We're currently working on a small 24V AUX-supply and have been testing the INN3167C-H101 in a normal flyback configuration. The schematic is pretty much as shown in the datasheet application-example. Unfortunately, we're struggling with audible noise, that's very significant. For the transformer, we've initially been testing with a Wurth 7508116314, with a primary inductance of 450 uH.
- At idle/no load, there is no audible noise and the switching-frequency is more or less random (just a few cycles here and there).
- With a bit of load, the switching-frequency is fixed and gives a clear noise, equal to the switching frequency. It starts around 10 kHz or so and increasing the load increases the switching-frequency. In this case there's no clear jitter, more or less just a fixed frequency.
- From 20-30 kHz, there's no audible noise
- At higher loads, the converter switches between 30 kHz and 100 kHz, which generates a clear wideband audible noise, in the range 10-20 kHz. Also, the load-regulation seems much worse and the average voltage drops.

The bias-winding uses all electrolytics and I've tried to replace the ceramic snubber capacitors, but the noise-source seems to be the transformer.

So I guess the question is:
At idle/light load: Is there anyway to force slightly higher switching-frequency or a bit of jitter, to minimize the audible noise?
At higher loads: The switching-frequency is 30/100 kHz, yet we clearly have a lot of noise in the audible-range, 10-20 kHz. Any good ideas what can be causing this?