Since I had had two meetings cancelled, I had a little time. I think it's also a fun set of data to have when one goes for comparisons between electronics in general.
Of course I invite the real measurement experts to chime in, this is just what I would do with my current level of ASR education...
(so go easy on me!)
Since we are measuring analog output of the DAC (not vibes, not feelings, not “soundstage energy”):
• Speakers stay out of the test (the room has opinions - we don’t invite it)
• Amp stays out of the test (Yamaha gets the night off)
• We capture RCA output of the Topping (cold, hard electrons only)
1. Lock the digital path
On both PCs:
- Install Topping USB driver (if used - yes, even if Windows “helpfully” says you don’t need it)
- Use WASAPI Exclusive (or ASIO) in your player (shared mode is where accuracy goes to die)
- Disable all DSP, EQ, volume normalization (no “enhancements”, “loudness”, or fairy dust)
- Set sample rate to match file (no resampling, no creative interpretation)
- Set DAC volume to fixed (100% - we don’t touch it again)
Then verify by playing a DTS test file - it must decode correctly downstream…
(or a 1 kHz sine if you prefer less violence)
If DTS breaks into digital shrieking, Windows touched the bits.
If it decodes cleanly, congratulations - the digital path is innocent.
We want to ensure identical digital PCM into the DAC, not “mostly identical, probably”.
2. Measurement gear/setup
• MOTU M2 (or Scarlett 2i2 - science doesn’t care which logo we prefer)
• Good RCA → TRS cables (coat hangers need not apply)
• REW and/or Audacity (free software, unlike audiophile cables)
Physical setup
PC (test) > USB > Topping E30 II Lite
Topping RCA out > Audio interface line-in
• Do not connect the Yamaha or speakers (they add drama, not data)
• Same RCA cables for both tests (no cable rolling, please)
• Same USB cable (this is about stopping arguments, not starting new ones)
3. Level match (non-negotiable)
Play a 1 kHz sine at -3 dBFS
In REW:
• Measure RMS level
• Adjust interface input gain until both PCs measure within 0.05 dB
We don’t skip this.
If levels differ, louder will “sound better” - invalid test.
4. Baseline capture
For PC #1:
Record:
60 seconds of silence (yes, silence is very informative)
60 seconds of 1 kHz tone
60 seconds of multitone
Save as:
PC1_silence.wav
PC1_tone.wav
PC1_sweep.wav
5. Repeat above for PC #2
And save as:
PC2_silence.wav
PC2_tone.wav
PC2_sweep.wav
Resist the urge to “optimize” anything in between.
6. Null test (the kill shot)
In Audacity or REW:
- Align PC1 and PC2 recordings (sample-accurate, no vibes-based alignment)
- Invert polarity of PC2
- Sum PC1 + inverted PC2
Interpretation:
• Residual below -90 dBFS = identical for all human and most alien ears
• Residual shows broadband noise = probably power or grounding
• Residual shows tones = clocking or EMI doing something silly
With E30 II Lite, expect minimal residual near the ADC noise floor.
6. Noise spectrum comparison
Compare silence captures and check for:
• 50/60 Hz hum (hello, power grid)
• USB packet noise (~1 kHz multiples)
• RF hash (10–100 kHz, where gremlins live)
Expectation:
• Differences <5 dB at < -100 dBFS = irrelevant, inaudible, and not worth arguing about
7. Expected result
Prediction:
• Null below -100 dBFS
• No spectral differences
If after all this someone still hears a difference…
…it’s not the computers. It’s expectation bias then - the most powerful audio component ever invented.
