it is probably unavoidable,.....because it sounds like the voltage drop across the body resistance of the output diode.......unless you did not design properly........what is duty cycle on full power?....is your primary current going up as far as it should on full power?

The first thing I would like to see is your schematic. Also the first thing to check is the bias voltage for the optocoupler on the primary side. Under No Output Load, you may have this voltage to be too low (therefore not enought bias for the opto transistor). If this is the case, you may fix your regulation by increasing the filter capacitor aftre the bias rectifier. Another solution is to increse number of turns on the bias winding. If this does not fix the problem, please post your shematic. Also you may take a look to the design examples section for designs using TL431. Thank you . PI-Worf

i tried with following configuration 1)0.1uF disc capacitor in bias 2)0.1uF is replaced with 10uF 3)external 12DC Constant power supply given to bias winding no changes found in load regulation(with load).from this i conculde no problem with bias voltage. i am using top261EN with 132khz (source & frequency pin short), i measured frequency in digital oscilloscope but i found frequency range is 55khz. i used TL431 loop configuration as per design example.plz suggest me loop compensation configuration which provide better load regulation.

tried with following configuration 1)0.1uF disc capacitor in bias 2)0.1uF is replaced with 10uF 3)external 12DC Constant power supply given to bias winding no changes found in load regulation(with load).from this i conculde no problem with bias voltage. i am using top261EN with 132khz (source & frequency pin short), i measured frequency in digital oscilloscope but i found frequency range is 55khz. i used TL431 loop configuration as per design example.plz suggest me loop compensation configuration which provide better load regulation

Would you please post your schematic or send it via a private message. This issue is definitely avoidable and the schematic will help us figure out what the problem is.

Cheers

PI-Chekov