Solution Finder Get Tech Support

on time too long with LN623

Posted by: RetoStraumann on

Hi
We use a LNK623 together with a custom EE13 Transformer (prim. 2mH) for a 1W 3.8V from 230V/50Hz supply and are struggling with two issues.
First: The feedback voltage is settling quickly (<2us) and the regulation is stable, but the on time of the fet is never below 1.6us. Even for no or very little load. The datasheet says it should go down to 0.7us. How is this possible?
Second: We would like to have the supply run continuously on 100kHz to prevent all audible noise (transformer is tip varnished), but need very high load (4W) to achieve this mode. How do we have to modify the transformer to get continuous pulses at 1W (not only at 4W)? Increase or decrease primary inductance? Change air gap? Something else?

Looking forward to get your reply.
Regards
Reto

Comments

Submitted by RetoStraumann on 04/16/2019

Hello? Can anybody help?

Submitted by PI-YY on 04/19/2019

Hello RetoStraumann,

Thank you for interested in using Power Integrations products.

the 0.7us is the minimum on time which mean the on time will always be bigger than 0.7us. So it is normal for you to get 1.6us.

If you want to get higher switching frequency at 1W, you might want to change the airgap to reduce the inductance. but you should not let the power supply run 100kHz it would trigger overload protection if the power supply continuously running at 100kHz. To avoid audible noise, any running frequency >25kHz should be fine. You can tune the inductance to achieve this.

Best Regards

Submitted by RetoStraumann on 04/30/2019

Thanks for your answer. We changed the air gap and we could improve the audible noise.
But there are still two things I don't understand: At the moment the supply switches always with 100kHz. Sometimes just one time, sometimes 3 times in a row (always with 100kHz) and then it waits a short moment and switches again. This wait states correspond to a frequency of 5kHz or 10kHz and are clearly audible. I assume I can not change the frequency, only make this wait states shorter. Correct?
And the second question. Why do I never see the 0.7us on time? Even with no load at all and very long wait states the on-time is still 1.6us. Does this makes sens? Under what condition do I see the 0.7us?
Best Regards
Reto

Submitted by PI-Kryten on 05/28/2019

Hi Reto,

Can you please share the PI XLS file for your design?

Best Regards,
PI-Kryten