Voltage mode CCM flyback has limitations on output capacitance

[Faringdon]

Do you agree that voltage mode CCM flyback has a hard limit on the amount of  output capacitance you may use?...(even if you compensate for it)
AN57 by Power Integrations puts across that this is so.
Do you agree with them?

[T3sl4co1l]

Might help to link the reference:
https://www.power.com/downloads/documents/an57.pdf
I don't see offhand where they discuss a "hard limit" but I'm not going to read it in detail just to explain how it's likely been misunderstood.

Also probably worth mentioning, many TOPSwitches are hysteretic, so the control compensation doesn't really matter very much.  Though I'm not familiar with the -GX in the article specifically.

Tim

[temperance]

It would be helpful to those who read your posts if you add all the required information in your posts instead of referring to application notes or spice simulations.

Like many others in here I do have enough stuff to read and study. I don't have time to read the app note in question and study what's in there or run your spice simulations to provide you with a sensible answer.

Back up your statements with calculations and other things and explain what you're thinking so people can judge if what you're on about might be correct or not. In other words: a question provoking a valuable discussion.

If you expect us to do your homework (the stuff we are usually getting paid for), then at least you could post the final outcome. Something you always omit and which I find pretty arrogant towards the people who provide you valuable feedback. Instead you just start your next ramble the next day.


Of course you will thank thank this post and do nothing about the way you communicate.

Do you agree?

[Faringdon]

Thanks,
AN57 says the resonant frequency between the effective secondary inductance (LE), and the output capacitor, should be >500Hz.....which , ayk, means that the output capacitor must be low enough in farads to achieve this......this is nonsense.......voltage mode flybacks have no constraint on output cap size.......as long as the cap is compensated for.

For voltage  mode flybacks in ccm, you should crossover at least 3 times above the said resonant frequency.....or well below it....and xover should also be 3x less than the RHPZ frequency.
May  you agree?

TOPswitch, when on full load is in pure voltage mode.

Also probably worth mentioning, many TOPSwitches are hysteretic, so the control compensation doesn't really matter very much.

Many power integrations flybacks are hysteretic type..but not TOPswitch.
___  -_____  -___
The statement made in AN57 is a fascinating one, and truly indicates a major new theory (if its true but it almost certainly isnt) about voltage mode control
The following "voltage mode vs current mode" by Mammano says nothing of the new theory in AN57
https://www.ti.com/lit/an/slua119/slua119.pdf?ts=1690059991837&ref_url=https%253A%252F%252Fwww.google.co.uk%252F

Also, neither of two Of Dr Basso's major books on SMPS control give any mention of it...and Dr Ridleys book on control (the smallish one) says nothing of it.

So,   much of the SMPS world awaits an answer....does Voltage mode control impose limits on the amount of output capacitance in VM Flybacks?....ie, given a certain transformer, are you then limited in output capacitance?......the answer should be no, because you can simply compensate for it........but....AN57...purports otherwise.

______________________ ___________-
Of more, related interest, is that on the Power Integrations Forum, one of the PI Apps engineers is actually saying that the  Modulator transfer function internal to the TOPswitch is secret....and the IP of Power Integrations.......which kind of nullifies AN57 altogether, because equations 22, 23 , 24 and 25 of AN57 purport to do just that (ie give the modulator transfer function of TOPswitch).

....So it makes you wonder...what are those equations all really about?

Here is the TOPswitch PI Apps guys response from their forum...
If you want a stable feedback loop, you will need to do measurements, as the TOPHX  changes operating mode as the load changes. The transfer functions are protected IP


___ ____
Not knocking PI....i  bet that any of their PI Expert or PI XLS designer designs run just fine.

Also, TOPswitch is now  updated by Innoswitch........which really does look to have some special features........and seems to be stable despite no feedback loop compensation effort being needed.
___________________________--- ___

And as a side note.....i actually tried to make the non-PI-Expert Topswitch design go unstable....and what i saw was quite interesting......there were some very uniform, regular, intermittent disturbancees, which only lasted some few seconds....then it stabilised out...it was almost as if there were some internal state machine inside the topswitch which was compensating for the deliberate instability that i had sewn in to the compensation components.

[mtwieg]

AN57 says the resonant frequency between the effective secondary inductance (LE), and the output capacitor, should be >500Hz

Yes they do make this recommendation a couple times. But nowhere do they imply that:

Voltage mode CCM flyback has limitations on output capacitance

This is you again interpreting off-hand recommendations as limitations. Reference designs and white papers often make recommendations that seem arbitrary. These are generally because the author thinks it's guiding the reader towards a better design without getting deep into the reasons why. Readers are free to ignore the recommendations if they feel they're not appropriate.

Obviously there's no fundamental reason a CCM flyback cannot work with a 10F output capacitor.

[Faringdon]

Thanks, it was indeed a recomendation at the bottom left  of page 4 of AN57. However, if such a recomendation is to be made, it must be backed up with some kind of reasoning...after all, this is the feedback loop we are talking about here...if it goes wrong, there can be potentially  100'000s of returned products.

As you ae cognisant, in a VM Flyback.....you have constraints already...eg, to stay away from RHPZ and the resonant point....so giving out willy nilly recomendations, i am sure you would agree, is a bad idea...as it may bring the cust into conflict with the other feedback loop requirements.



And what about the transfer function for the error amplifier and modulator for topswitch designs........is the one in AN57 to be trusted?
...If you may, take a look at equn 23 (AN57)...this is refuted by other versions of this equation in other documents , (ie docs not from Power Integrations.)

As such.....Why does AN57 include the open loop gain of the TL431 in the feedback equation (23), whereas AND8334 from onsemi does not include it? (AND8334 page 5)...Equn 23 of AND8334  does not contain the "KTL431" factor.

[mtwieg]

However, if such a recomendation is to be made, it must be backed up with some kind of reasoning...

The authors aren't obligated to do any such thing, especially if they make it clear that they are making a recommendation, rather than a requirement. And you are not obligated to apply their recommendations either.

And what about the transfer function for the error amplifier and modulator for topswitch designs........is the one in AN57 to be trusted?
...If you may, take a look at equn 23 (AN57)...this is refuted by other versions of this equation in other documents , (ie docs not from Power Integrations.)

As such.....Why does AN57 include the open loop gain of the TL431 in the feedback equation (23), whereas AND8334 from onsemi does not include it? (AND8334 page 5)...Equn 23 of AND8334  does not contain the "KTL431" factor.

It's unclear which equation in AND8334  you are comparing with. At a glance, the two circuits are different, and therefore the transfer functions should be different. I don't notice any errors in either.

The authors of AN57 perhaps thought that the open loop gain of the TL431 might be of significance to the reader, while the authors of AND8334 did not. So what?

[langwadt]

is this like betteridge's law, default answer is no

[temperance]

Think about what will happen if the output cap is not charged to the required Vout before the soft start period has been completed.

[Faringdon]

The authors aren't obligated to do any such thing, especially if they make it clear that they are making a recommendation, rather than a requirement.

Thanks, but as you know, having a limitation on the output capacitance, would  tend to mean that the output LC resonant frequency would be higher. This is a problem, since as Basso says on page 532 of his Book (Designing control loops for linear and switching power suppplies), the crossover frequency in voltage mode should be some 3x the output LC resonant frequency. Also, ayk, the crossover frequency should be some 3x lower than the RHPZ (if the SMPS  has RHPZ). If you limit Cout, then you can end up with a extremely small range in which you can cross over with stability.

They can say what they like, (ie they gave a recomended limitation on the Cout value) of course, i agree, but its unwise in this case, and unhelpful to the customer.