Headphone jack circuitry needs a deep inspection

[Djhg2000]

That's much more reasonable, and thanks for the effort. It's not exactly unknown for DACs to misbehave with 0dB(FS) signals so that may be what's happening here. Please post sufficient info on test method and settings for others to reproduce your results.
Edit: and what load is this into?

The Scarlett Solo 3rd Gen is specified at 60 kOhm in line input mode so that's a very small load to drive. I'll see if I can wire up headphones as well next time I'm at work, I don't think I have enough adapters to do that at home.



Currently this is connected through a 3.5mm to RCA cable (left channel from the phone) into an RCA to 6.3mm plug adapter (second channel on the audio interface). All faders in alsamixer set to -0dB unless otherwise specified. The analog gain setting on the input is set to match close to -3dB in the Audacity VU meter, with exception for when clipping occurs in which case it's set slightly below the level where it starts clipping (not applicable for these particular tests but it happened a couple of times during other experiments).

As a complete side note, the previous graphs with Lollypop turned out to not even be with a flat EQ. That wasn't the entirety of the issue though and the remainder of it is being tracked here: https://gitlab.gnome.org/World/lollypop/-/issues/2564

[wibble]

Thanks for the added details. The details for REW would be good too - which parameters you used for generating the frequency sweep and for the subsequent analysis. I haven't got round to soldering up the necessary test bits, but if I do it should be easy to include a dummy or real load to see how that changes things.