Coupled / group inductor in computer power supplies

[tchicago]

If I take apart a typical computer power supply, I find that there is no individual voltage stabilization for most high current outputs. Such as +5V, +3.3V and 12V just come out of the separate transformer secondary windings that have proportional number of turns, rectified and filtered. The stabilization is implemented with the primary side taking the feedback from just one of the outputs. So theoretically, if there is a high power consumption on the line that is used for feedback, it may cause other outputs to overvoltage, as the primary side will compensate for the higher power consumption, affecting the unrelated outputs.

However, all these outputs also run through one additional large inductor, which has multiple windings - one per each output.

Some sources call this inductor a "group stabilization inductor" implying that this arrangement somehow compensates the imbalance of the output power consumption. It is also said that one output can thus "borrow" power from the other via that inductor. I have not seen any theory explanation of how it works though.

Does anyone know what this inductor is actually doing? Is it just a cost cutting measure - to put filtering inductors for each output on one core? Or it is indeed some clever way to stabilize the output voltage?

[picture from Wikipedia with such shared inductor marked]

[opampsmoker]

So i guess its a forward converter...and the output inductor is coupled...ie all outputs use the same inductor  core and are thus coupled together.
Yes this does help them to cross regulate better.
Ill come back with some stuff on it.

Here on page 3 it mentions it
https://www.ti.com/lit/an/slua119/slua119.pdf?ts=1607676263808&ref_url=https%253A%252F%252Fwww.google.com%252F

..its DN-62 by unitrode, and written by Mammano

..as it says, its current mode control  that benefits from  the coupled inductors, not voltage mode control.

If you think of a flyback converter....it naturally kind of cross regulates because all the secondaries are coupled in the flyback core which is essentially a coupled inductor.....a forward converter  transformer is not a "coupled inductor".

If we ponder it, the most loaded  secondary is going to basically "rob flux" from the other windings via the coupled inductor...so they dont go into overvoltage..at least not as much.
Basically, if a coil of inductance L has current I  in it..then it will have flux density B in it.....but  a lightly loaded secondary will "see" a bigger inductor because theres loads of flux in the core from the heavily loaded secondary...thus the lightly loaded secondary "sees" a bigger inductor than normal...and so  for it, the current rise is impeded more...so it doesnt get overvoltaged as much as it would have.

Attached is an ltspice sim you can fiddle with...make the loads the same and the vouts are the same for coupled and uncoupled versions...but adjust the loads of each split coil and watch and see...the coupled one magically cross regulates them well even though they are differently loaded......the magic of the coupled inductor is out!...bottled by LTspice. LTspice is free download

[T3sl4co1l]

As said above, yes, it does actually help.

Additionally, only 5/12 cross-regulation is needed -- and they don't need to be terrifically precise, or unevenly loaded for the most part.  3.3V is in fact regulated separately!  This is done with clever use of some old technology, the key part being that black heat-shrink component in the middle, between the two transformers -- a saturable reactor or magamp.  During forward-bias, this component saturates after some time delay, the time depending on the bias current set during reverse-bias, controlled typically by a TL431 error amp or something like that.

Tim

[tchicago]

Thanks, opampsmoker, T3sl4co1l,

I think I understand: if the coupled inductor can be looked at as a current transformer, then the higher load1 will induce a current in the lower load2 inductor, opposite to the current flowing into the load2, canceling a part of it.
If that coupled inductor is looked at as a voltage transformer, then the voltage applied on the load1 side of the inductor in such a way, that on the load2 it will appear as a series voltage source generating an opposite voltage, lowering the voltage on the load2. Very cool.

That is all of course if that inductor is connected after the rectifier before the capacitor. This is something I missed when looking at the sample schematics of the PC PSU. So if there is a filtering inductor, then it should be of course after the filtering capacitor.

And the saturable reactor thing looks more complicated, I guess (I never understood the reactors well). On the picture, it is connected between primary and secondary (next to the optocoupler). So I think it has more than one winding - one on the primary side and one on secondary.