KSTR
Major Contributor
Thanks for these new measurements, and after seeing that I now have to step back from my previous impression that the TotalDac is a reasonably well engineered piece of gear. It's not, ghost notes in the audio band hitting -60dB (0.1%) is not acceptable.I was going to investigate something else in Totaldac but discovered some serious performance issues not covered earlier.
The test is simple. A 10 kHz tone generated by the Audio Precision analyzer, fed to both Grace Design Balanced DAC ($150) and Totaldac d1-six ($13,400). An ideal DAC would generate just the 10 kHz tone and nothing else. Here is what we get from both DACs, with XLR Output:
View attachment 30379
As you see I have set the bandwidth to a hard limit of 20 kHz. So no argument please about ultrasonic tones. Our concern here is the audible band.
The Grace Design Balanced DAC produces the expected results from a well engineered DAC. We have our pure tone at 10 kHz and tiny spurious tones at lower than -135 dB.
The picture is wildly different with d1-six. It produces spurious tones both above and below the 10 kHz tone we fed it. The ones below are especially problematic because they are not perceptually masked. Indeed, you may not have good enough hearing to perceive the 10 kHz tone but hear the lower spurious tones since they land in our most sensitive spectrum (2 to 5 kHz).
The tone also has broad shoulders instead of sharp narrow one in Grace Design. That indicates random but low frequency disturbances, usually in clock but could also be reference voltage modulation.
We also have mains leakage which we don't have any sign of with Grace Design.
None of the extra tones in d1-six are harmonic in nature. The second harmonic of a 10 kHz tone is 20 kHz which is cut off from the test.
Note that this test was setup and I simply moved the cables from one DAC to the other and re-measured.
The designer talks about 96 kHz sampling being better. It was not. But more on this later.
The above test is at 48 kHz. To see if we have some timing related issues, I changed the sampling rate to 44.1 kHz and got this (everything else the same):
View attachment 30382
As expected, Grace Design performance does not change. But the performance of the Totaldac d1-six degrades further with tons more spurious tones. We saw this in our original dashboard in the review.
Again notice that so much garbage is created at lower frequency than our 10 kHz tone.
I happen to turn on the FIR filter and was surprised that it highly magnified these spurious tones. Here is that output at 96 kHz sampling:
View attachment 30383
Wow. Now we have unwanted tones rising to just -65 dB or so relative to our 10 kHz tone!!! What the heck is going on? Why would a simple high frequency equalization also cause this?
You want to see ugly? Here it is at 88.2 kHz:
View attachment 30384
In whose book this is high-fidelity?
Who here dares explaining to me how these extra distortions make good sound? I fed the thing 10 kHz and I get all of the above in the bargain?
When we can design a nearly perfect DAC for just $150, how upside down the audio world is to produce what the Totaldac d1-six outputs?
Remember, all of this is with hard cut off of 20 kHz. So none of the arguments prior to this post about unfiltered NOS matter. These are all audio-band issues. NOT ultrasonics.
I am disgusted and going to sleep now.....
While it would be interesting to further investigate how these spuriae exactly change with frequency, frequency relation to sample rate, and level, I'd also say now: game over, case closed.