Current clamp circuit added to SMPS output

[Flyback]

Hi,
I am trying to do some reverse engineering on a 1kW power supply. Its 36Vin and 28Vout (step down converter)
I only have the not-very-informative schematic as attached. I don’t have any part values.

I am told it was designed as a voltage output SMPS, but the customer was using it to charge batteries, without telling the supplier. As such, the supplier has now decided to now fit a current clamp circuit (ringed in the attached) to the output of the SMPS, so that it doesn’t supply an overcurrent into the battery.

The thing is the “control circuit” ( as attached), looks a bit unusual to me. The current and voltage error amplifiers have unusal RC’s coming off their inverting inputs for a start. There was always a TRIM input, going to the TRIM pin of the DCDC module used, and this TRIM input is now being used for the added-in current clamp circuit to act on.

I unfortunately don’t have the datasheet for the DCDC module.

I am told that the existing current_error_amplifier can’t be used to do the output current clamping because it is “not accessible” (too difficult to get to).

The reference voltages for the voltage and current error amplifiers look to be connected to a degree. They are not independent. Do you know why they might have done this?

Do you know what’s going on here?

 

[Diver300]

The RC networks (R5, C2, R27 & C10) are probably to do with the changeover between voltage limit and current limit.

Clearly either U4 or U6 can sink current and turn on U5. Normally the output of one of those will be sinking current and the other will be near 5 V. When the load changes and the power supply transitions from constant voltage to constant current, the output of U6 will fall, so it will start sinking current. Soon after the output voltage will fall, and the output of U4 will rise, so U4 will stop sinking current.

I can't work it out in more detail but it looks like there is a possibility of oscillation at the change from constant voltage to constant current, and the RC networks may be to stop the oscillation.