AN3180
Application note
A 200 W ripple-free input current
PFC pre-regulator with the L6563S
Introduction
A major limitation of transition-mode-operated PFC pre-regulators is their considerable input
ripple current, which requires a large differential mode (DM) line filter to meet EMI
requirements. The ripple-steering technique, with its ability to reduce an inductor ripple
current, theoretically to zero, can be very helpful in reducing the need for DM filtering in any
offline switching converter and in PFC pre-regulators in particular, where DM noise is an
issue as there is no electrolytic capacitor just after the bridge. This application note
illustrates the use of this technique, providing both the theoretical base and practical
considerations to enable successful implementation. Furthermore, it shows the bench
results of the EVL6563S-ZRC200 demonstration board, a 200 W ripple-free PFC preregulator based on an L6563S controller, designed according to the criterion proposed in
this application note.
Figure 1.
September 2010
EVL6563S-200ZRC 200W PFC demonstration board
Doc ID 17273 Rev 1
1/39
www.st.com
Contents
AN3180
Contents
1
Basic topologies with zero-ripple current . . . . . . . . . . . . . . . . . . . . . . . . 4
2
Zero-ripple current phenomenon: theory . . . . . . . . . . . . . . . . . . . . . . . . 6
3
Sensitivity of zero-ripple current condition . . . . . . . . . . . . . . . . . . . . . 10
4
Zero-ripple current phenomenon: practice . . . . . . . . . . . . . . . . . . . . . . 13
5
Capacitor selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6
A 200 W ripple-free input current PFC pre-regulator . . . . . . . . . . . . . . 19
7
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
8
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Appendix A Electrical equivalent circuit models of coupled inductors and
transformers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Appendix B Measuring transformer and coupled inductor parameters . . . . . . 36
Revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2/39
Doc ID 17273 Rev 1
AN3180
List of figures
List of figures
Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.
Figure 8.
Figure 9.
Figure 10.
Figure 11.
Figure 12.
Figure 13.
Figure 14.
Figure 15.
Figure 16.
Figure 17.
Figure 18.
Figure 19.
Figure 20.
Figure 21.
Figure 22.
Figure 23.
Figure 24.
Figure 25.
Figure 26.
Figure 27.
Figure 28.
Figure 29.
Figure 30.
EVL6563S-200ZRC 200W PFC demonstration board . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Some basic topologies with zero-ripple current characteristics . . . . . . . . . . . . . . . . . . . . . . 4
Smoothing transformer and related currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Coupled inductor a = k ne model under zero-ripple current conditions. . . . . . . . . . . . . . . . . 6
Coupled inductor a = ne/k model under zero-ripple current conditions. . . . . . . . . . . . . . . . . 7
Coupled inductor a = n model under zero-ripple current conditions . . . . . . . . . . . . . . . . . . . 8
Ripple-current attenuation as a function of the error sources for various winding coupling 12
Examples of high-leakage magnetic structures (cross-section) . . . . . . . . . . . . . . . . . . . . . 13
Two-section slotted bobbin suggested for the realization of a coupled inductor - top view 14
Two-section slotted bobbin suggested for the realization of a coupled inductor - side view14
Partial ripple cancellation: still under compensated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Partial ripple cancellation: overcompensated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
200 W PFC pre-regulator with ripple-free input current: electrical schematic. . . . . . . . . . . 20
Harmonic emissions and conformity to JEITA-MITI standards . . . . . . . . . . . . . . . . . . . . . . 23
Harmonic emissions and conformity to EN61000-3-2 standards . . . . . . . . . . . . . . . . . . . . 23
200W PFC pre-regulator with ripple-free input current: typical performance . . . . . . . . . . . 24
Line current and voltage @ full load (200 W) - Line current and voltage @ 115 Vac 200 W . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Line current and voltage @ full load (200 W) - Line current and voltage @ 230 Vac 200 W . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
AC and DC winding currents nominal input voltages and full load (200 W) - AC and DC
winding currents @ 115 Vac - 200 W . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
AC and DC winding currents nominal input voltages and full load (200 W) - AC and DC
winding currents @ 230 Vac - 200 W . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Conducted EMI @ Vin = 110 Vac, Pout = 100 W. Limits: EN50022 class B precompliance EMI test @ 115 Vac - 200 W . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Conducted EMI @ Vin = 110 Vac, Pout = 100 W. Limits: EN50022 class B precompliance EMI test @ 230 Vac - 200 W . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Coupled inductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Electrical equivalent circuit of coupled inductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Model of coupled inductors with a = n (a=n model) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Model of coupled inductors with a = ne (a = ne model) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Model of coupled inductors with a = 1 (a =1 model, or T-model) . . . . . . . . . . . . . . . . . . . . 33
Model of coupled inductors with a = k ne (a = k ne model) . . . . . . . . . . . . . . . . . . . . . . . . . 34
Model of coupled inductors with a = ne/k (a = ne/k model) . . . . . . . . . . . . . . . . . . . . . . . . . 34
Winding connections: aiding flux (left), opposing flux (right). . . . . . . . . . . . . . . . . . . . . . . . 37
Doc ID 17273 Rev 1
3/39
Basic topologies with zero-ripple current
1
AN3180
Basic topologies with zero-ripple current
Coupled magnetic devices have been around since the early days of electronics, and their
application to power switching circuits dates back to the late 70's with the experiments on
the Cuk converter, from which “magnetic integration” originated. With this technique,
inductors and transformers are combined into a single physical structure to reduce the
component count, usually with little or no penalty at all on the converter's characteristics,
sometimes even enhancing its operation. During initial experiments on the Cuk converter
the zero-ripple current phenomenon was first observed. The technique derived by the use of
this phenomenon is known as ripple-steering or ripple cancellation. Besides providing an
excellent discussion, also gives an interesting historical outline of the subject (see
References 1). The application of the zero-ripple current phenomenon is of considerable
interest in switching converters, where there are at least two reasons why it is desirable to
minimize inductor ripple currents. Firstly, lowering ripple current in inductors reduces the
stress on converter capacitors, resulting in either lower associated power loss or more
relaxed filtering requirements. Secondly, and often more importantly, most converter
topologies have pulsating current at either input or output or both, and most applications
require low conducted noise at both ports, because of EMC requirements or load
requirements.
Figure 2.
Some basic topologies with zero-ripple current characteristics
9LQ
9LQ
9LQ
9RXW
9RXW
9RXW
)O\EDFNFRQYHUWHUZ LWKQRQSXOVDWLQJLQSXWFXUUHQW
%RRVWFRQYHUWHUZ LWKQRQSXOVDWLQJLQSXWFXUUHQW
9LQ
)RUZ DUGFRQYHUWHUZ LWKQRQSXOVDWLQJLQSXWFXUUHQW
9LQ
9LQ
9RXW
9RXW
9RXW
)O\EDFNFRQYHUWHUZ LWKQRQSXOVDWLQJRXWSXWFXUUHQW
%RRVWFRQYHUWHUZ LWKQRQSXOVDWLQJRXWSXWFXUUHQW
)RUZ DUGFRQYHUWHUZ LWKQRQSXOVDWLQJRXWSXWFXUUHQW
!-V
This issue is commonly addressed with the use of additional LC filters, whose impact on
both the overall converter size and cost is not at all negligible, not to mention their interaction
with the small-signal dynamics which sometimes cause poor dynamic response issues or
even stability issues. In particular, in offline converters, where EMC regulations specify limits
to the amount of conducted and radiated emissions, a technique like ripple-steering which
makes the input current non-pulsating or nearly so, therefore eliminating most of the
differential mode conducted noise, is advantageous as it enables the reduction in EMI filter
size and complexity, especially in its differential filtering section (Cx capacitors and
differential mode inductors).
Reducing Cx capacitors to a minimum brings an additional benefit to applications with tight
specifications on standby consumption: Cx capacitors cause a considerable reactive current
to flow through the filter, which is a source of additional and unwanted loss (even 0.1 W or
more at high line); furthermore, the discharge resistor which, for safety, must be placed in
parallel to Cx can be higher. As a result, both losses are minimized.
4/39
Doc ID 17273 Rev 1
AN3180
Basic topologies with zero-ripple current
All converter topologies are capable of producing zero-ripple current phenomenon, provided
there are two or more inductors which have equal (or, more generally, proportional) voltages
of equal frequency and phase. Some topologies, such as Cuk and SEPIC, have two
inductors, which can be coupled on a common magnetic core: so they immediately lend
themselves to ripple-steering. The other basic topologies - buck, boost, buck-boost, flyback
and single-output forward - have typically a single inductor and then, to reduce its ripple
current to zero, they must be modified with the addition of a second winding wound on the
same inductor core. Moreover, this additional winding must be connected in such a way as
to have the same voltage as the winding where the ripple current is to be cancelled.
Figure 2 shows examples of how to modify some of the basic topologies to achieve zeroripple inductor current, to have non-pulsating current at either the input or the output. In all
examples it is possible to recognize the addition of a cell, a two-port circuit commonly
termed a smoothing transformer and shown separately in Figure 3. This cell is able to divert,
or steer, the AC component (the ripple current) from the externally accessible DC winding,
to the AC winding (cancellation winding) whose current conduction path goes directly to the
input port, leaving only DC current flowing through the DC winding and the output port. Note
that the denomination of input and output ports of the smoothing transformer cell is different
from that of input and output of the converter where the cell is applied.
Figure 3.
Smoothing transformer and related currents
, W
, W
W
W
GFZ LQGLQJ
/6
, W
, W
DFZ LQGLQJ
9
,DF W
&6
,DF W
9
W
!-V
This application note, after reviewing the theoretical base, considers the realization of zeroripple inductor current to minimize the input ripple in a TM PFC pre-regulator. In these
systems, the large input ripple is one of the major limitations to its use at higher power
levels.
Doc ID 17273 Rev 1
5/39
Zero-ripple current phenomenon: theory
2
AN3180
Zero-ripple current phenomenon: theory
Zero-ripple current in one of a two-winding coupled inductor, having self-inductances L1 and
L2, can be achieved if the coupling coefficient k, given by:
Equation 1
M
k=
L1 L 2
(M is their mutual inductance), and the effective turns ratio ne defined as:
Equation 2
ne =
L2
L1
is such that either k ne = 1 or k = ne, provided the windings are fed by the same voltage.
To confirm this, it is convenient to consider the a = k ne coupled-inductor model (refer to
Appendix A) with the terminals excited by proportional voltages v(t) and αv(t) having the
same frequency and phase, shown in Figure 4. This is the only condition to be imposed on
the terminal voltages, their actual waveform is irrelevant.
Figure 4.
Coupled inductor a = k ne model under zero-ripple current conditions
L W
Y W
/ N
N QH
DY W
/
LGHDO
L W
DY W
!-V
Figure 4 shows that, in order for the secondary ripple current (i.e. di2(t)/dt) to be zero, the
voltage across the inductance L2 (1-k2) must be zero, that is, the voltage on either side of it
must be the same. Thereby:
Equation 3
k ne v(t) = α v(t)
6/39
⇒
Doc ID 17273 Rev 1
k ne = α
AN3180
Zero-ripple current phenomenon: theory
Figure 5.
Coupled inductor a = ne/k model under zero-ripple current conditions
L W
Y W
/ N
Q
N
H
N /
L W
N
DY W
QH
DY W
LGHDO
!-V
Similarly, considering the model of Figure 5, again excited by proportional voltages v(t) and
αv(t), it is equally apparent that, in order for the primary ripple current to be zero the voltage
across the inductance L1 (1-k2) must be zero, that is, the voltage on either side of it must be
the same:
Equation 4
v(t) =
k
α v(t)
ne
⇒
k
α =1 ⇒
ne
k α = ne
If in Equation 3 and 4 α=1, which means that equal voltages are impressed on either side of
the coupled inductor, we find the above mentioned assertion. As α=1 is the most common
condition found in switching converters, from now on this is the only case that is taken into
consideration, therefore:
●
k ne = 1
●
k = ne
condition for zero-ripple secondary current
condition for zero-ripple primary current
Note that, as k < 1, to obtain a zero-ripple secondary current it must be ne > 1, that is L2 >
L1, while to obtain a zero-ripple primary current it must be ne < 1, that is L1 > L2; and so
ripple current cannot be reduced to zero in both windings simultaneously. In Figure 4 and 5,
note also that the inductance of the winding, where zero-ripple current is achieved, is
irrelevant, since there is no ripple current flowing (only DC current can flow). As a
consequence, the zero-ripple current winding reflects an open circuit to the other one, so
that the inductance seen at the terminals of that winding equals exactly its self-inductance.
The designation of which winding is the primary or the secondary is purely conventional.
Therefore, we consider only one zero-ripple current condition and arbitrarily assume the
condition to be assigned to the secondary winding:
Equation 5
k ne = 1
which, consistent with the terminology used for the smoothing transformer of Figure 3, is
termed DC winding, while the primary winding is termed AC or cancellation winding.
Equation 5, considering Equation 1 and 2 can be written in different equivalent ways:
Doc ID 17273 Rev 1
7/39
Zero-ripple current phenomenon: theory
AN3180
Equation 6
k ne = k
L2
M
=
=1
L1 L1
Equation 5 and 6 are noteworthy because of their concision in expressing the conditions for
zero-ripple current phenomenon to occur, but unfortunately its physical nature is not shown.
To provide some physical insight, let us consider the a = n coupled inductor model (n is the
physical turn ratio N2/N1) excited by equal terminal voltages v(t), shown in Figure 6.
Figure 6.
Coupled inductor a = n model under zero-ripple current conditions
L W
/O
/O
Q
L W
L0 W
Y W
/0
Y
W
Y W
LGHDO
Y W
!-V
Proceeding with the same technique, in order for the ripple current i2(t) to be zero, the
voltage across the secondary leakage Ll2 must be zero, that is, the voltages on either side of
Ll2 must be equal to one another. On the other hand, if i2(t)=0 the voltage impressed on the
primary side of the ideal transformer v'(t) is given by the ratio of the inductive divider made
up of the primary leakage inductance Ll1 and the magnetizing inductance LM; the voltage
applied to the left-hand side of Ll2 is equal to nv'(t). Then, there is zero-ripple current on the
secondary side of the coupled inductor if the following condition is fulfilled:
Equation 7
LM
n v(t) = v(t)
L l1 + L M
⇒
LM
L
n = M n =1
L l1 + L M
L1
which is equivalent to Equation 5 and 6, as can be easily shown, considering that LM = M/n.
Equation 7 provides the desired physical interpretation of the zero-ripple current condition: it
occurs when the turn ratio exactly compensates for the primary winding leakage flux, so that
the primary winding induces, by transformer effect, a voltage identical to its own excitation
voltage on the secondary winding; and so, if this is externally excited by the same voltage,
no ripple current flows through it.
The extensions of this interpretation to the case of zero-ripple primary current (just reflect
the magnetizing inductance LM to the secondary side) and to that of proportional excitation
voltages (α ≠ 1) are obvious.
8/39
Doc ID 17273 Rev 1
AN3180
Zero-ripple current phenomenon: theory
To summarize the main results of this brief analysis:
1. Ripple-current can be reduced to zero in either winding of a two-winding coupled
inductor, but not simultaneously
2. The only conditions imposed on the voltages that excite the windings, in order to get
ripple current steering, is that they are proportional, with the same frequency and
phase
3. Provided the above condition on the terminal voltages is met, the zero-ripple current
occurrence in one winding depends on a proper choice of the leakage inductance
associated with the other winding only
4.
As a winding where zero-ripple current occurs is virtually open for AC currents, the
inductance seen at the terminals of the other winding is exactly equal to its selfinductance.
Doc ID 17273 Rev 1
9/39
Sensitivity of zero-ripple current condition
3
AN3180
Sensitivity of zero-ripple current condition
In real-world coupled inductors it is unthinkable to reduce the ripple current in a winding to
exactly zero and produce a perfect ripple steering. There are two basic reasons for this:
●
Zero-ripple condition mismatch. In practice, the inductance of a winding is determined
by the number of turns and the average permeability of the associated magnetic circuit.
The turn ratio can assume only discrete values (ratio of two integer numbers) and it is
difficult to control the average permeability to achieve the exact value that allows the
meeting of the condition in Equation 7 or any equivalent (Equation 6). Even if this may
be obtained in occasional samples, manufacturing tolerances cause the actual value to
deviate from the target in mass production
●
Impressed voltage mismatch. In real operation, there are several factors that cause the
two windings to be excited by voltages that are not exactly equal to one another, such
as the voltage drop across the winding resistance (neglected so far), or the mere
inability of the external circuit to do so. For example, in the smoothing transformer, the
finite capacitance value of CS and its ESR cause an impressed voltage mismatch even
in the case of ideal windings. To evaluate the residual ripple current it is convenient to
use the a = k ne model in Appendix A, already used (Figure 4) for deriving the zeroripple current condition. Based on that, it is possible to write:
Equation 8
d i 2 ( t ) v 2 ( t ) − k ne v 1 ( t )
=
dt
L2 1− k 2
(
)
which, after a simple algebraic manipulation, can be re-written as:
Equation 9
⎤
⎡
⎥
⎢
di 2 ( t )
1
⎢ v 2 (t) − v1 (t) + ( 1 − k ne )v1 (t) ⎥
=
2
1
42
43
1
4
4
2
4
4
3
dt
L 2 1 − k ⎢Im pressed voltage
⎥
Zero ripple condition
⎥⎦
⎢⎣ mismatch
mismatch
(
)
Although the inductance of the DC winding is theoretically irrelevant to the phenomenon
itself (the inductance could even be zero), Equation 7 and 8 show that this inductance is
significant in practice, because it determines the actual residual ripple current resulting from
the unavoidable aforementioned mismatches. More precisely, these equations highlight the
need for a high-leakage magnetic structure, so that a low coupling coefficient k maximizes
the “residual” inductance L2 (1-k2).
Note that a low value for k also means that the value of ne that meets the zero-ripple
condition is higher: for a given primary inductance L1, this implies a higher value of L2, and
so further contributing to keeping the ripple low.
With the aim of assessing the amount of ripple attenuation in case of a non-ideal
cancellation, considering an assigned value for L1 is an important practical constraint. As
previously stated, the inductor ripple current on the cancellation winding is just determined
by its inductance L1 and this ripple is seen by the power switch of the converter. This means
that, if L1 is unchanged, the converter circuit still operates exactly under the same conditions
even with the use of the additional coupled inductor.
10/39
Doc ID 17273 Rev 1
AN3180
Sensitivity of zero-ripple current condition
If the attenuation A is defined as the ratio of the residual ripple di2(t)/dt, given by Equation 7
or 8, to the ripple that would be there without the coupled inductor (di1(t)/dt =v1(t)/L1, equal
to the actual ripple on the cancellation winding, as L1 is unchanged), it is possible to write for
the worst case scenario:
Equation 10
di 2 ( t )
⎛ Δv(t)
⎞
L di 2 ( t )
=ρ⎜
+ δ⎟
A = dt = 1
⎜ v ( t)
⎟
di1 (t) v1 (t) dt
⎝ 1
⎠
dt
where Δv(t) = v2(t)-v1(t) is the absolute voltage mismatch, δ = k ne - 1 is the zero-ripple
condition mismatch (absolute and relative values coincide) and the factor ρ is given by:
Equation 11
ρ=
k2
1
(1 + δ)
2
1− k 2
In Figure 7 the attenuation A is plotted for different values of the relative voltage mismatch
Δv(t)/v1(t) and of the coupling coefficient k, as a function of the zero-ripple condition. From
the inspection of these plots, it is apparent that a low coupling coefficient is essential for a
good attenuation even if the zero-ripple condition is not exactly met. To achieve attenuations
always greater than 10-12 dB even with a tolerance of ±10 % on the value of δ and 10 %
voltage mismatch, the coupling coefficient k must be around 0.7. Lower k values would lead
to a higher insensitivity of the zero-ripple condition due to mismatches but lead to more
turns for the DC winding, which could become an issue in terms of inductor construction.
Note that in case of under-compensation (δG%@
N
$>G%@
N
N
N
N
Y
Y
N
Y
Y
G>@
N
N
N
N
N
$>G%@
N
N
Y
Y
N
$>G%@
G>@
Y
Y
G >@
G >@
!-V
5.
A resonance frequency, close to:
Equation 12
fr =
1
2π L1CS
for small zero-ripple condition mismatches, exists where the gain peaks and attenuation
may be lost. However, the parasitic resistances damp this peak so that it is not high, even
though its effect might be visible in the waveforms, should that frequency be stimulated;
anyhow, to get the desired attenuation at the switching frequency and above, the practical
values of CS are such that the peak occurs at a much lower frequency, where EMC
regulations do not generally apply.
12/39
Doc ID 17273 Rev 1
AN3180
4
Zero-ripple current phenomenon: practice
Zero-ripple current phenomenon: practice
Before giving details of the practical realization of a coupled inductor able to provide ripple
steering, it is useful to draw some conclusions of considerable practical interest from the
theoretical analysis carried out in the previous sections:
1.
The fundamental point is that to achieve zero-ripple current in a coupled inductor with
low sensitivity to parameter spread, a high-leakage magnetic structure is needed,
which is contrary to the traditional transformer design practice. Then, in the important
case of inductors realized with a gapped ferrite core plus bobbin assembly, the usual
concentric winding arrangement is not recommended, although higher leakage
inductance values can be achieved by increasing the space between the windings.
However, it is difficult to obtain a stable value, because it depends on parameters (such
as winding surface irregularities or spacer thickness) which are difficult to control. Other
methods are used, such as placing the windings on separate core legs (if using an EE
core) or side-by-side on the same leg. The latter arrangement is possible with both EE
and pot cores (see Figure 8). They permit much more stable leakage inductance
values, because they are related to the geometry and the mechanical tolerances of the
bobbin, which are much better controlled. At this point there is the practical issue of
finding suitable bobbins for such arrangements: large production quantities could make
the case for a custom product, but using something readily available in the market is a
good choice anyway. Slotted bobbins, like those illustrated in Figure 9, are quite
commonly available from several manufacturers; they easily lend themselves to a sideby-side winding arrangement, and they are considered here.
Figure 8.
2.
Examples of high-leakage magnetic structures (cross-section)
The fundamental action of the smoothing transformer is to split up the current into its
DC component, which flows in the DC winding, and its AC component, which flows in
the AC winding. In this way the total rms current in the windings is unchanged, hence
the total copper area needed to handle the two currents separately is quite close to that
of a single inductor carrying the total current. As compared to a single inductor, no core
size increase is typically expected because of insufficient winding window area, except
for a few marginal cases where the slight decrease of window area due to slotting
becomes critical or where the DC winding has many more turns. It is also worth noting
that the DC winding can be made with a single wire, as its residual AC current is low;
only the AC winding is made with litz or multi-stranded wire. This minimizes the
additional cost of the added winding.
Doc ID 17273 Rev 1
13/39
Zero-ripple current phenomenon: practice
Figure 9.
AN3180
Two-section slotted bobbin
suggested for the realization of a
coupled inductor - top view
Figure 10. Two-section slotted bobbin
suggested for the realization of a
coupled inductor - side view
6HFRQGDU\ SULPDU\
ZLQGLQJVORW
3ULPDU\ VHFRQGDU\
ZLQGLQJVORW
6HFRQGDU\ SULPDU\
ZLQGLQJVORW
3ULPDU\ VHFRQGDU\
ZLQGLQJVORW
!-V
3.
!-V
As a result of the aforementioned current splitting action, the same happens to the
magnetic flux: the AC winding is responsible for generating the AC component of the
flux and the DC winding for its DC component, but their sum is the same as in a single
inductor carrying the total current. This means that no core size increase is expected
because of flux density limitations (core saturation) or core loss reasons. Also the core
size selection approach is the same.
To complete the picture, an important property of the “gapped ferrite core plus bobbin”
assembly should be recalled. The reluctance of the leakage flux path is constant for a given
core geometry and is independent of the gap thickness, lgap: essentially, it is a function of
the physical dimensions of the core and the distance between the windings. This implies
that the leakage inductance of the primary winding, Ll1, depends only on the turn number
N1 and does not change if lgap is adjusted. However, the total primary inductance L1= Ll1+
LM is strongly affected by the gap thickness, but it is only the magnetizing inductance LM that
changes. With reference to the zero-ripple current condition in Equation 7:
Equation 13
L − L l1
LM
n= 1
n =1
L l1 + L M
L1
once the ferrite core and the associated bobbin are defined, it is thereby possible to control
LLl1 and LM separately, acting on N1 and lgap respectively, then the secondary turn number
N2 is determined so that the ratio n = N2/N1 meets the condition in Equation 7.
14/39
Doc ID 17273 Rev 1
AN3180
Zero-ripple current phenomenon: practice
At this point, all the elements needed to outline a step-by-step practical design procedure
are in place. For the details of steps 2 to 4, please refer to the algorithm described in
References 3.
1.
The case of a conventional inductor is considered. From the circuit design determine
the required inductance value L1, the maximum peak short-circuit current and,
considering full load conditions, the rms, DC and AC inductor currents, recalling that
between them the following relationship holds:
Equation 14
IAC =
2
2
IRMS
− IDC
In the case of a flyback transformer the winding to consider is the one on the side where the
ripple current is to be cancelled. For the forward converter the inductor to consider is the
output filter inductor.
2.
Assuming that an EE-shaped ferrite core with a slotted bobbin is used, determine the
maximum flux density and the maximum flux swing the core is operated at. Tentatively
select a core size and determine the loss limit, both in the winding (PCu) and the core
(PFe). Calculate the turn number N1 of the AC winding in such a way that the desired
inductance value L1 can be obtained without exceeding either the core saturation limits
or the permitted core losses. Determine the required gap length
3.
Calculate the conductor size for the AC winding considering that its resistance must be:
Equation 15
R ac ≤
PCu
I2AC
To minimize skin and proximity effects, use litz or multi-stranded wire.
4.
Calculate the conductor size for the DC winding considering that its resistance must be:
Equation 16
R dc ≤
PCu
2
IDC
As DC current flows through this winding a single wire is used.
5.
Wind the entire AC winding in one slot and, in the other slot, wind a couple of layers of
the wire which is used for the DC winding. Temporarily assemble the core set. If the two
half-cores are not gapped, use a gap value close to that calculated, to make the
measurements that follow under conditions as close to those of a finished sample as
possible. Consider also that a small gap amplifies measurement errors
6.
Using any of the methods described in Appendix B, measure the leakage inductance
LLk (referred to the AC winding) obtained in this way and calculate a first-cut value for
N2:
Doc ID 17273 Rev 1
15/39
Zero-ripple current phenomenon: practice
AN3180
Equation 17
N2 = N1
L
L − L LK
where, in this case, L is the measured inductance of the AC winding. Add 5 % to the
result, to account for the slight coupling improvement that there is as the DC winding is
entirely in place, and round up to the next integer. Let this number be N2
7.
After removing the layer used for the preliminary measurement and the core, wind the
N2 turns of the DC winding. Do not permanently fix the end of the wire because N2
should be greater than the value that meets the zero-ripple condition and some turns
need to be taken out in the following step
8.
Reassemble the core and adjust the air gap so as to get the required value L1.
Measure the leakage and the magnetizing inductances (see Appendix B) and check if
(7) is met. If not, remove one turn and repeat the step until the condition in Equation 7
is close to being met
9.
Connect the coupled inductor to the converter, power it on and measure the ripple on
the DC winding. It should be quite small. If it looks like that in Figure 11 the turn number
of the DC winding has been reduced too much. If it is not possible to add more turns to
the DC winding, either the air gap must be reduced, if possible (paying attention to
saturation!), or the turns number of the AC winding must be reduced. If the ripple looks
like that in Figure 12, where the ripple is 180° out-of-phase, there are still too many
turns on the DC winding and they should be further reduced. Also in this case it is
possible to fine tune by adjusting the gap, if possible. Only now can the DC winding
wire end be permanently fixed to the bobbin pin. Record the final values of N1, N2 and
lgap.
Figure 11. Partial ripple cancellation: still
under compensated
Figure 12. Partial ripple cancellation:
overcompensated
GFZ LQGLQJFXUUHQW
GFZ LQGLQJFXUUHQW
DFZ LQGLQJFXUUHQW
DFZ LQGLQJFXUUHQW
JDWHGULYH
JDWHGULYH
8QGHUFRPSHQVDWHG NQH
GFZLQGLQJZLWKWRRIHZWXUQV
!-V
2YHUFRPSHQVDWHG NQH !
GFZLQGLQJZLWKWRRPDQ\WXUQV
!-V
Note that the first four steps of this procedure are the same as those for a conventional
inductor. Starting from step 5, the procedure becomes empirical and quite tricky and might
also be time-consuming. Fortunately, this work needs to be carried out only for the first
prototype.
There is still an important practical issue to consider. When the two half cores are
assembled, even if they are kept together well, they are typically quite loose inside the
16/39
Doc ID 17273 Rev 1
AN3180
Zero-ripple current phenomenon: practice
bobbin, unless mounting clips or gluing are used (quite impractical to use during the cutand-try phase). The position of the core inside the bobbin, especially along the direction of
the legs, is critical because it changes the position of the air gap with respect to the
windings. Moving the core causes significant variations of the magnetizing inductance,
therefore it is recommended to fix the cores in a stable position as close as possible to that
of the finished sample before doing any inductance measurement or test on the converter
circuit.
It is now important to determine how much spread can be expected in mass production.
Side-by-side winding arrangements in slotted bobbins are quite commonly used when
making common-mode chokes. For wires which are not too thin (say, above 0.2 mm) with a
good winding machine it is not difficult to have a tolerance of around 4-5 % on the leakage
inductance. As for the self-inductance of one winding, using gapped cores the tolerance on
the AL factor (nH/turn2), deducible from ferrite core catalogs, is 5 % or better for air gaps
>0.1mm for small cores (e.g. an E30/15/7) and for air gaps >0.4 mm in case of bigger cores
(e.g. an E55/28/21). An additional 3 % tolerance has to be typically considered due to the
above mentioned displacement of the core inside the bobbin. It is therefore possible to
assume 5 % tolerance for Ll1 and 8 % tolerance on L1.
Turn ratio n is not subject to production spread, the rounding error just introduces a fixed
mismatch in the zero-ripple current condition. Again, using gapped cores, where the air gap
cannot be adjusted to compensate for the mismatch, the rounding error is 0.5 turns,
therefore the maximum absolute error that it introduces on n is ≤ 0.5/N1 and the maximum
relative error is (0.5/N1)/n = 0.5/N2.
With some simple algebraic manipulations it is possible to find the relative error on the zeroripple current condition in Equation 7 due to the spread δl1 and δ1 of Ll1 and L1,
respectively:
Equation 18
δ = ( n − 1)
δ1 − δ l1
1 + δ1
As n is slightly greater than 1 the effect of the spread of Ll1 and L1 is attenuated. For
example, with n=1.3, the resulting maximum spread is δ = -4.2 % to +3.6 %.
With good approximation, it is possible to consider that this tolerance band is centered on a
value shifted by the rounding error. Referring to Figure 7, the attenuation factor A degrades
more quickly for negative values, therefore it is better to round N2 to the upper integer, to
provide a “positive offset” to δ and be in a region where A changes less. Furthermore, if
N2=50 turns (rounded up), the rounding error is +1 %, so that the total tolerance band is -3.2
%, +4.6 %.
Doc ID 17273 Rev 1
17/39
Capacitor selection
5
AN3180
Capacitor selection
It is obvious that the capacitance of the smoothing transformer capacitor should be as large
as possible and its ESR as low as possible to minimize the impressed voltage mismatch.
Although this assertion is always true, the use in a PFC pre-regulator to minimize the input
current ripple poses an important limitation to the capacitance value that can be used.
In Figure 2 it is shown that in the zero-ripple input current configurations there is a DC path
between the capacitor and the input port of the converter. This means that, at low frequency,
the capacitor appears as if directly connected to the input rectifier bridge; therefore, in a
PFC stage it adds to the capacitor normally placed just after the bridge and contributes to
the increase in total harmonic distortion (THD) of the low-frequency line current as well as to
lowering the power factor (PF).
Actually, the capacitor placed after the bridge can be reduced because of the ripple
reduction caused by the smoothing transformer; but the value of the smoothing capacitor
must, in any case, be selected trading off the filtering performance of the smoothing
transformer against the need for a low THD of the line current. The peak-to-peak voltage
ripple Δv appearing across the capacitor, which is essentially twice the voltage mismatch, is
related to the AC ripple amplitude ΔI:
Equation 19
Δv pk −pk =
ΔI
8 fsw C s
which is a maximum at the top of the sinusoid at minimum line voltage, where the amplitude
of the AC current is maximum and the switching frequency is close to its minimum value.
Substituting the values that can be derived by the PFC relationships [4], it is possible to find
the following expression for the relative voltage mismatch:
Equation 20
L
Δv(t)
= 1
V1(t) max 4 C s
⎛ Pinmax
⎜
⎜ Vin 2
min
⎝
2
⎞
Vout
⎟
⎟ Vout − 2 Vin
⎠
min
Typically, the values normally used for the capacitor after the bridge (5÷15 nF/W) lead to a
good compromise. Polypropylene capacitors are recommended for their low ESR.
18/39
Doc ID 17273 Rev 1
AN3180
6
A 200 W ripple-free input current PFC pre-regulator
A 200 W ripple-free input current PFC pre-regulator
Figure 13 shows the electrical schematic of the EVL6563S-ZRC200 demonstration board, a
200 W zero-ripple-current PFC pre-regulator (see BOM in Table 1) based on the boost
topology modified by adding the cancellation winding to the boost inductor and a capacitor,
to make a smoothing transformer to minimize the input ripple. Table 2 lists its electrical
specifications, while Table 3 gives the details of the boost inductor.
The design of the pre-regulator is carried out exactly in the same way as for a traditional
stage. The coupled inductor, the crucial element of this application, is designed following the
previously outlined procedure.
The controller chip used is the L6563S [5], suitable for implementing transition-modecontrolled PFC pre-regulators up to 250-300 W. Other notable semiconductor devices used
are the STF12NM50N (a second generation MDmeshTM 0.38 /500 V N-channel MOSFET)
and the STTH5L06 (a Turbo2 ultrafast diode, optimized for use in transition-mode PFC
stages).
A prototype of the pre-regulator has been built and evaluated on the bench. The diagrams of
Figure 14, 15 and 16 show its typical performance; Figure 17 to 22 show some significant
waveforms.
Doc ID 17273 Rev 1
19/39
&1
0.'6
)
Doc ID 17273 Rev 1
Q
&
+7$9
N
5
0
5
5
0
5
N
X)
5
&
S)
Q
Q
N
&
&
&
Q
Q
0
&9UPV@
:
:
:
3RZHU)DFWRU3)
9LQ>9UPV@
:
:
:
7RWDOKDUPRQLFGLVWRUWLRQ7+'>@
9LQ>9UPV@
24/39
Doc ID 17273 Rev 1
!-V
AN3180
A 200 W ripple-free input current PFC pre-regulator
Figure 17. Line current and voltage @ full load Figure 18. Line current and voltage @ full load
(200 W) - Line current and voltage
(200 W) - Line current and voltage
@ 115 Vac -200 W
@ 230 Vac -200 W
Figure 20.
Figure 19. AC and DC winding currents
nominal input voltages and full load
(200 W) - AC and DC winding
currents @ 115 Vac - 200 W
AC and DC winding currents
nominal input voltages and full load
(200 W) - AC and DC winding
currents @ 230 Vac - 200 W
The waveforms of Figure 17 to 20 are taken at full load. The images in Figure 17 and 18
show the mains voltage and current drawn from the mains: this is quite sinusoidal, as in a
standard transition mode PFC converter. In fact its harmonic content is well below the
maximum allowed by current regulations (JEITA-MITI and European EN61000-3-2), as can
be seen from Figure 14 and 15. Also the power factor and the total harmonic distortion
parameters are as good as for a standard circuit without the cancellation winding, as can be
seen from Figure 16, where also the efficiency measurements are plotted, showing values
always greater than 91.7 %.
The zoomed image of Figure 19 (@115 Vac) shows a residual ripple current in the DC
winding of 180 mA pk-pk, against a 6.053 A pk-pk of the AC winding current (which would
flow in the boost inductor in the conventional topology arrangement), and so resulting in
0.18/6.053 = 29.7*10-3 (-30.5 dB) attenuation.
Doc ID 17273 Rev 1
25/39
A 200 W ripple-free input current PFC pre-regulator
AN3180
The zoomed image of Figure 20 shows a residual ripple current in the DC winding of 114
mA pk-pk, against a 2.962 A pk-pk of the AC winding current, and so resulting in
0.114/2.962 = 38.5*10-3 (-28.3 dB) attenuation.
The EMI filter used for reducing the line conducted interferences is a single cell filter made
up of two X capacitors and a common mode inductor (see the electrical schematic in
Figure 13: CX1, CX2, L1) and two further Y capacitors (CY1, CY2). In standard transition
mode PFC converters (without the cancellation winding) a double cell filter is usually
necessary to reduce the EMI emission to levels within the maximum allowed by current
regulations. The effect of ripple steering is particularly conspicuous up to about 2 MHz, with
more than 20 dB attenuation provided. For higher frequencies, where common mode noise
is dominant, its effect becomes negligible and emissions must be controlled resorting to the
usual noise control techniques. The EMI measurement results are shown in Figure 21 and
22: the emission level with peak detection is well below the quasi-peak and average limits
envisaged by EN50022 class B.
Figure 22.
Figure 21. Conducted EMI @ Vin = 110 Vac,
Pout = 100 W. Limits: EN50022
class B - precompliance EMI test @
115 Vac - 200 W
26/39
Doc ID 17273 Rev 1
Conducted EMI @ Vin = 110 Vac,
Pout = 100 W. Limits: EN50022
class B - precompliance EMI test @
230 Vac - 200 W
AN3180
7
Conclusions
Conclusions
The ripple-steering technique, with its ability to reduce an inductor ripple current
theoretically to zero, has been discussed and its theoretical base has been outlined. The
construction of a coupled inductor is addressed and a practical design guide given to enable
its successful implementation.
A practical application of the ripple-steering technique has been shown: it refers to a
Transition-mode controlled PFC pre-regulator where the technique has been used to
minimize the input ripple current, which is a major limitation to the use of this simple and
efficient topology at higher power levels.
The experimental results have shown that the ripple-steering technique is very effective in
reducing the input ripple current and that considerable savings can be achieved in the
differential-mode input EMI filter, with the positive side effect of minimizing the loss due to
the filter itself.
Doc ID 17273 Rev 1
27/39
References
8
28/39
AN3180
References
1.
“A 'Zero' Ripple Technique Applicable to Any DC Converter”, Power Electronics
Specialists Conference, 1999. PESC 99. 30th Annual IEEE (1999 Volume 2)
2.
“The Coupled Inductor Filter: Analysis and Design for AC Systems”, IEEE Transactions
on Power Electronics, Vol. 45, No. 4, August 1998, pp. 574-578
3.
“Inductor and Flyback Transformer Design”, Unitrode Magnetics Design Handbook
(MAG100A), Section 5
4.
AN966; L6561R, Enhanced transition mode Power Factor Corrector, application note
5.
L6563S; Enhanced transition-mode PFC controller, datasheet
Doc ID 17273 Rev 1
AN3180
Electrical equivalent circuit models of coupled inductors and transformers
Appendix A
Electrical equivalent circuit models of
coupled inductors and transformers
A system of coupled inductors is a set of coils that share one or more common magnetic
paths because of their proximity. Because of this, magnetic flux changes in any one coil do
not only induce a voltage across that coil by self-induction, but also across the others by
mutual induction.
Accurate descriptions of coupled inductors use the reluctance model approach and its
derivations, which closely represent the physical structure of the magnetic element. This
approach is especially useful when dealing with complex magnetic structures, which is not
the case under consideration. Here a simpler method is used based on the terminal
equations describing the electrical behavior of the magnetic structure.
From an electrical standpoint, a system of m coupled inductors, is defined by m coefficients
of self-inductance, relating the voltage across any inductor to the rate of change of current
through the same inductor, and m·(m-1) coefficients of mutual inductance, equal in two by
two, relating the voltage induced across any inductor to the rate of change of current in
every other inductor.
Considering the important practical case of coupled inductors wound on the same core of
magnetic material, each inductor is commonly termed “winding”. Focusing on the case m=2,
a system of two coupled inductors, which are designated as the primary and the secondary
winding, is a linear, time-independent two-port circuit described by the following branchconstitutive equations:
Equation 21
v1(t)
v 2 (t)
=
M d i1(t)
M L 2 dt i 2 (t)
L1
where L1 and L2 are the self-inductances of the primary and the secondary windings
respectively, and M is the mutual inductance. Winding resistance is assumed to be
negligible.
Unlike L1 and L2, which are inherently positive, M can be either positive or negative,
depending on the voltage polarity of the windings relative to one another: a positive rate of
change of the current in one winding can induce a voltage either positive or negative in the
other winding. As shown in Figure 23, this is indicated by dot notation, which follows three
important rules:
1.
Voltages induced in any winding due to mutual flux changes have the same polarity at
dotted terminals
2.
Positive currents flowing into the dotted terminals produce aiding magneto-motive
forces
3.
If one winding is open circuited and the current flowing into the dotted terminal of the
other winding has a positive rate of change, the voltage induced in the open winding is
positive at the dotted terminal.
Based on rule 1 and on the sign convention of the terminal voltages and currents of two-port
circuits, it is easy to see that for the coupled inductors in Figure 23 on the left M>0, while for
those on the right M0 (and 0 k 1) always, the extension to the case M LO, therefore, as already mentioned:
Equation 32
M=
L A − LO
4
⇒
k=
L A − LO
4 L1 L 2
The advantage of this method is its low sensitivity to winding resistance and to the
impedance of the wire used for connecting the windings. It is not recommended for low
values of k because in that case it would be given by the difference of two similar quantities,
and the error might be high.
Whichever method has been used, the parameters of the equivalent circuit of Figure 25 (a =
n, assuming n is known) can be readily calculated from the third group of Figure 26 with
obvious symbolism change:
Equation 33
M
⎧
⎪ LM = n
⎪⎪
M
⎨ L l1 = L1 − L M = L1 −
n
⎪
2
=
−
=
L
L
n
L
L
⎪ l2
2
M
2 − nM
⎪⎩
The parameters of the equivalent circuit of Figure 26 (a = ne, useful in case n is not known)
are given by Figure 28, here re-written for convenience:
Equation 34
⎧ L a = ( 1 − k ) L1
⎪
,
⎨ L μ = k L1
⎪ L = (1− k ) L
2
⎩ b
with ne =
Doc ID 17273 Rev 1
L2
L1
37/39
Revision history
AN3180
Revision history
Table 4.
38/39
Document revision history
Date
Revision
14-Sep-2010
1
Changes
Initial release.
Doc ID 17273 Rev 1
AN3180
Please Read Carefully:
Information in this document is provided solely in connection with ST products. STMicroelectronics NV and its subsidiaries (“ST”) reserve the
right to make changes, corrections, modifications or improvements, to this document, and the products and services described herein at any
time, without notice.
All ST products are sold pursuant to ST’s terms and conditions of sale.
Purchasers are solely responsible for the choice, selection and use of the ST products and services described herein, and ST assumes no
liability whatsoever relating to the choice, selection or use of the ST products and services described herein.
No license, express or implied, by estoppel or otherwise, to any intellectual property rights is granted under this document. If any part of this
document refers to any third party products or services it shall not be deemed a license grant by ST for the use of such third party products
or services, or any intellectual property contained therein or considered as a warranty covering the use in any manner whatsoever of such
third party products or services or any intellectual property contained therein.
UNLESS OTHERWISE SET FORTH IN ST’S TERMS AND CONDITIONS OF SALE ST DISCLAIMS ANY EXPRESS OR IMPLIED
WARRANTY WITH RESPECT TO THE USE AND/OR SALE OF ST PRODUCTS INCLUDING WITHOUT LIMITATION IMPLIED
WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE (AND THEIR EQUIVALENTS UNDER THE LAWS
OF ANY JURISDICTION), OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.
UNLESS EXPRESSLY APPROVED IN WRITING BY AN AUTHORIZED ST REPRESENTATIVE, ST PRODUCTS ARE NOT
RECOMMENDED, AUTHORIZED OR WARRANTED FOR USE IN MILITARY, AIR CRAFT, SPACE, LIFE SAVING, OR LIFE SUSTAINING
APPLICATIONS, NOR IN PRODUCTS OR SYSTEMS WHERE FAILURE OR MALFUNCTION MAY RESULT IN PERSONAL INJURY,
DEATH, OR SEVERE PROPERTY OR ENVIRONMENTAL DAMAGE. ST PRODUCTS WHICH ARE NOT SPECIFIED AS "AUTOMOTIVE
GRADE" MAY ONLY BE USED IN AUTOMOTIVE APPLICATIONS AT USER’S OWN RISK.
Resale of ST products with provisions different from the statements and/or technical features set forth in this document shall immediately void
any warranty granted by ST for the ST product or service described herein and shall not create or extend in any manner whatsoever, any
liability of ST.
ST and the ST logo are trademarks or registered trademarks of ST in various countries.
Information in this document supersedes and replaces all information previously supplied.
The ST logo is a registered trademark of STMicroelectronics. All other names are the property of their respective owners.
© 2010 STMicroelectronics - All rights reserved
STMicroelectronics group of companies
Australia - Belgium - Brazil - Canada - China - Czech Republic - Finland - France - Germany - Hong Kong - India - Israel - Italy - Japan Malaysia - Malta - Morocco - Philippines - Singapore - Spain - Sweden - Switzerland - United Kingdom - United States of America
www.st.com
Doc ID 17273 Rev 1
39/39
工商网监
湘ICP备2023018690号
