SMPS Feedback control

Where can I find some good theory on SMPS feedback ckts.
I designed a flyback smps using TOPswitch, but don't know how exactly the feedback ckt works.
I think there is current as well as voltage feedback.
What exactly does feedback do? Does it change the PWM duty cycle?
Voltage FB is taken at the output capacitor. (I think) Current feedback is taken at the secondary rectifier diode. Why?
The FB is thru TL431 and an optocoupler.
ckt here -http://obrazki.elektroda.pl/20_1289496620.jpg

There is a single feedback path on TOPswitch flyback switch/controller.

Yes, feedback modifies PWM of switcher based on voltage regulation on output. The opto coupler isolates secondary from primary side AC mains. Units with 'M' input can be set to current limit on the primary side. For more precise current limit you can add another opto coupler with control fed from current sense resistor on output prior to voltage sense point. Output of two opto coupler are paralleled and fed to single feedback into 'C' control. Some cellphone chargers do this to set the charge rate current for their LiIon batteries.


**broken link removed**

Thanks for the reply.
The part I understand is the voltage divider taken from the output thru resistors R4 and R5.
This divided voltage goes to TL431, which has internal 2.5V reference. The difference between voltage divider output and 2.5v is the error voltage.
Depending on the magnitude of the error, the PWM duty cycle varies.
Am I right so far?
The part I don't understand is - why is the optocoupler anode connected before the inductor. Mainly the opto diode-R1 and R2 loop is the part I don't get.
Also what is the need for R6 and C1 in the TL431 ckt.

Can anyone answer please?

C9, R6 and R1 are providing a loop gain compensation. They are necessary for stable operation. R2 causes a bias current. It may improve the large signal behaviour according to the TOP204 properties. Analysis of TL431 circuits has been discussed in previous edaboard threads, if I remember right.

The part I don't understand is - why is the optocoupler anode connected before the inductor.

Click to expand...

It's not always possible to give a clear explanation. Some circuit details represent arbitray designer decisions. They must not be necessarily required or optimal. For a better understanding of possible design variants, you should test their effect with your own circuits or at least in a simulation.

I think one particular IC from LT does the sensing on Flybacks automatically without any external hardware. the video is on youtube. I do not recollect the partname though.

FvM,

I understand C9 and R6. They are just opamp feedback elements. Why they are used I don't know.
Also, what do you mean by R2 causes a bias current. Bias current for what? The opto or the TL431.
I thought R1 was to limit the current thru the opto diode.

They are just opamp feedback elements. Why they are used I don't know.

Click to expand...

So you didn't yet understand the concept of loop frequency compensation. Without C9 and R6, the TL431 would amplify the error signal by it's open loop gain, which would effectively cause the TL431 to set the LED current to either fully on or fully off, before the PWM could even react on the feedback signal.

I thought R1 was to limit the current thru the opto diode.

Click to expand...

Yes, that's also true.

what do you mean by R2 causes a bias current. Bias current for what? The opto or the TL431.

Click to expand...

Obviously for the TL431.

R2 is connected at the output of TL431. Why does the TL431 need biasing at the output. I don't get it. What am I missing?

Where can I read about loop frequency compensation. Any app notes or websites. I will try google in the mean time.

Why does the TL431 need biasing at the output.

Click to expand...

I didn't say that it's needed, I just guessed about a possible effect. It's not generally used in similar circuits. See my comment about arbitrary designer decisions.

Where can I read about loop frequency compensation.

Click to expand...

You can refer to general control theorie. The RC combination results in a PI controller characteristic.