Flyback converter voltage stability issue

[Ravi Kumar S]

I have designed flyback convertor using pi expert, the output is 5v and 12v both are isolated output. 5v has control circuit using TL741 shunt regulator. The problem is i can only get 12v when 5v consumption is 300mA. If 5v rail is 1A or above 12v rail will be above 18v. Please anyone help me to understand the mistake and fix the issue. It’s my first try. I have attached schematic image for reference

[Someone]

No mistake, only cross-regulation

[nctnico]

What happens if you load the 12V rail with a couple of hunderd mA? Goes it back into regulation? What you see on the output might be caused by the transformer pushing a little bit power into the 12V output due to the leakage inductance. If there is no load on the 12V, the output will rise.

[mtwieg]

These sort of multi-output flybacks typically don't regulate well with very light loads. Light load cross regulation is mainly a function of coupling between the output windings, IIRC.

Poor cross regulation with high loads is more likely because of transformer secondary resistances (and possibly poor transformer coupling as well).

Can you share details on your transformer construction?

[Ravi Kumar S]

Thank you for support. That is the problem it was designed to provide upto 4A. But it not stable. If i drive 5V at 200mA 12v output with 1A. 5v is table and 12v become 11v with 1A.

[Ravi Kumar S]

Hi  mtwieg,
I have attached transformer construction details

[Ravi Kumar S]

How to reduce cross regulation with separate ground. I needed isolated output for my project.
Please let me know is it possible.

[nctnico]

TI has an appnote on this:
https://www.ti.com/lit/an/snva562/snva562.pdf

Likely there are many similar appnotes to be found that can help with the design. But the regulation for the 12V output will never be super accurate. From what I've seen from multi-output power supplies is that 12V typically is somewhere between 11V and 13V. If you need 12V to be accurate then you could consider increasing the voltage a bit more and adding a linear post-regulator.

[Faringdon]

The only real way you can get any kind of half decent  regulation in those 2 outputs , is if they are always only ever both fully loaded on constant load all the time.
You should try and arrange the turns so that the duty cycle woudl be the same for each when on the full load.
There are all kinds of tricks to "improve" cross regulation in any SMPS....pretty much none of them really work.
There is one method using active inductors, to allow diode conduction or not.
Then theres  bridge converters with coupled output inductors...but its all pretty rotten.
You can also try multiple interleave winding.......but it takes much room on the bobbin.

[mtwieg]

Looks like a reasonable design overall. Could maybe get better coupling between outputs if you split the 12V secondary into two windings and sandwiched them around the 5V secondary.

Bifilar wire for the secondaries would be even better coupling, but might not be practical from a construction standpoint (especially if the secondaries have only a few turns in them).

The turns ratio between the secondaries is 2:1, so I would only expect around 10.5-11V on the 12V output when both outputs are equally loaded (assuming the rectifier diodes drop ~0.7V).

What are your actual requirements for cross regulation?