300W pushpull not working

[cupoftea]

Hi,
We are doing a 300W pushpull
24vin, 32vout, 350khz.
(as atatched LTspice and jpeg).

The attached is how we have it, with opto feedback exactly as shown......its just giving 2.56v on the output....the opto driver (LT1431) is pulling so much current through the opto_diode that the output voltage is only 2.56v.

Incidentally, i once got rid of the opto feedback, and did direct vout divider feedback back to the LTC3723-1 (obviosly non isolated), and it THEN works fine.....its just when we put in the opto feedback like the attached it doesnt work........as discussed, the opto driver is pulling too much current through the opto diode unfortunately.
Do you know what might be wrong?...i changed the LT1431 and it still doesnt work.

..PS, i know the attached sim is unstable, but at least it gives >2.56 vout.....we could work on the instability, but at the moment, the opto is just braking the beck out of it, for some reason....

 

[Easy peasy]

It is possibly oscillating due to too much gain

put 560E across the opto led

check the feedback loop is not open circuit - or some other similar fault

check the opto is not dead

change R8 -> 2k2

 

[Easy peasy]

Also - in the rreal world - you need a good film / foil cap with short leads from the Tx pri site centre tap to gnd, with the fets nearby in a tight loop.

--- Updated Dec 19, 2022 ---

p.s. cap on comp to gnd should perhaps be not larger than 22nF

 

[cupoftea]

Thanks, i will try it, also, i wonder why they (in LTC3723-1 datasheet) suggest a resistor from VREF to the opto emitter?.....is it just to "get some more current flowing" in the opto, and therefore use the higher CTR?

 

[Easy peasy]

V ref via R to opto collector, forming the mid point and connection to the IC, opto can then pull this point to gnd for zero output .... ( opto emitter goes to gnd )

 

[cupoftea]

Thanks, i see what you mean, but if the FB pin of LTC3723-1 is grounded, (which it is) then the LTC3723-1 internal error amplifier will pull the COMP pin high anyway?

 

[Easy peasy]

400uA too low, gives the opto too much gain, adding the R allows say 2mA total ....