Switched Mode Power Conversion

Prof. L. Umanand
Department of Electronics System Engineering
Indian Institute of Science, Bangalore

Lecture - 35
Regulation of Multiple Outputs – I

Good day to all of you. So today, we continue from where we left off from the last class,
remember that we were discussing the topic of trying to regulate multiple output windings.
So, most of the power supplies generally have more than one winding, it could be a 5 volt
winding, 15 volt + - 15 volt supplies, 3.3 volts. So many, various of these standard
voltages, which are needed to drive many of the control boards or some other applications.
So, in general you will have 3 to 5 isolated power outputs, but you could, you can only
control one of the outputs through the duty cycle control.

So, we were discussing that what do we do with the other outputs, which are unregulated,
and how do we regulate them. We discussed couple of methods; one of them was to put a
linear regulator at the output of the unregulated power output, power supply output. The
linear regulator will have a small differential between its input, regulated input and the
final output, and this small differential voltage is what is going to cause a power loss.

However, we discussed that, we mentioned that we would put the linear regulator in those
windings, which are processing lower power, as compared to the control winding which is
processing very high power. The other method was to replace the linear regulator with a
non-isolated DC-DC converter like a buck converter, which has a local control, which is
supplied by either the output voltage or the unregulated voltage of the power supply output,
there by trying to achieve regulated final output. We shall today discuss 2 other methods.
(Refer Slide Time: 03:06)

The method of coupled inductor and another method, the method which is normally called
the magnetic amplifier method. So let us take these methods one by one, and try to
understand how they operate. Let me take example of a forward converter, isolated forward
converter.

(Refer Slide Time: 03:59)

We have the supply V in there is a forward converter transformer, which is connected in

this fashion to a power semiconductor switch. I am indicating the power semiconductor
switch by a BJT; however this BJT could as well be an MOSFET or an IGBT.
Now, the forward converter has a demagnetizing winding, which is connected in this
fashion, is coupled with the primary, however the dot polarities are opposite in this fashion.
And there are other windings, now these are the secondary windings that I will be drawing
winding 1, winding 2 and so on there could be many other winding, just now for the
purpose of discussion. Let us say that we have 2 windings which means 2 isolated power
supply outputs.

Now, the dot polarities for these are in this fashion and they are then connected to the
standard buck type topology, because forward converter is a buck derived converter. So
this is one of the outputs, and we shall call as V01, the other output similarly, another buck
derived topology to that secondary output connected in this fashion. And with some load
resistance and we shall call that one as V02, V02. Now, under the normal conditions the,
the primary and I am going to mark that in a different color, so these windings will all be
on the same core. This will be on the same core and so also the demagnetizing winding in
this fashion.

So, all the green colored windings represent one transformer, the forward converter
transformer. Now the inductance I am going to represent it with another color, now this
inductor and this inductor are wound on the same core they are coupled. So, these two are
coupled, and that is why it is called the coupled inductor method, coupled inductor method.

So, let us do some more naming this is secondary VS 1, this is secondary VS 2, I will call
this as point A, and the voltage here is VA, I will call this as point B and the voltage here
is VB. This is inductor L1 and this is inductor L2 they are coupled, so in the design of this
particular coupled inductor. Let us make the following put the following constraint let us
say VS 1, so let us say VS 2 / VS 1 = n, what I mean is this ratio to this ration is n, do not
confuse it to the ratios between secondary’s.

So, let us say if this is the primary winding, this is the secondary winding, this is the
demagnetizing winding, and this is the secondary winding 2, which is NS 2 / NS 1. So this
ratio is the ratio between these two windings n. So, we will try to maintain the same ratio
n even for the inductor windings these two inductor windings meaning if, if you have NL
2 by NL 1, where NL 2 is the number of turns on inductor, for the inductor L2, and NL 1
the number of turns on the inductor L1.
So, the idea here is that we try to maintain the ratio of the secondary inductors, windings
and the ratio of the secondary windings themselves to be same. The inductors are coupled
with this ratio the secondary windings are coupled with the same ratio that is the basic
concept here. And now, now let us say that one of the windings is controlled and only one
of the winding needs to be regulated by the coupled inductor method. Now, let us take V0
2, V0 2 is actually VB - VL 2, V0 2, the voltage here this voltage - the drop across the
inductor VL 2. Now, what is the relationship between VB and VA?

Now, VB = n VA, because V S2 =nV S1 this is coming from here. And what is the
relationship between VL 2 and VL 1? VL 2 = n V L1 and that is coming from this. So,
therefore V0 2 = n V A - n V L 1 = n (V A - V L 1), by substituting for V B this here, and for
V L 2 there so that you get n V A - V L 1. Now, this is nothing but V0 1, we may write
here V0 1 = V A - V L 1, V A - the drop across L 1 would be V0 1. So that is what we are
indicating here. So, therefore, this becomes V0 1 and overall V 2 = n V0 1.

So, it is an interesting relationship, now let us say we control one of the outputs through
duty cycle control. Then the other output is related only by n, which is a fixed physical
parameter which is the winding ratio and automatically the other also be controlled,
because of the coupling, so this basically is the concept. So if you feedback this through
to the opto into the control circuitry here, + -, we have a V0 reference from which V0
reference specify that it is V0 1 reference, then the output of that goes through PI
controller.

And then goes through PWM drive circuit to, so that the duty cycle here gets controlled
by sensing of V0 1, once V0 1 is regulated, V0 2 and V0 1 are regulated in this fashion by
only a fixed factor n. So, if V0 1 is regulated V0 2 is also automatically regulated, so this
way regulated output of the other winding are also achieved, so this is an interesting
method. However, in practice you will not get such a tight regulation as indicated by this
equation to understand that we have made some assumptions inherently, saying that the
diodes are ideal or if they are not ideal these voltages cancel off and they do not affect the
final output.
(Refer Slide Time: 19:54)

But in actuality if you see, when we write V0 2 = V B - VL 2, it is actually V S 2 - V diode

drop - V L 2. Now, this diode drop is nothing but the diode drop of this component and the
diode drops of the components is actually obtained in this fashion. So if I am having the I-
V characteristic of this diode, of this diode static characteristic. And let us say we have the
forward characteristic which is something like this for the diode. We are not so much
interested in the reverse characteristic for now, now depending upon the amount of current
that flows.

So, let us say we have some particular amount of current is2 that flows the operating point,
oh sorry, I have wrongly named the axis I am going to correct that this is, this is the V and
this is the I, and that is the sum current, is2 flows here. The voltage across the device VDS2
is at this value, now for the first winding which is given as V A - V L 2, V L 1 this is V S 1 -
V D s 1 - V L 1. Now this diode drop corresponds to this component here, and this component
will be carrying a current is1 through the secondary and let us say it is higher power and
therefore, higher current.

So, it is somewhere here is1 and its corresponding diode drop VDS1 will be much different
from VDS2. And therefore, and therefore, there will be a mismatch between these values
they will not exactly cancel as we have put the equations like this. So, if you include into
this those VD drops, you will see that there is cross mis regulation which occurs, and it
will not the tightly regulated like this equation the idealized equation as shown here. So I
just indicating to you what is the effect come in due to just one component, so the voltage
drop across V A and also across L 1 can vary, what we have not considered here further is
the resistance, winding resistance of the inductance.

So, if this is a small value of resistance due to the winding of the inductance, these two
would cause a non ideal term to come in, the voltage drop across this is nothing but iL if
this is L into RL. So in actuality the voltage here would be VS - VD - V L - i L r L =V0. So
this term and this term are load dependant and they will be different in both the windings,
so there will not be an exact match in coupling between these two in terms of the coupling
ratio n.

Because of that what you could say is you will get a regulation better than not regulating
and the regulation of the regulated one would worsen a bit compared to not coupling.
However by coupling you will get a regulation in both the windings. To an extent that it is
better than not regulating, but not as could as the linear regulation solution that we
proposed earlier. But in the linear regulated case there is dissipation so the efficiency will
be slightly lesser, but in the case coupled inductor case the efficiency will be better.

(Refer Slide Time: 27:34)

So, let us see one more technique, the magnetic amplifier technique, method. Here the core
principle is the manner in which we are going to saturate an inductor; we are going to put
an extra inductor component. And will call that one as saturable inductor that is we will
allow that to saturate indicated like that so this is called a saturable inductor, we would
like it to saturate under certain conditions. Now, if you see that we have a current iL that
is flowing through the inductor we will call this as LS, the saturable inductor. If you look
at this inductor its static characteristic called B-H curve, B is called the flux density which
= phi flux by the area of the cross section of the core. And on the x axis, we put H the
magnetic field, which is given by N i L by mean magnetic length the path length.

Now, if you take the B-H curve like that, you may have already seen or studied earlier the
B-H curve for all the magnetic material is in this fashion. So, this area is the area where
the core has saturated, both here in the positive side and the third quadrant around here.
This is the linear operating region. And you see that when it traverses in one direction it
takes this path, and when it is coming down it takes the other path and this is called
hysteresis effect. Anyway hysteresis effect is only going to cause loss, this is a kind of a
low pass filter effect we will try to understand this without the Hysteresis. So, what I am
going to do is erase this and make much simpler curve, so this is the saturation region a bit
more idealistic to understand, it is easier and this is the B-H curve. So beyond this it will
saturate let me call B sat, - B sat so what it basically means is if you give iL if you give a
value of current i L greater than i L 1, then this inductor will saturate.

Now, to this inductor if I now apply a voltage, to this inductor if you now apply a voltage
like that through some circuit. And if that voltage is VL then the current iL is given by
1
𝐿
∫ 𝑣𝐿 𝑑𝑡, and if VL is a constant you will see that the current is increasing if VL is constant,
VL current iL will increase in linear manner, it keeps on increasing and as iL keeps on
increasing till it reaches let us say the limit iL 1. The moment it reaches this limit, the core
saturates, it will not no longer go like that. In fact at that point, it will start taking a point
in this fashion the current will steeply increase, the voltage across the inductor will 1 it
𝑑∅
becomes 0, because there is no .
𝑑𝑡

𝑑∅ 𝑑∅
Thus the voltage across the inductor is also 𝑁 𝑑𝑡 , and in this zone, = 0. And therefore,
𝑑𝑡

the voltage across the inductor become 0 this falls down and because falls down the current
will shoot because the external circuits is 0 impedance, 0 voltage and if it is not limited
the current can shoot. So, we say that at this point the inductor has saturated the voltage
across it, as become 0 as it is now a short circuit it is as though the inductor is not there.
So this is the principle that we will be using. We will make the inductor to provide
impedance for some period of time.
Let us say dt, and then make it saturate and make it into a plain conductor during which
time it will not act as an impedance. And it will allow the power, the power to, the voltage
to pass along to the other side, so basically the inductor is being used as a magnetic switch.
Therefore, it is called a magnetic amplifier, and will see how we will use this concept that
we just discussed in the context of our converter regulation.

(Refer Slide Time: 36:56)

Let us take the typical forward converter, so let me put in this circuit switch you are
familiar; we have a demagnetizing winding which we have included. Now, let us say we
have one winding and to that winding, we now include the, let us say we have the other
winding which is the regular controlled winding with the standard buck derived circuit.
And this is actually the one that is fed back and used for control. Now, the other
unregulated winding is to be controlled in the following manner using saturable inductor,
now here let us include the saturable inductor Ls and then we have our regular circuit
diode. So, this is; this is the unregulated winding V0, this is V0 1 which is regulated. There
is one more item at this point; you are going to feed in some controlled current source
,which is being taken from the output, and fed in here; we may also put a diode here such
that it is only in one direction.

So, this is the circuit that we need to discuss now. How does this operate? Now, let us say
the switch Q. So Q is ON, when the Q is ON, the dot end becomes positive, the dot end
here becomes positive and at this point just before the inductor was freewheeling, but the
inductor here is presenting an impedance. So, there is a voltage drop across this therefore;
this voltage does not carry forward and appear here to turn off this diode. So the voltage
drop across this is entirely matched by this supply voltage, the inductor is continuing to
freewheel.

However, because of this voltage there is a current flow, where there is a voltage across
this and current flow in this direction which starts flowing in this fashion, in this path.
Now, you may ask how is it going against this diode? If the inductor current this
freewheeling is let us say for example, 10 amps and if this, this saturable inductor current
is around 1 amp, then it is 10 – 1=9 amp flowing in this direction. So, as long as this current
is less than this inductor freewheeling current, you will have a path through this to reach
back again here. So as this inductor current is increasing at some point this will saturate.
The moment which saturates this becomes a short, and once this becomes a short this
voltage appears at this point and reverse biases the diode. Then it becomes a normal buck
converter operation where the power is put into this inductor, charging it up.

And then when this Q turns OFF, when this Q turns OFF, this dot end becomes negative
and the non dot end becomes positive, the inductor is freewheeling in this fashion. This
diode is off because this dot end is positive, this would come in here and turn off this diode.
Now as far as this inductor is concerned, through this inductor there will be a current flow
in the reverse direction. Look at this source. From the output there is a DC current, a DC
current which is made to flow through in this direction. This is off, therefore, it will flow
in this direction and back again to its negative point.

So, it will start flowing in this manner during the time when this is off. So when the current
is flowing in this reverse direction through the saturable inductor, it is bringing down the
flux within the core. When the current is flowing in the negative direction in this area, it is
bringing the inductance flux value negative, lower somewhere here. So, let us say the
current as, current value is up to this point. If the current value negative value is at this
point, then the inductor would have reached here. If the negative value of the current is
there, then the inductor flux would have reached here or if it was here the inductor flux
would have reached here. So, you can make the flux value within the core to be anywhere
on the locus. It could be any flux value.
Now when you give a positive voltage, when you give a positive voltage; when this is
turned on, when this is turned on. This is positive and the voltage across the inductor is
positive the same value V in, n Vin. The current rate is decided by voltage across the
𝑑𝑖𝐿
inductor by L, will be . If this is fixed then this current rate will be more or less fixed,
𝑑𝑡

of course this is; this is got from the unregulated voltage V in. So more or less the current
rate will be near about fixed current rate, but the more important point is that the flux rate
also is connected to the current rate. So flux will also increase from the point it was left
previously during the off state, and keep continuing from that point to saturation. So note
that if the inductor were brought to much lower value with the same current rate, it will
take a longer time for it to go up to reach saturation.

If the inductor was brought down to a value here it will take that much longer time to reach
saturation. So, which means that, the inductor will present an impedance for a longer time
in a period to the input source voltage, which means that this will be freewheeling for a
longer time, which means in the standard buck converter notation, this will be the buck
converter, the buck derived converter; this portion will be in the off time portion. If the
inductor saturates less then it takes; is brought to a negative value relative, relatively higher
up towards the origin, then it will take a shorter time to reach saturation.

And therefore, this will lose its inductance property sooner and the voltage will come
across here sooner and on time is more. And therefore, you have a modulation there, so
which can be used for adjusting V0 2 locally here independent of this duty ratio, so that is
the principle we are trying to follow. Let, let us understand with some waveforms what is
happening, let me draw some waveforms for you. Consider, let me make some space. This
is your V0 2. Now let us have the voltage across V LS the voltage across that and let us
also write the flux waveform in the core of the saturable reactor.

And then let us also see what is the voltage that is appearing at this point, and will call that
point c and voltage at point c. So let us observe these three, now let us divide the entire
into time segments. During d t s, the switch Q is on during (1 – d) T s the switch is off. So
during d t s, and shall mark that portion (1 – d)T s and so on, repeats d t s, (1 - d )T s. Now
this d corresponds to d t s here, what you apply as the signal to the transistor Q. Now, let
us say that the moment the transistor Q is turned on, the voltage across V LS, the arrow
direction shown, has taken some positive value like that. And let us say, the flux in the
core was 0 to start with meaning we are here. This is the point where we are beginning.

And the flux from here starts increasingly linearly, so let us say the V LS is constant starts
increasing in this fashion. At some point here it saturates. Let me call that as ∅𝑠𝑎𝑡 . And the
𝑉
rate at which it increase is whatever voltage you have V S, 𝑁𝐿𝑠 , where 𝑁𝐿𝑠 is a number of
𝑠

turns on saturable inductor. That is the rate at which it increase. The moment it saturates
here, the inductance will lose its inductance property, the voltage across it becomes 0 and
it goes on there like that.

So, this continues still the end of the on time, at which point the Q is switched off. The
moment Q is switched off, the voltage goes negative, because the dot end becomes
negative and the non dot end becomes positive this is freewheeling, and the voltage across
the inductor starts taking a negative value. And this negative value is coming due to the
application of this, there is negative current flowing through this, negative DC current, and
this DC current is flowing through opposite direction there has to be a negative voltage
appearing across it.

And the effect of the negative flowing through that is to pull the flux down, the flux is
linearly decreasing, it is getting pulled down to this value, to some level. And now, let us
say it is gone lesser than 0 and now you again switch on Q; switch on Q, this jumps
positive. And this increases at some rate it has to go from negative, you see earlier it had
to go from 0 to ∅𝑠𝑎𝑡 ; now with the same rate these two are parallel, almost parallel lines,
these two are parallel lines. It has to go to saturation, it will take a longer time to saturation
and it becomes zero here and so on and this will start coming down.

Now, this rate at which it is coming down is controlled by the current source rate I here,
so that this current source will actually determine this fall rate. So therefore, you can bring
it to whatever level you would like to have the inductor come down to either here or here
or here or here. And then from there this will go parallel to this line, this will be parallel
lines because here the voltage is determined by the output. So, you could get actually
different points of times when the inductor can actually go down, go to 0, like that. So,
now let us look at the, output wave shape. You see the output at VC, at VC if you look at
here, it will be VS - this and when the inductor saturates during this time, that voltage
whatever VS voltage come down to C.
So at that time the inductance has saturated, there is a voltage across C and so also here at
this time inductance has saturated, there is a voltage across C. And if you vary the current
that is allowed through the inductor to reset, you can get different voltages and, therefore,
different outputs here. So this is the basic concept, we will study this further in the next
class, and try to evolve a circuit to implement this in the next class.

Thank you for now.