
CSR Only
This mode is not generally recommended,
but may reduce cycle slipping in some
applications. Although the theoretical lock
time is decreased, due to the decreased loop
bandwidth during Fastlock, cycle slips can be
reduced or eliminated.
It never makes sense to use a lower charge
pump current during Fastlock than in the
steady state.
Note that if the charge pump current and cycle slip reduction circuitry are engaged in the same proportion, then it
is not necessary to switch in a Fastlock resistor and the loop filter will be optimized for both normal mode and
Fastlock mode. For third and fourth order filters which have problems with cycle slipping, this may prove to be
the optimal choice of settings.
Determining the Loop Gain Multiplier, K
The loop bandwidth multiplier, K, is needed in order to determine the theoretical impact of fastlock/CSR on the
loop bandwidth and also which resistor should be switched in parallel with the loop filter resistor R2. K = K_K ·
K_Fcomp where K is the loop gain multiplier K_K is the ratio of the Fastlock charge pump current to the steady
state charge pump current. Note that this should always be greater than or equal to one. K_Fcomp is the ratio of
the Fastlock comparison frequency to the steady state comparison frequency. If this ratio is less than one, this
implies that the CSR is being used.
Determining the Theoretical Lock Time Improvement and Fastlock Resistor, R2’
When using fastlock, it is necessary to switch in a resistor R2’, in parallel with R2 in order to keep the loop filter
optimized and maintain the same phase margin. After the PLL has achieved a frequency that is sufficiently close
to the desired frequency, the resistor R2’ is disengaged and the charge pump current is and comparison
frequency are returned to normal. Of special concern is the glitch that is caused when the resistor R2’ is
disengaged. This glitch can take up a significant portion of the lock time. The LMX2471 has enhanced switching
circuitry to minimize this glitch and therefore improve the lock time.
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
19
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
The change in loop bandwidth is dependent upon the loop gain multiplier, K, as determined in section 4. The
theoretical improvement in lock time is given below, but the actual improvement will be less than this due to the
glitch that is caused by disengaging Fastlock. The theoretical improvement is given to show an upper bound on
what improvement is possible with Fastlock. In the case that K < 1, this implies the CSR is being engaged and
that the theoretical lock time will be degraded. However, since this mode reduces or eliminates cycle slipping, the
actual lock time may be better in cases where the loop bandwidth is small relative to the comparison frequency.
Realize that the theoretical lock time multiplier does not account for the fastlock/CSR disengagement glitch,
which is most severe for larger values of K.
Loop Gain
Multiplier, K
Loop Bandwidth
Multiplier
R2’ Value
Lock Time
Multiplier
1:8*
0.35
open
× 2.828
1:4*
0.50
open
× 2.000
1:2*
0.71
open
× 1.414
4:1
2.00
R2/1.00
× 0.500
8:1
2.83
R2/1.83
× 0.354
16:1
4.00
R2/3.00
× 0.250
32:1
5.66
R2/4.65
× 0.177
* These modes of operation are generally not recommended
Using Fastlock and Cycle Slip Reduction (CSR) to Avoid Cycle Slipping
In the case that the comparison frequency is very large ( i.e. 100 X ) of the loop bandwidth, cycle slipping may
occur when an instantaneous phase error is presented to the phase detector. This can be reduced by increasing
the loop bandwidth during frequency acquisition, decreasing the comparison frequency during frequency
acquisition, or some combination of the these. If increasing the loop bandwidth during frequency acquisition is
not sufficient to reduce cycle slipping, the LMX2471 also has a routine to decrease the comparison frequency.
RF PLL Fastlock Reference Table and Example
The table below shows most of the trade-offs involved in choosing a steady-state charge pump current
(RF_CPG), the Fastlock charge pump current (RF_CPF), and the Cycle Slip Reduction Factor (CSR).
Parameter
Advantages to Choosing Smaller
RF_CPG
1. Allows capacitors in loop filter to be smaller values
making it easier to find physically smaller components and
components with better dielectric properties.
Advantages to Choosing Larger
Phase noise, especially within the loop bandwidth of the
system
will be slightly worse for lower charge pump currents.
2. Allows a larger loop bandwidth multiplier for fastlock, or
a higher cycle slip reduction factor.
RF_CPF
CSR
20
The only reason not to always choose this to 1600 µA is to This allows the maximum possible benefit for fastlock.
make it such that no resistor is required for fastlock. For
3rd and 4th order filters, it is not possible to keep the filter
perfectly optimized by simply switching in a resistor for
fastlock.
Do not choose this any larger than necessary to eliminate
cycle slipping. Keeping this small allows a larger loop
bandwidth multiplier for fastlock.
This will eliminate cycle slips better.
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
The above table shows various combinations for using RF_CPG, RF_CPF, and CSR. Although this table does
not show all possible combinations, it does show all the modes that give the best possible performance. To use
this table, choose a CSR factor on the horizontal axis, then a fastlock loop bandwidth multiplier on the vertical
axis, and the table will show all possible combinations of steady state current, Fastlock current, and what resistor
value (R2’) to use during Fastlock. In order to better illustrate the cycle slipping and Fastlock circuitry, consider
the following example:
Crystal Reference
10 MHz
Comparison Frequency
10 MHz × 2 = 20 MHz (OSC2X = 1)
Output Frequency
1930 – 1990 MHz
PLL Loop Bandwidth
10 KHz
Loop Filter Order
4th ( i.e. 7 components )
The comparison frequency is 20 MHz and the loop bandwidth is 10 KHz. 20 MHz is a good comparison
freqeuncy to use because it yields the best phase noise performance. This ratio of the comparison frequency to
the loop bandwidth is 2000, so cycle slipping will occur and degrade the lock time, unless something is done to
prevent it. Because the filter is fourth order, it would be difficult to keep the loop filter optimized if the loop gain
multiplier, K was not one. For this reason, choosing a loop gain multiplier of one makes sense. One solution is to
set the steady state current to be 100 µA, and the fastlock current to be 1600 µA. The CSR factor could be set to
1/16 and reduce this ratio to 2000/16 = 125. However, using 100 µA charge pump current has phase noise that
is significantly worse than the higher charge pump current modes. A better solution would be to use 200 µA
current and 1600 µA X2 ( using PDCP = X2 Fastlock ), since the 200 µA mode will have better phase noise.
Depending on how important phase noise is, it could make sense to use a higher steady state current. Using 800
µA steady state current provides much better phase noise than 200 uA ( about 5 dB ), but then the cycle slip
reduction factor would need to be reduced to 4. In general, it is good practice to use the PDCP = X2 fastlock
mode whenever cycle slip reduction is used, so that the best phase noise can be achieved. If the ¼ CSR factor
is used, then the ratio of comparison frequency to loop bandwidth in fastlock is reduced to 250. There may be
some cycle slipping, but the phase noise benefit of using the higher charge pump current may be worth it. If
phase noise is even more important, it might even make sense to have a steady state current of 1600 µA and
use a CSR factor of ½ and the PDCP mode of X2 Fastlock. Another consideration is that the comparison
frequency could be lowered in the steady state mode to reduce cycle slipping. This sacrifices phase noise for
lock time. In general, using Fastlock and CSR is not the same for every application. There is a trade-off of lock
time vs. phase noise. It might be tempting to try to achieve the best Fastlock benefit by using a K value of 32.
Even if the loop filter could be kept well optimized in Fastlock, this hypothetical design would probably switch
very fast when the Fastlock was engaged, but then when Fastlock is disengaged, a large frequency glitch would
appear, and the majority of the lock time would consist of waiting for this glitch to settle out. Although this would
definitely improve the lock time, even accounting for the glitch, the same result could probably be obtained by
using a lower K value, like 8, and having better phase noise instead.
Capacitor Dielectric Considerations for Lock Time
The LMX2471 has a high fractional modulus and high charge pump gain for the lowest possible phase noise.
One consideration is that the reduced N value and higher charge pump may cause the capacitors in the loop
filter to become larger in value. For larger capacitor values, it is common to have a trade-off between capacitor
dielectric quality and physical size. Using film capacitors or NP0/CG0 capacitors yields the best possible lock
times, where as using X7R or Z5R capacitors can increase lock time by 0 – 500%. However, it is a general
tendency that designs that use a higher compare frequency tend to be less sensitive to the effects of capacitor
dielectrics. Although the use of lesser quality dielectric capacitors may be unavoidable in many circumstances,
allowing a larger footprint for the loop filter capacitors, using a lower charge pump current, and reducing the
fractional modulus are all ways to reduce capacitor values. Capacitor dielectrics have very little impact on phase
noise and spurs.
FRACTIONAL SPUR AND PHASE NOISE CONTROLS FOR THE LMX2471
The LMX2471 has several bits that have a large impact on fractional spurs. These bits also have a lesser effect
on phase noise. The control words in question are CPUD[2:0], FM[1:0], and DITH[1:0]. It is difficult to predict
which settings will be optimal for a particular application without testing them, but the general recipe for using
these bits can be seen.
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
21
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
A good algorithm is to start with a 3rd order fractional modulator (FM=3) and dithering disabled. Then depending
on whether phase noise, fractional spurs, or sub-fractional spurs are most important, optimize the settings.
Integer spurs and fractional spurs are nothing new, but sub-fractional spurs are something unique to delta-sigma
PLLs. These are spurs that occur at a fraction of the frequency of where a fractional spur would appear.
First adjust the delta-sigma modulator order. Often increasing from a 2nd to a 3rd order modulator provides a
large benefit in spur levels. Increasing from a 3rd to a 4th order modulator usually provides some benefit, but it is
usually on the order of a few dB. The modulator order by far has the greatest impact on the main fractional
spurs. If the loop bandwidth is very wide, or the loop filter order is not high enough, higher order modulators will
introduce a lot of sub-fractional spurs. The second order modulator usually does not have these sub-fractional
spurs. The third order modulator will introduce them at ½ of the frequency where one would expect to see a
traditional fractional spur, thus the name "sub-fractional spur". The fourth order modulator will introduce these
spurs at ½ and ¼ of where a traditional fractional spur would be. If the benefit of using a higher order modulator
seems significant enough, it may make sense to try to compensate for them using the other two test bits, or
designing a higher order loop filter. Be aware that the impact of the modulator order on the spurs may not be
consistent across tuning voltage. When the charge pump mismatch is not so bad, the lower order modulators
may seem to outperform the higher order modulators, but when the worst case fractional spurs are considered
over the whole range, often the higher order modulator performs better.
Second, adjust with the CPUD[2:0] bits. Setting this bit to maximum tends to reduce the sub-fractional spurs the
most, however, it may degrade phase noise by up to 1 dB.
Third, experiment with the dithering. When dithering is enabled, it may increase phase noise by up to 2 dB.
However, enabling dithering may also reduce the sub-fractional spurs. Also, sometimes both the fractional spurs
and the sub-fractional spurs can be unpredictable with dithering disabled. This is because the delta-sigma
sequence is periodic, but the starting point changes. Dithering takes these problems away. When the fractional
numerator is 0, enabling dithering typically hurts spur performance, because it is trying to correct for spur that are
not there.
Fourth, consider experimenting with the loop filter order and comparison frequency. In general, higher order loop
filters are always better, but they require more components. Often, the best spur performance is at higher
comparison frequencies as well. The reason why this is the last step is not because it has the least impact, but
because it takes more labor to do this than to change the FM[1:0], CPUD[2:0], and DITH[1:0] bits.
Although general trends do exist, the optimal settings for test bits may depend on the comparison frequency and
loop filter. Also the output frequency in important. In particular, the charge pump tuning voltage is relevant. The
recommended way to do this is to test the spur levels at the low, middle, and high range of the VCO, and use the
worst case over these three frequencies as a metric for performance. Also, it is important to be aware that all the
rules stated above have counterexamples and exceptions. However, more often than not, these rules apply.
PROGRAMMING DESCRIPTION
GENERAL PROGRAMMING INFORMATION
The descriptions below describe the 24-bit data registers loaded through the MICROWIRE Interface. These data
registers are used to program the R counter, the N counter, and the internal mode control latches. The data
format of a typical 24-bit data register is shown below. The control bits CTL [3:0] decode the register address. On
the rising edge of LE, data stored in the shift register is loaded into one of the appropriate latches (selected by
address bits). Data is shifted in MSB first. Note that it is best to program the N counter last, since doing so
initializes the digital lock detector and Fastlock circuitry. Note that initialize means it resets the counters, but it
does NOT program values into these registers. Upon a cold power-up, it is necessary to program all the
registers. The exception is when 22-bit is not being used. In this case, it is not necessary to program the R7
register.
MSB
LSB
DATA [21:0]
23
22
CTL [3:0]
4 3
Submit Documentation Feedback
2
1
0
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
Register Location Truth Table
The control bits CTL [2:0] decode the internal register address. The table below shows how the control bits are
mapped to the target control register.
C3
C2
C1
C0
DATA Location
x
x
x
0
R0
0
0
0
1
R1
0
0
1
1
R2
0
1
0
1
R3
0
1
1
1
R4
1
0
0
1
R5
1
0
1
1
R6
1
1
0
1
R7
1
1
1
1
R8
Control Register Content Map
Because the LMX2471 registers are complicated, they are organized into two groups, basic and advanced. The
first four registers are basic registers that contain critical information necessary for the PLL to achieve lock. The
last 5 registers are for features that optimize spur, phase noise, and lock time performance. The next page
shows these registers.
Quick Start Register Map
Although it is highly recommended that the user eventually take advantage of all the modes of the LMX2471, the quick start register map is
shown in order for the user to get the part up and running quickly using only those bits critical for basic functionality. The following default
conditions for this programming state are a third order delta-sigma modulator in 22-bit mode with no dithering and no Fastlock.
REG
ISTE
R
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
DATA[19:0] ( Except for the RF_N Register, which is [22:0] )
R0
RF_N[10:0]
3
2
1
0
C3
C2
C1
C0
0
0
0
1
0
0
1
1
RF_FN[11:0]
RF_R[5:0]
0
R1
RF_
PD
1
RF_FD[11:0]
R2
IF_P
D
IF_P
R3
0
0
1
0
1
R4
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
R5
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
1
R6
0
0
0
0
0
0
0
0
0
0
1
1
0
0
0
1
0
0
0
0
1
0
1
1
R7
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
1
R8
0
0
0
0
1
1
0
0
0
0
0
0
1
0
0
0
1
1
1
0
1
1
1
1
8
7
6
5
4
3
2
1
0
C3
C2
C1
C0
0
0
0
1
0
0
1
1
0
1
0
1
0
1
1
1
1
0
0
1
1
0
1
1
1
1
0
1
IF_C
PG
IF_N[16:0]
RF_CPG[3:0]
IF_R[14:0]
Complete Register Map
The complete register map shows all the functionality of all registers, including the last five.
REG
ISTE
R
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
DATA[19:0] ( Except for the RF_N Register, which is [22:0] )
R0
RF_N[10:0]
R1
RF_
PD
1
R2
IF_P
D
IF_P
RF_FN[11:0]
RF_R[5:0]
RF_FD[11:0]
IF_C
PG
IF_N[17:0]
R3
0
R4
CSR[1:0]
R5
0
0
0
1
0
0
0
0
R6
0
0
0
0
0
RF_
CPT
RF_
CPP
IF_C
PT
R7
RF_CPG[3:0]
IF_R[14:0]
RF_CPF[3:0]
RF_TOC[13:0]
RF_FD2[9:0]
IF_TOC[11:0]
IF_C
PP
FDM
FM[1:0]
ATPU[1:0]
OSC
2X
OSC
RF_FN2[9:0]
MUX[3:0]
0
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
23
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
R8
0
0
0
0
www.ti.com
DITH[1:0]
0
0
0
0
0
0
PDCP[1:0]
0
0
CPUD[2:0]
0
1
1
1
1
R0 REGISTER
Note that this register has only one control bit. The reason for this is that it enables the N counter value to be
changed with a single write statement to the PLL.
REG
ISTE
R
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
DATA[19:0] ( Except for the RF_N Register, which is [22:0] )
R0
RF_N[10:0]
3
2
1
0
C3
C2
C1
C0
FN[11:0]
0
RF_FN[11:0] -- Fractional Numerator for RF PLL
Refer to section 2.8.1 for a more detailed description of this control word.
RF_N[10:0] -- RF N Counter Value
The RF N counter contains a 16/17/20/21 prescaler. Because there is only one selection of prescaler, the value
that is programmed is simply the N counter value converted into binary form. However, because this counter
does have a prescaler, there are limitations on the divider values.
RF_N[10:0]
RF_N
RF_C
≤64
RF_B
RF_A
N values less than or equal to 64 are prohibited.
65-66
Possible only with a second order delta-sigma engine
67-70
Possible with a second or third order delta-sigma engine.
71
0
0
0
0
1
0
0
0
1
1
1
72
0
0
0
0
1
0
0
1
0
0
0
...
.
.
.
.
.
.
.
.
.
.
.
2039
1
1
1
1
1
1
1
0
1
1
1
20402043
Possible with a second or third order delta-sigma engine.
20442045
Possible only with a second order delta-sigma engine.
>2045
N values above 2045 are prohibited.
R1 REGISTER
REG
ISTE
R
23
22
R1
RF_
PD
1
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
DATA[19:0] ( Except for the RF_N Register, which is [22:0] )
RF_R[5:0]
RF_FD[11:0]
3
2
1
0
C3
C2
C1
C0
0
0
0
1
RF_FD[11:0] -- RF PLL Fractional Denominator
The function of these bits are described in section 2.8.2.
RF_R [5:0] -- RF R Divider Value
The RF R Counter value is determined by this control word. Note that this counter does allow values down to
one.
R Value
24
RF_R[5:0]
1
0
0
0
0
0
...
.
.
.
.
.
.
63
1
1
1
1
1
1
Submit Documentation Feedback
1
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
RF_PD -- RF Power Down Control Bit
When this bit is set to 0, the RF PLL operates normally. When it is set to one, the RF PLL is powered down and
the RF Charge pump is set to a TRI-STATE mode. Because the EN pin and ATPU[1:0] word also controls power
down functions, there may be some conflicts. The order of precedence is as follows. First, if the EN pin is LOW,
then the PLL will be powered down. Provided this is not the case, the PLL will be powered up if the ATPU[1:0]
word says to do so, regardless of the state of the RF_PD bit. After the EN pin and the ATPU[1:0] word are
considered, then the RF_PD bit then takes control of the power down function for the RF PLL.
R2 REGISTER
REG
ISTE
R
23
22
21
R2
IF_P
D
IF_P
IF_C
PG
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
DATA[19:0] ( Except for the RF_N Register, which is [22:0] )
IF_N[17:0]
3
2
1
0
C3
C2
C1
C0
0
0
1
1
IF_N[16:0] -- IF N Divider Value
The IF N divider is a classical dual modulus prescaler with a selectable 8/9 or 16/17 modulus. The IF_N value is
determined by the IF_A , IF_B, and IF_P values. Note that the IF_P word can assume a value of 8 or 16. The
RF_A and RF_B counter values can be determined in accordance with the following equations.
B = N div P
A = N mod P
B≥A is required in order to have a legal N divider ratio
Here the div operator is defined as the division of two numbers with the remainder disregarded and the mod
operator is defined as the remainder as a result of this division. For the purposes of programming, it turns out
that the register value is just the binary representation of the N value, with the exception that the 4th LSB is not
used and must be programmed to 0 when the 8/9 prescaler is used.
IF_N Programming with the 8/9 Prescaler
IF_N[16:0]
N
Value
IF_B
400 × BW
R5 REGISTER
REG
ISTE
R
23
22
21
20
19
18
R5
0
0
0
0
0
0
17
16
15
14
13
12
11
10
9
8
7
6
5
4
DATA[19:0] ( Except for the RF_N Register, which is [22:0] )
0
0
IF_TOC[11:0]
3
2
1
0
C3
C2
C1
C0
1
0
0
1
IF_TOC[11:0] IF Timeout Counter for Fastlock
The IF_TOC word controls the operation of the IF Fastlock circuitry as well as the function of the FLoutIF output
pin. When IF_TOC is set to a value between 0 and 3, the IF Fastlock circuitry is disabled and the FLoutIF pin
operates as a general purpose CMOS TRI-STATE output. When IF_TOC is set to a value between 4 and 4095,
the IF Fastlock mode is enabled and FLoutIF is utilized as the IF Fastlock output pin. The value programmed into
IF_TOC represents the number of phase comparison cycles that the IF synthesizer will spend in the Fastlock
state.
IF_TOC[11:0]
Fastlock Mode
Fastlock Period [Charge Pump
Cycles]
FLoutIF Pin Functionality
0
Disabled
N/A
High Impedance
1
Manual
N/A
Logic “0” State
Forces IF charge pump current to
4 mA
2
Disabled
N/A
Logic “0” State
3
Disabled
N/A
Logic “1” State
4
Enabled
5
Fastlock
…
Enabled
…
Fastlock
4095
Enabled
4095
Fastlock
R6 REGISTER
REG
ISTE
R
23
22
21
20
19
18
R6
0
0
0
0
0
RF_
CPT
17
16
15
14
13
12
11
10
9
8
OSC
2X
OSC
7
6
5
DATA[19:0] ( Except for the RF_N Register, which is [22:0] )
RF_
CPP
IF_
CPT
IF_
CPP
FDM
FM[1:0]
ATPU
[1:0]
MUX
[3:0]
4
3
2
1
0
C3
C2
C1
C0
1
0
1
1
MUX[3:0] Frequency Out & Lock Detect MUX
These bits determine the output state of the Ftest/LD pin.
MUX[3:0]
0
28
0
0
0
Submit Documentation Feedback
Output Type
Output Description
High Impedance
Disabled
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
0
0
0
1
Push-Pull
General purpose
output, Logical “High”
State
0
0
1
0
Push-Pull
General purpose
output, Logical “Low”
State
0
0
1
1
Push-Pull
RF & IF Digital Lock
Detect
0
1
0
0
Push-Pull
RF Digital Lock
Detect
0
1
0
1
Push-Pull
IF Digital Lock Detect
0
1
1
0
Open Drain
RF & IF Analog Lock
Detect
0
1
1
1
Open Drain
RF Analog Lock
Detect
1
0
0
0
Open Drain
IF Analog Lock
Detect
1
0
0
1
Push-Pull
RF & IF Analog Lock
Detect
1
0
1
0
Push-Pull
RF Analog Lock
Detect
1
0
1
1
Push-Pull
IF Analog Lock
Detect
1
1
0
0
Push-Pull
IF R Divider divided
by 2
1
1
0
1
Push-Pull
IF N Divider divided
by 2
1
1
1
0
Push-Pull
RF R Divider divided
by 2
1
1
1
1
Push-Pull
RF N Divider divided
by 2
OSC -- Differential Oscillator Mode Enable
This bit selects between single-ended and differential mode for the OSCin and OSCout* pins. When this bit is set
to 0, the RF R and IF R counters are driven in a single-ended fashion through the OSCin pin. Note that the
OSCin and OSCout* pin can not be used to drive a crystal. When this bit is set to 1, the OSCin and OSCout*
pins are used to drive these R counters differentially. In some cases, spur performance may be better when this
is set to differential mode, even if the R counters are being driven in a single-ended fashion. Current
consumption in differential mode is slightly higher than when in single-ended mode.
OSC2X -- Oscillator Doubler Enable
When this bit is set to 0, the oscillator doubler is disabled TCXO frequency presented to the IF R counter is
unaffected. Phase noise added by the doulber is negligible.
ATPU -- PLL Automatic Power Up
This word enables the PLLs to be automatically powered up when their respective registers are written to. Note
that since the IF Powerdown bit is in the IF register, there is no need to have an ATPU function activated by the
R2 word.
ATPU
RF PLL
IF PLL
0
No auto power up
No auto power up
1
Powers up when R0 is written to
No auto power up
2
Powers up when R0 is written to
Powers up when R0 is written to
3
Reserved
Submit Documentation Feedback
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
29
OBSOLETE
LMX2471
SNAS213A – MAY 2004 – REVISED MAY 2004
www.ti.com
FM[1:0] -- Fractional Mode
Determines the order of the delta-sigma modulator. Higher order delta-sigma modulators reduce the spur levels
closer to the carrier by pushing this noise to higher frequency offsets from the carrier. In general, the order of the
loop filter should be at least one greater than the order of the delta-sigma modulator in order to allow for
sufficient roll-off.
FM
Function
0
Fractional PLL mode with a 4th order delta-sigma modulator
1
Disable the delta-sigma modulator. Recommended for test use only.
2
Fractional PLL mode with a 2nd order delta-sigma modulator
3
Fractional PLL mode with a 3rd order delta-sigma modulator
FDM -- Fractional Denominator Mode
When this bit is set to 0, the part operates with a 12- bit fractional denominator. For most applications, 12-bit
mode should be adequate, but for those applications requiring ultra fine tuning resolution, there is 22-bit mode.
Note that the PLL may consume slightly more current when it is in 22-bit mode.
FDM
Bits for Fractional
Denominator/Numerator
Maximum Size of Fractional
Denominator/Numerator
0
12-bit
4095
1
22-bit
4194303
IF_CPP -- IF PLL Charge Pump Polarity
When this bit is set to 1, the phase detector polarity for the IF PLL charge pump is positive. Otherwise set this bit
to 0 for a negative phase detector polarity
IF_CPT -- IF PLL Charge Pump TRI-STATE Mode
This bit enables the user to put the charge pump in a TRI-STATE ( high impedance ) condition. Note that if there
is a conflict, the ATPU bit overrides this bit.
RF_CPT
Charge Pump State
0
ACTIVE
1
TRI-STATE
RF_CPP -- RF PLL Charge Pump Polarity
For a positive phase detector polarity, which is normally the case, set this bit to 1. Otherwise set this bit to 0 for a
negative phase detector polarity.
RF_CPT -- RF PLL Charge Pump TRI-STATE Mode
This bit enables the user to put the charge pump in a TRI-STATE ( high impedance) condition. Note that if there
is a conflict, the ATPU bit overrides this bit.
RF_CPT
Charge Pump State
0
Active
1
TRI-STATE
R7 REGISTER
REG
ISTE
R
R7
30
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
DATA[19:0]
RF_FD2[9:0]
RF_FN2[9:0]
Submit Documentation Feedback
7
6
5
4
3
2
1
0
C3
C2
C1
C0
1
1
0
1
Copyright © 2004, Texas Instruments Incorporated
Product Folder Links: LMX2471
OBSOLETE
LMX2471
www.ti.com
SNAS213A – MAY 2004 – REVISED MAY 2004
Fractional Numerator Determination { RF_FN2[9:0], RF_FN[11:0], FDM }
In the case that the FDM bit is 0, then the part operates in 12-bit fractional mode, and the RF_FN2 bits become
don’t care bits. When the FDM is set to 1, the part operates in 22-bit mode and the fractional numerator is
expanded from 12 to 22-bits.
Fract
ional
RF_FN2[9:0]
Num
erato
r
( These bits only apply in 22- bit mode)
RF_FN[11:0]
0
In 12- bit mode, these are don’t care.
In 22- bit mode, for N