I'm giving my incompetent 2 cents here.
Imho the name of the problem here might be: "TPR22" spec sheet is totally misleading (read wrong) vs its real features.
Inside the box there is a WHS9415 dac-amp chip and apparently nothing else. No downstream amp I mean, unlike on the Meizu.
I dug I bit around
The WHS94xx family includes 4 chips.
9420 has the same specs as 9415 with the addition of ANC circuitry. Fair enough.
This distinction is interesting insofar as online I found this :
1543460058893017551.pdf "WHS9420 USB Interface Audio Codec Device Specification"
Page 22: Single Ended Output Performance, at the bottom:
- Full Scale Output Performance (PCMI):
0,95- 0,99 Vrms
- Output power (32Ohm load) 27 - 30mW
- Supported loads
4 - 32 Ohm <----- VEEERY different from 16-600 in'i ?
Put like that, the TPR22 has simply never been what it is apparently marketed for, that is, a dac amp dongle supporting headphones (too)
I personally use it to drive my E5000 (14Ohm) on which it produces great SPL, and very resolved sound already at 28-30/100 volume scale. With no perceivable distortion artifacts noise etc.
Yes, the volume scale appears absolutely "worng".
My suspect was that the chip's volume control is exclusively linear vs logarithmic.
After reading Amir's findings, may it be a wrong wiring or sort of ? It looks like someone set a "volume endscale" of 2V onto a chip which is only able to deliver 1V - that would sound "consistent" with Amir's finding it going nuts just about after halfscale (?)
As I said, my 2 cents nothing more