Hi all, I’m just getting started with my QA403, and I thought it would be a nice idea to try and repeat some of the datasheet measurements on my WiiM Pro Plus (just because I’ve got it lying around). I’m not having a huge amount of success, though…
The reference measurements I am using is from this ASR thread.
The WiiM is configured for optical in, RCA out. Output volume is fixed at 100%, with line out level at 2Vrms.
Parameter 1: noise
Shorting the inputs, I can get a self-noise of the QA403 of -119dB A-weighted. That seems spot on. To measure the WiiM, I have used a BNC → RCA adapter from left line out to L+ (all other inputs shorted). Full scale input is set to 0dB.
My result: -115dB A-weighted
ASR result: -122dB A- weighted
I would therefore have expected my result to be -119dB, thus limited by the QA403, given the WiiM itself should measure up to -122dB
For testing the WiiM, I have used mirror mode to output the test tone over SPDIF.
SE loopback: -106.5dB
My result: -91.7dB
ASR result: -115.2dB
Again, I would expect my result to be roughly equal to the loopback result, given the true THD+N is as low as -115dB.
Before I go too much further trying to test more complex measurements, is there a simple explanation (very likely user error) for why I’m not getting the values that I expect?
You’ve got a huge raft of noise there, seems to be grouped at odd multiples of 13Hz or so, that’s dominating the THD+N reading. Maybe investigate where this is originating?
The spike at ~330Hz is intriguing too, plus the peaks 1 and 2 etc which aren’t signal harmonics, but odd harmonics of the 330Hz signal - 330 is a subharmonic of the 1kHz signal - definitely odd.
I note that there can be a problem with mirror mode and windows transcoding (quick search revealed this) - perhaps that can explain the dirty signal?
I have a Wiim but haven’t measured it (it went straight into service doing the job it was designed to do with Spotify).
The Wiim is based on the TPA3255, isn’t it? The A-weighted output noise with the TPA3255 is 85uV (AW), which is about -81 dBV. I have confirmed that on TI’s reference design before. If someone is measuring 40 dB better than TI’s reference design, they are likely not measuring what they think they are measuring. Either the gain is turned way down, or the amp is in power saving mode and not actually doing anything.
Note TI’s measurement was made with the input shorted, A weighting and an AES17 filter. You don’t need the AES17 filter to measure this. But you will need the inputs shorted.
thinking on this more, a quick measurement you could do is drive into the wiim with a -60 dBV 1 kHz tone. Confirm you hear that coming out of a connected speaker. It will be extremely faint, but should be perceptible if the room noise isn’t excessive.
And then, add a N+D measurement tile and measure across the speaker using scope probes (Wiim L+ to L+ In the QA403, and Wiim L- to L- on the QA403). This will report you the noise of the entire spectrum except where your tone is. Do an a-weighting and that should get you pretty close to the TI number I’ll bet.
Also, I agree with @MarkT that something is going on wrt thd and it looks like transcoding. The giveaway is those tones you have marked aren’t harmonics. They are potentially (probably?) byproducts of a sample rate conversion.
Finally, like most TPA3255 designs, I suspect the + and - outputs of the wiim are idling at the supply voltage / 2. So, if you ground the - output, you will fault the TPA3255. I point this out (maybe too often) because it tripped me up badly when I first was measuring the TPA3255. It’s a wonderful part. But very atypical output topology which can cause a lot of confusion.
Good luck, and please report back. These types of measurements you are embarking upon a interesting to lots of folks I suspect.
Actually it’s the WiiM Pro Plus streamer I am measuring, not the WiiM amp! (Although a project on the to-do list is build a TPA3255-based amp, so the resources you have linked will be very useful in the medium term…)
I’m actually currently developing an audio DSP; Windows completely messing with the audio signal, doing SRC here there and everywhere, is a familiar source of annoyance to me.
Re noise: for what it’s worth, I just tried a mains USB-C power supply for the WiiM, which is slightly worse than the plot in the OP which is using a USB-C cable plugged into my computer (as it includes mains hum).
I have also swapped RCA cables. The cheaper, longer one I used in the OP yields no really measurable harmonics. The shorter Van Damme “Pro Grade” has spikes at 100Hz and 200Hz…
Re THD+N: turning off “Audio enhancements” in Windows didn’t make a difference, and I can’t see any other special effects being applied in any of the settings.
Manually setting the sample rate in Windows to 192KHz has got rid of the 330Hz subharmonic (changing the sampling rate in the QA40x software doesn’t make a difference, but I’ve matched it at 192KHz here). But now I’m, getting peaks at multiples of 93Hz >_>
For reference, I did get the software sort of working on Linux using Mono, but I couldn’t get the generator to mirror the output to my computer’s digital out (it did manage analogue out, though, for some reason).
Hi @Michael1, sorry about that! I thought this was the class D version.
To measure the WiiM, I have used a BNC → RCA adapter from left line out to L+ (all other inputs shorted). Full scale input is set to 0dB.
For the noise measurement, I’d short the input to the Wiim Pro rather than drive it from the L+. The reason is that a short will always be lower noise than the output stage of the analyzer set to “off”
But now I’m, getting peaks at multiples of 93Hz >_>
The harmonics are now int he correct place, but excessive. And the 93 Hz are weird too. You might need to play around with Windows exclusive mode and try to force 48k. If the Wiim has an asio driver, use that with REW (don’t use a 3rd party asio driver as those often revert to Windows under the cover). And then, you could drive a sine wave created in Audacity at the native sample rate and see how it does.
Just a small update - with a fully fresh start and following all the instructions carefully, I’m now getting -102.5dB on the THD+N, with all the weird noise and harmonics gone. I can get on with measuring other things now
