Adding max current control to isolated AC/DC flyback converter!?

[thomasx]

Hi

I have an isolated AC/DC flyback controller design that I would like to add a maximum current regulation to the output on. I.e. the AC/DC is designed to give 12V 7A, but there are occasions when the load resistance will be so low that the current will be higher than 7A at 12V, in these cases I would like to limit the current to 7A or slightly above by lowering the output voltage, such that the current stays at 7A. When the resistive load increases such that the voltage can rise to 12V and the current drops below 7A, it shall go back to normal operation.

I am thinking perhaps this could be accomplished by a current sense resistor on the output, some kind of regulator and then an optocoupler to transfer the current control feedback back to the switching regulator and "inject" the current feedback into the switch controller feedback, such that the controller gets an increase in the feedback signaling that it needs to lower the voltage.

Does this sound like a feasible way to go about this?

Are there other smarter ways to do it?

[f4eru]

yes, that is a common way to do. You can even use the same opto, there are exampless of schematics around.
One important thing: if you have a flyback, with an AUX primary winding supplying the PRI controller, take care to get proper supply for the reg chip even when the output voltage goes down.
https://www.eetimes.com/implementation-of-the-primary-side-regulation-in-flyback-converters-part-1-of-2/

Now, often, you just rely on the primary current limitations, which is often good enough for protection of the circuit

[thomasx]

Great, thanks, I'll have a look into this!

[Faringdon]

...as f4eru says,  you just shovel your sense res on the output.....shovel that into an extra  error amplifier...and it will clamp your output to 7A (or whatever you want).
...and vout will fall.
st.com actually make a derivative of the TL431 with two opamps in it exactly for your purpose. I cant remember the name of the chip...something like TM101 or something.
Give me a shout if you want an LTspice sim of what you want...and i will send it to you....so you can see it in action...the current clamp error amplifier is normally saturated high, and out of the circuit, but as soon as the current tips above 7A, it swings into action and overrides the main  error amplifier...uusually just a bit of diode OR-ing between the two.

TSM101
https://www.st.com/resource/en/datasheet/tsm101.pdf

[thomasx]

Hi

Thanks for your help!

I think the reason for my question in the start of this thread was a bit hasty and based on some lack of knowledge

I am working with a flyback controller based on TI's UCC28C54Q1 controller, with a transformer with an auxiliary winding giving 15,5v, no opto-coupler from the secondary side. 230VAC input and 12V 7A DC output.

The controller has a voltage feedback, 2,5V and a current sense input, max 1V.

The controller has overcurrent limiting.

So I would have thought that the voltage feedback from the aux winding would help keep the output voltage at the expected 12V, and the current sense would manage the overcurrent protection part.

For a while there I thought I had it working pretty good, set at 13,5v at no load dropping to about 12v at full load and dropping voltage with 7A current at overload, with the exception of to much powerloss in the snubber circuit. So I swapped the 700v mosfet switch transistor for a 1700 volt SiC Mosfet one and recalculated the snubber circuit. And now the snubber circuit powerloss is not causing any overheating at least. But I seem to have totally lost the voltage regulation as it drops below 10 V even at 30w load, which is less than half of the max output power.

at no load I measure 13,6 volt output, Vaux 16,4, Vcs = 100mV
at 0,8A load I  measure 11,65 volt output, Vaux 16,4, Vcs = 300mV
at 2,3a load I measure 9,75 volt output, V aux 16,4, Vcs = 550mV

The original MOSFET was the IPX65R190CFD
The SiC MOSFET is the IMBF170R450M1

I adapted the design so that I am using the sense output from the SiC MOSFET. Sense resistor connected to source and then the MOSFET sense pin connected to the controller sense input via the filter

My design is basically the same as the application example i the UCC28C54Q1 datasheet, but with different component values of course. I am running a switch frequency of 42kHz.

It was originally based on a Webench design that was totally wrong, so I have struggled quite a bit to get where I am right now.

Any ideas on what I should look for to find the reason for the voltage drop so I can find a solution to that?

MOSFET Datasheets
https://www.mouser.se/datasheet/2/196/Infineon_IPX65R190CFD_DS_v02_07_en-1485079.pdf
https://www.mouser.se/datasheet/2/196/Infineon_IMBF170R450M1_DataSheet_v02_03_EN-3362317.pdf

UCC28C54Q1 Datasheet
https://www.ti.com/lit/ds/symlink/ucc28c54-q1.pdf?ts=1712065214192&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FUCC28C54-

[Faringdon]

It sounds like the sense output from the sic fet is not what you think it is.
Can you not just put in a standard fet, and use the current sense resistor in the source to do your (current mode) current limiting?
Have you used a output diode with trr<75ns?
Have you used a rcd clamp diode with trr<75ns.
Have you used a suitable RC current sense filter?
Whats your layout like?
Did you do interleave winding, and put the bias coil "Inside the sandwich"? (for good coupling)

How high does your pri switchign node go?....and is all your sensitive circuitry distanced and/or shielded from this noise_terror node?

SMPS layotu doc as attached shud you wish

Also attached a flyback offline in LTspice...you can fit it to your requirement, and get that working first, then go to the bench, it will save time.Sorry the diodes are old and need changing and youll need to insert your values. remember the leakage L.

[thomasx]

Hi

I don't think I misunderstood the SiC sense pin, it's basically the same as source but with a separate conductor from the substrate to avoid high current effects on the sense signal. So the sense resistor is connected between source and ground and the sense pin goes to the sense filter on to the sense input.

My sense signal into the controller looks good I think, no spikes higher than the sense signal itself.

There's a bit of noise on the FB signal though, perhaps that's it. But I am still a bit confused that at 9,75 output voltage the auxiliary voltage is still at 15,5 volts, is that normal? It's a Wurth transformer, so I don't know exactly how it is wound, but to the best standards for the application I would assume.

The RCD clamp diode is a UFS180Je3/TR13 with trr max 60nS

Yes I think the RC current sense filter is suitable 1k/470pF, I have also added leading edge blanking which was not there from the beginning. It didn't make any noticeable difference on the output though. For the leading edge blanking I used an NMOS with 100pF/33ohm for the timing. 15 volt Vdd/OUT.
The output diodes are V8P10-M3_86A and I couldn't find trr in the datasheet. What is the effect if the reverse recovery time is too long in the output diodes (two in parallell in this case).

The layout may be the culprit here. If it is, shouldn't I be able to measure somewhere the noise/spikes causing the issues?

Thanks for the doc and LTSpice file, I will set this up and have a look.