OK, understood.Since the digital part of the test set up wasn't changed between any of these tests, any constant clock drift between DAC and ADC should not be an issue as it would be the same in all captures. In fact, it's safe to turn off drift measurement/correction in DeltaWave (and I've done it for @GXAlan 's recordings) -- it made no measurable or audible difference in the result. There are much larger analog errors than the 1-2ppm clock drift. This was also verified by capturing two files from each of the amps and then comparing them to each other -- the RMS null was well over -100dB (edit: meant to say well lower than -100dB)
IME this is a rare case when clock stability of two independent clocks is good enough for a deep null (despite any clock rate mismatch which is irrelevant when it remains constant, like you said). Devices need to have fully settled thermally and there must be zero disturbance during the test (like vibration and of course thermal conditions). Basically we have tested the clock stability of the SACD and the ADC as being truly excellent. It also of course depends on the length of the recording.
@GXAlan, any chance you might publish the recordings?
Excellent idea, using a splitter cable. The ADC's channel differences can be checked beforhand for a baseline, together with @dualazmak 's idea of double checking criss-cross.The sync issue is addressed by using just one channel (say L) from each amplifier into the ADC in stereo mode.
To dig in really deep, we would need to compare a DUT's output directly to the digital source signal. This will also include the errors from the DAC and ADC used but the baseline resolution of this can be determined, and the contribution of the DAC and ADC can even be factored out, by comparing DAC-->ADC to DAC-->DUT-->ADC (and preferably both recorded sample-synced). Prerequisites must be fullfilled, the main one being identical levels at the ADC (and identical polarity etc). When the differences are large enough, DW can handle this nicely.If so, without placing enormous additional workload on Amir, what additional testing would you suggest to include? Preferably it should be possible to refine the result to a single figure metric for easy comparison between devices.
This is what I did in my linked tests and by this I uncovered an ever so sligthly compression effect of my DAC (the signal slightly modulating the reference voltage of the DAC). In hindsight it also manifested itself in standard THD measurement at low frequency where the distortion is higher but from that alone it is impossible to infer it came from a dynamic Vref modulation rather than an actual simple nonlinearity.
A single figure of merit seldom works as a general concept from which a general behavior could be extrapolated, IHMO. For one specific test (the test signal -- music), tough, it is certainly possible to get a single value like PK-metric which has sufficient validity. Looking -- and actually listening to -- the residual offers more insight IME.
Still I think the verifcation run is neccesary to have a reliable result.
- So, you have two DUTs and get positive DBT result (with level matching, training period etc). I would think we all do agree that a DBT is required to establish that there actually *is* a difference worth further investigation.
- As explained, then try to emulate one DUT with the other by adding the residual to the input signal, do sanity check if it now measures the same (with the same test) to good enough precision (difference not way larger that run-to-run variations).
- Do another DBT (e.g., but not restricted to, an ABX) and when it is positive again (differences still reliably heard), chanced are high that something else is going on and we have missed it in this test.
The DBT is the critical part, the testers not knowing what is being played, otherwise prior knowledge might easily bias the result ("we have established that the emulation of amp A via amp B measures the same the real amp A, so why should there be any perceptual difference?").