• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Measuring DX7 Pro+ using Cosmos ADC, or Why You Shouldn't Blindly Trust SINAD Scores

BlinkDL

Member
Joined
Apr 2, 2022
Messages
7
Likes
40
My Cosmos ADC is A-grade and set to 3.5 Vrms. All RMAA measurements are 24/192KHz.

TLDR: The THD performance of DX7 Pro+ is great at -1.0dBFS but worse at -1.2dBFS. All ESS DACs have this "oscillating THD" issue (see https://archimago.blogspot.com/2022/07/hi-res-thdn-vs-output-level.html).

[DX7 Pro+ XLR @ 0db] to [Cosmos ADC], stereo mode:
1664639860122.png


[DX7 Pro+ XLR @ 0db] to [Cosmos ADC], mono mode (ignore crosstalk - it's meaningless for mono mode)
1664639869177.png

These are probably the best RMAA scores ever, especially considering I am using Cosmos ADC without APU.

So all good? Not so simple.

Firstly, the USB isolation of DX7 Pro+ is great, but not perfect. Note I am using battery-powered laptop for measurement, so only the DAC is connected to AC power in the whole measurement system.

Noise level with usual USB connection:
1664639909890.png


Noise level with optical-isolated USB - much cleaner:
1664639917142.png


The -3dbFS 1khz sine has some HF noises (even after USB-isolation) but probably inaudible:
1664639975184.png


Now here comes the interesting part.

I switch to REW for 384KHz measurements.

The -1.0dBFS 1khz sine is exceptionally good: THD+N = -123.4dB in mono mode. This is APX555 level result.
1664639963297.png


HOWEVER, if you set it to -1.2dBFS, the THD is quite worse – an increase from -129.3 to -123.5 !

1664640025291.png

This is because all “SINAD king” DACs are using THD compensation, but it is only effective for certain volumes.

Hence I suggest in the future we should measure SINAD for [0dBFS, -0.1 dBFs … -1dbFS] and show all of them, to see the average and worst-case scores.

Now you might ask: is it possible that Cosmos ADC is especially good at -0.27 dBFS and bad at -0.47 dbFS? So I produce the same signal level using DAVE.

DAVE [email protected] dBFS = -125.2dB, [email protected] dBFS = -125.8dB. It’s more consistent.
1664640036883.png

1664640042805.png

DAVE has higher noise floor (and rising noise below 50Hz), but the HF spectrum is cleaner.

Note: I am showing all devices at their best. DAVE’s RCA has 4.7dB lower noise (so the XLR results will be worse), and DAVE’s USB must be isolated (or there will be more HF noises).
DAVE’s real THD+N is -120 (because here I set DAVE to -1dB volume), which is noisier than ES9038pro DACs, but with very competitive THD.

DX7 Pro+ XLR IMD:
1664640216898.png


DAVE RCA IMD - cleaner harmonics (do note the noise floor is higher):
1664640226320.png


The THD+N of a modern DAC is dominated by noise. I believe THD is more audible than noise, and we should use a better score like [THD * 2 + Noise] instead of [THD + Noise].
 
Last edited:
Top Bottom