Opto-isolated offline Flyback feedback loop.

[ocset]

Hi,

This  post concerns offline flyback controllers that output a current from their COMP pin for feedback loop control purposes.  The COMP pin gets pulled down by a common emitter connected opto. Now, all  small signal feedback loop equations for error amplifiers deal with this situation by assuming a “pullup resistor”  connected to the COMP pin pulling up to Vcc of the controller. However, as discussed,  many controllers simply have a current source as described, and no actual pullup resistor. However, one still has to derive an “effective pullup resistor “ value for use in the feedback loop equation. How do you do this? Is the following method OK?.....
…………..----------------------……………….--------------------

The attached is a LTspice simulation of an LT1243 based offline flyback with common emitter opto connection feedback. The error amp in the LT1243 is set up as a ~1mA current source by grounding the FB pin. (the LT1243 is pin-for-pin like the UCC28C43)

Also attached is the small signal open loop feedback loop calculation document in excel (for this converter).
This shows that the “effective pullup resistor” at the COMP pin of the LT1243 must be ~10k.

{this was asserted because setting the pullup resistor to 10k in the excel doc revealed a crossover frequency exactly correspondent to what is shown in the LTspice simulation (attached). The simulation demonstrates the feedback loop crossover frequency by way of its vout deviation following a sudden load transient}.

The question is, the “effective pullup resistor” value of 10k is correct….but how is it calculated?

As an estimation, I would say that the fact that the error amp is outputting a current, means that in actual fact the “effective pullup resistor” value will vary with the load on the converter? Would you agree that it is pertinent to simply calculate the “effective pullup resistor” value by dividing the voltage at the output of the error amp (ie at the COMP pin) by the current value that it is outputting?…and do this for the case of maximum load?

...you then end up with the highest value that the "effective pullup resistor" could be. This is a good value to take because it will give the least phase margin, and so represents a conservative calculation. Would you agree?