smps design 2

[Steve]

Hi again,
A problem in push-pull converters are to achieve exactly the same
magnitude of switch current in both halves of the primaries to avoid
"flux walking". Is there any other ways than go for current-mode, to
achieve balance in this respect?
One thing I though of; is it possible to "take a snapshot" of the
primary-current through the "top" switch by a sample and hold circuit,
and drive the "low" switch until its current is equal?
This would mean that only the top switch is pure pulsewidth modulated
while the low switch is current controlled with respect to the top
switch.
Regards,
Stefan

 

[John Popelish]

Hi again,
A problem in push-pull converters are to achieve exactly the same
magnitude of switch current in both halves of the primaries to avoid
"flux walking". Is there any other ways than go for current-mode, to
achieve balance in this respect?
One thing I though of; is it possible to "take a snapshot" of the
primary-current through the "top" switch by a sample and hold circuit,
and drive the "low" switch until its current is equal?
This would mean that only the top switch is pure pulsewidth modulated
while the low switch is current controlled with respect to the top
switch.

Active balancing schemes are certainly possible, but the common
solution is to put a capacitor in series with the primary, yo
accumulate the difference and turn it into a voltage difference that
forces a balance.

 

[Steve]

John Popelish skrev:

Active balancing schemes are certainly possible, but the common
solution is to put a capacitor in series with the primary, yo
accumulate the difference and turn it into a voltage difference that
forces a balance.

Ok, I realize the capacitor is possible to use for half and full bridge
designs, but is it really applicable for a push pull design were power
is applied on the center-tap of the transformer?

 

[John Popelish]

Ok, I realize the capacitor is possible to use for half and full bridge
designs, but is it really applicable for a push pull design were power
is applied on the center-tap of the transformer?

Sorry, no. I read push pull and thought half bridge. My mistake.

 

[Genome]

Hi again,
A problem in push-pull converters are to achieve exactly the same
magnitude of switch current in both halves of the primaries to avoid
"flux walking". Is there any other ways than go for current-mode, to
achieve balance in this respect?
One thing I though of; is it possible to "take a snapshot" of the
primary-current through the "top" switch by a sample and hold circuit,
and drive the "low" switch until its current is equal?
This would mean that only the top switch is pure pulsewidth modulated
while the low switch is current controlled with respect to the top
switch.
Regards,
Stefan

You didn't explain that one very well so JP got the wrong idea until you
did. After all a push-pull convertor doesn't have a top switch...... Anyway,
before I get all pedantic and start dreaming about Ms Frostrup, I've lowered
my sights, there 'may' be a way.

During the switch dead time output current is isolated on the secondary side
as it recirculates through the diode things. In the mean time magnetising
current is flowing somewhere in the primary.... it's all idealised this is.

So, what you do is sample and integrate the sensed primary current during
the dead time with appropriate steering to the appropriate integrator based
on which switch was just turned off. Then you take the difference of that
and use it to adjust the on time of the switches in some sort of way.....

Anyway, current mode control makes the 'problem' go away.

However..... it does not work in a current fed push-pull convertor (or other
types of current fed convertors) because the input current is fixed so it
cannot differentiate between output and magnetising current.

That was a source of a giggle for me because there are lots of mil-spec
convertors out there that use current fed topologies and they are all
basically buggered as a result, unless you do something about it which they
don't.... so they are buggered. Don't tell anyone though and don't go flying
in flying things ;-).

Hmmmm.. If you want to play and look then get into that LTspice stuff. Keep
your models ideal and poke about, it can be very insightful.

DNA

 

[[email protected]]

Hi again,
A problem in push-pull converters are to achieve exactly the same
magnitude of switch current in both halves of the primaries to avoid
"flux walking". Is there any other ways than go for current-mode, to
achieve balance in this respect?
One thing I though of; is it possible to "take a snapshot" of the
primary-current through the "top" switch by a sample and hold circuit,
and drive the "low" switch until its current is equal?
This would mean that only the top switch is pure pulsewidth modulated
while the low switch is current controlled with respect to the top
switch.
Regards,
Stefan

It doesn't have to be balanced exactly. There is usually some dead
time. Most designs use current limiting for each pulse.

 

[Tony Williams]

Ok, I realize the capacitor is possible to use for half and full
bridge designs, but is it really applicable for a push pull
design were power is applied on the center-tap of the transformer?

Years ago that would have been done using two transformers
with centre-tapped primaries in parallel (more or less).

There would be the main power transformer, using a soft
magnetic material, and an auxiliary timing transformer
wound on a hard (near square loop) material.

The timing transformer would be designed to saturate well
before the main transformer and would run unloaded (usually
base drive power only), so that that the centre-tap current
was mainly mag current, allowing an easy sensing of the
approaching Bsat.

 

[legg]

Hi again,
A problem in push-pull converters are to achieve exactly the same
magnitude of switch current in both halves of the primaries to avoid
"flux walking". Is there any other ways than go for current-mode, to
achieve balance in this respect?
One thing I though of; is it possible to "take a snapshot" of the
primary-current through the "top" switch by a sample and hold circuit,
and drive the "low" switch until its current is equal?
This would mean that only the top switch is pure pulsewidth modulated
while the low switch is current controlled with respect to the top
switch.

It's volt-seconds that you're trying to equalize; differences in
magnetizing current are the result, not the source of the problem.

The primary current reflected from the secondary will have a peak
value that's roughly constant and determined by the output current
through the output inductor. Magnetizing current is superimposed on
this in the primary.

Gapping the core is often a sufficient precaution if the switches are
resistive ie fets. Their increased voltage drop, to the increased
peak-to-peak magnetizing current acts to balance the driven voltage.

The average voltage on each phase is measurable on a coupled winding -
their difference can be used to affect phase duty cycle to force
equality.

The instantaneous current is measurable on each phase, and differences
can be used to modify duty cycle without going for complete
curent-mode control.

Depending on your type of modulator, simply adding current information
into either the voltage ramp or voltage feedback lines could suffice.

You have to ask youself, in the end, just how much fiddling around you
are prepared to do with a circuit that may never have optimum
characteristics for your specific application. Difficulties of
modifying the control circuit to a current-mode version may be trivial
in many cases, as the basic elements for doing so are already present
in the voltage-mode control circuit.

Sticky keyboards in this weather......

RL