Read here about my findings on Dosmix TPR22 is, and its quite mysterious goods and bads.
I can easily say that the most mysterious of all aspects can be summarised as follows: how can two dongles having so similar specs such as Meizu HIFI DAC Pro and Dosmix TPR22 behave so dramatically differently in terms of power delivery ?
In addition to my own direct findings all regarding low-impedance and very-low-impedance IEMs, other owners of the same product (or one of its siblings – being called Chartek TPR22, or BGVP T01 / T02) recently ran into equivalently difficult to explain experiences on the opposite end of the impedance spectrum: so much more powerful Dosmix is vs Meizu on low end IEM, as much it is disappointing on 150 or 300 Ohm loads, where it shows extremely weak output.
Recently a review came up on Audio Science Review about the TPR22. Read it all of course following the link, but in summary:
- SINAD is very good
- Max output volume is 1Vrms.
- Very low, and surprisingly so.
- Meizu’s (and other “good” dongle competitors claiming equivalent specs) reach 2V.
- Max output of 1V is reached at 52/100 volume.
- Above that it’s all distortion!
- There’s a lot of jitter, but it’s inaudible so who cares.
- Meizu is similar.
- Max power delivered vs 300 Ohm load = 2.8mW.
- That’s ridiculously low ! Meizu, by comparison, delivers 10mW.
- Max power delivered vs 33 Ohm load = 26mW.
- Much, much better. Meizu is at 49mW.
- Output impedance: 0.6 Ohm.
- Exceptional ! Meizu is @1.3 Ohm
In short: the plot thickens. Then suddendly… a ray of light.
The TPR22 (and its apparent siblings: Chartek TPR22 and BGVP T01 & T02) are all declaredly based on Qualcomm’s WHS9415 DAC-AMP chip.
No downstreamp amp chip is integrated in there – unlike what happens inside the Meizu, featuring a TI OPA1622.
Here Qualcom publishes WHS94xx family’s spec sheet. Interestingly enough, there’s no mention to supported impedance load ranges.
From that spec sheet and further digging around, we understand that 9415 is the second-best chip in the family. 9420 has 2 additions: ANC circuitry (for the microphone part ofc) and a “differential operation mode” option, which if I well understood is the base for a possible balanced-out implementation.
Such similarity is very important, in light of the fact that after quite some digging I found this crucial document: “WHS9420 USB Interface Audio Codec Device Specification”
There, at page 22, in the “Analog HPH single-ended outpu performance” table, the butler gets nailed:
Supported output load: 4 to 32 Ohm ! Whaaaa? It’s 16 – 600 Ohm on marketing papers!! No way ! Four to thirty-two Ohm !
So very simply put: TPR22 never has been what someone marketed it for. It is actually (and interestingly) a quite unique category variant: it only supports low-impedance, even very low-impedance loads (IEMs), for which it acts as a specialised, ultra-budget, effective tool.
It (now obviously) fails miserably on higher impedance ( > 32 Ohm) loads.
On low impedances instead, it’s super effective – as reported in the specs, as additionally measured by ASR who found an exceptionally low Output Impedance which is particularly important vs lowend loads, and 100% consistently with my findings.
I thought to open a digression here on why and how an amp with TPR22 specs is “better at driving low impedance” IEMs, and actually I did write it. Then I realised it became too long as a digression. So I published it under a separate post. Here it is.
So is TPR22 the endgame dongle?
Certainly not.
Firstly, as I tried to outline here above, it simply is not designed to properly power higher than 32Ohm impedance drivers. And that would be already more than enough.
Futhermore, it’s got a few rough edges. One above all its volume control.
One aspect appears consistent accross my findings and those of all other users I spoke with is that volume progression is “odd” at best.
It appears linear, not logaritmic.
Looking at the already mentioned ASR’s report, another suspicion could be a wrongly set endscale value?
Be as it may, the issue is serious: a wrong move on the source device while listening to music risks to shoot a huge current spike into the drivers (burning them), and a terrible SPL shout into the listener’s ears seriously risking auditory damages. That’s quite worrisome tbh.
That said, assuming to keep its limitations under close “manual control”, I simply am not aware of any other option right now to add good-SQ E5000 (and possible Tin P1) driving ability to an existing source with the same ease, and even at 3 times TPR22’s price.