@matt, really great having pc mirroring back. I was wondering what sorts of numbers you were seeing with the E70 you bought fairly recently.
No. After trying lots of permutations on attenuation, I still can’t get snr better than 99db. That’s with balanced out, pins 2 and 3 into l+ and l- on my qa402. Tried pin 1 grounded and not to bnc shell. Not any significant difference.
Sorry, I might be missing something. I asked if you are seeing something unexpected, and you replied “no” but then you indicated you can get SNR better than 99 dB. Are those statements contradictory?
Not sure what results I should expect from the QA but a little surprised that the snr was so low and that I couldn’t find a configuration to get a better number given that I get so much better from another adc.
Hi @Moto, Are you sure that Windows isn’t doing a rate transcoding? That is a certain way to botch the SNR. If your DAC is showing 44.1 and your QA40x SW is at 48, then Windows will unhelpfully transcode it for you, and the results aren’t good. Most DACs that you plug into Windows default to a pretty lousy mode of operation, and the only way to know is via spectrum analysis.
Additionally, you have to make sure your windows volume is set to max, that Windows hasn’t decided to add “bass boost” or EQ etc.
So I left the city where the E70 is but here I have a QA403 and smsl su-9n. Using the same config, at 4v(-.5db) in to the QA(12dbv), atten at 18db, both QA and windows/dac at 48k full volume, no effects on, I get thd -128db and snr 110.5.
The results with Cosmos adc are thd -128.6 and snr 120.2db.
Is that what I shoukd expect from the QA403?
I don’t know as I don’t have this DAC. What are you seeing for THDN and SNR if you use an app like REW to generate the tone and the QA403 to measure the tone? In other words, do you think the mirror generation is deficient?
I thought you had the E70 Velvet which is very similar.
I can’t get a wiring config that gives me a decent balanced output measurement.
Using the rca out I get thd -121.7db and snr 116.4 when using PC mirroring. When I use REW to generate the tone, I get thd -122.7db and snr 117.2. This difference while not significant, is consistent over many runs.
The above is with a 5.8 dbv input signal at 6dbv fs.
If I change to 12 dbv fs of course the thd is better but the snr worse. The differences above between pc mirroring and REW remain though.
What is the optimal cable config for measuring balanced out from the dac into the QA403? I am still struggling with that.
Last question. In pc mirroring, if the QA is in Stop, but the GEN button is pressed is the signal meant to be on to the external dac?
Yes, I have an E70 but I haven’t used it since we did the last round of experiments here.
For cabling, I’d take the XLR and split it into the + and - (make sure the the adapter isn’t merging the - with ground). And then run the + and - into the + and - on the QA40x. And the ground to the BNC shell.
I just checked, and when mirroring, if the GEN button is toggled, then the mirroring starts/stop.
@matt, that is what I tried in the first example in this thread. I have an xlr connector with test clips on the leads and used that to connect to the QA + - inputs. Tried with the braid clip on and off the bnc shell.
Rca still much better.
Maybe I need to make up a cable with xlr and 2 bnc connectors. Hmmm.
Hi @Moto, I’ve been doing some experiments with different audio modes on the E70 Velvet looking into the mirror mode issues you’ve pointed out. I have the latest Thesycon drivers installed. But I cannot get Wasapi Exclusive to give anything closed to Wasapi Shared mode.
I wonder if this is some of the confusion?
Here is -1 dBFS in Wasapi Shared:
And here is -1 dBFS is Wasapi Exclusive
In both cases there’s a notch + gain so the peaks aren’t as you’d expect. But the E70 hardware is set to -1 dB with the big knob, Windows volume is 100%, and SigGen1 was -1 dBFS. But you can see DAC noise is mostly the same. The harmonics are what is hurting the THDN. And that says that re-sampling in Windows is happening.
I will try playing with that. Why did you set the knob on the e70 to -1dbfs? I thought that should be at 0 dbfs and siggen 1 at -1dbfs would then send a -1dbfs signal to the e70 in mirror mode.
I will try playing with that. Why did you set the knob on the e70 to -1dbfs?
No reason in particular.
Here’s another plot, this time with volume knob at 0 dB. This uses a 1 dB step from -10 to -1 dB, and then a 0.1 dB step from -1 to 0 dB
Was this through a notch?
Yes, it’s the notch in the QA480 tweaked for noise with a balanced front end (same +18 dB post-notch gain). Here’s dynamic range (127.8 dB) of the E70 unweighted
Weighting improves the above about 3 dB. The 30 dBV reference is comprised of 12 dBV from E70 and 18 dB notch gain).
Here the same notch with inputs shorted (no weighting). The notch box might benefit from some additional gain (another 6 dB to bring it to 24 dB).
Back on the wasapi issue, using the topping supplied driver for the e70, there are no switches for wasapi modes in the driver dialogue in device manger in windows 11.
It’s a bit buried in Win11. From System->Sound look for “More sound settings”
And then
If you untick that, then various apps will be forced to use shared mode no matter what they want. If you tick that, then certain apps can ask for exclusive mode. In exclusive mode, you won’t have to worry about another app playing sounds over your test tones. For example, the next release of the QA40x software will allow you to opt for Exclusive mode, and then if you start playing a youtube video, the youtube app will be blocked from playing on that particular sound card (the primary device you are testing).
But what I’m seeing is that once you grab exclusive mode, the THDN gets degraded a lot.
This seems odd, because so many hi-fidelity players are claiming to use wasapi exclusive mode precisely so that you get bit-exact playback…
Don’t know how I missed that! It seems to make no difference in measurements on my machine with the QA.
If I use REW with the cosmos adc and use the java interface instead of an asio one, then shared is worse than exclusive.
The Thesycon driver is asio right? I wonder how that plays in the scheme of things.