problems in stabilization of feedback in smps full bridge

[falcosms]

Hi there!

i´m having problems to stabilize a feedback for smps full bridge 1200 W.

here is the schematic attached.

i´m using the uc3524 and optocoupler. i tried many values of R and C in TL431 and in the compensation of ic controller, but without sucess.

anybody can help me?

tks

Marcelo

 

[MicrosiM]

Whats the problem exactly?

 

[falcosms]

hello

i can´t stabilize the feedback and consequently the output voltage. if i remove the isolation by optcoupler, putting both gnd´s togheter and make a simple resistor divider in IC, works fine, but with isolation and tl431, i can't make the output stable.

 

[chacaton]

Dear,

Try to use some decoupling caps on pin 1 of TL431.If feedback is not a little filtered some noise can cause mis-function of 431, 10nF between cathode and pin ref (1 & 3 for you).
Sometimes you do not require compensation at all (R105, C90) try also without them.

I can saw that your pin number for 431 is not he same of on my designs pls verify .... I use without any problem TL431AIDBZ with 1:Cathode, 2:ref, and 3: anode (to GND).

Regards from France!

 

[KX36]

chacaton, your post seems strange. You recommend adding a capacitor between the cathode and ref of the TL431 for decoupling and filtering and removing the RC that's already there for compensation. It just seems strange you suggest adding a cap and removing a cap from the same place claiming they're for different purposes. The capacitor in that place isn't primarily for noise immunity and it's certainly not a decoupling cap in the traditional sense of the term, it is for compensation.

falcosms, your efforts so far look commedable. I can see that you've obviously read something on the TL431 to get the circuit you have, with its separate shunt regulated power rail and type II RC compensation network. This is necessary because the output voltage is higher than the 36V max of a TL431. and I think you aren't mistaking it for a zener or comparator as many do even in industry. Here are some of your problems:

1 - no output inductor. See my forum signature. You need one.
2 - SG3524 is an obsolete TL494 based voltage mode controler IC from the 70s. Make no mistake, even though the second error amplifier inputs on that chip are now named for a role in overcurrent protection, its still extremely poor and doesn't make the loop current mode. Current mode control is what you want. it's better for the full bridge transformer, tends to have a better performance and is easiler to compensate across an isolation boundary. You've already got most of the comonenents in the right places apart from IC specific ones, hopefully mostly just some values might need to change. I like the UCC28083 for push pull and full bridge.
3- The 10k + 1uF to ground on the compensation pin of the IC is wrong. This IC uses a standard opamp for its error amp, not a transconductance amplifier so compensation is not to ground, but more importantly you're not using that error amp you're using the TL431, so tie off the error amp on the IC whichever way is correct for that IC, inject the optocoupler signal straight in after the error amp which would be the comp pin on that IC, and make sure you know whether the optocoupler should be wired common emitter or common collector for whatever IC you're using. [Of course, as I said don't use that IC at all, but on the next IC, don't make the same mistake. ]

Read this too.
www.onsemi.com/pub/Collateral/TND381-D.PDF‎