Ethan Winer was doing his cable test with a sansa clip as a source which doesn't do 24bit. Nobody questioned that. I think it is time to stop asking for 24 bit.
Quote on 16bit vs 24bit.
As for clipping, there is no such thing. If you have to find a fault in me, probably the only one is that I shouldn't respond to the thread.
I am not going to respond to this topic any further.
There's no reason to get upset when someone is asking questions about your test procedure. It is better to do a 24 bit ADC recording, even when testing 16-bit DACs. It is also reasonable to record at a higher sampling rate than the DACs. It is important to use a higher quality ADC than the DACs you are trying to measure. All of these are normal, reasonable expectations in testing, especially since you are trying to analyze low-level differences between these devices. Your measuring device, the ADC, should have lower noise and distortion than the devices you're measuring. If it's not, your results are likely measuring the interface, instead.
There's significant distortion in your second 1kHz capture. Clipping or not, this is not normal and needs to be explained if this is really what the DAC produced. It shows a problem with the device, the signal, or the test procedure.
Ethan's null tester is an analog device, so it works with whatever source you feed it. Since it was used to test for differences between cables, there is absolutely no reason to argue about the quality of the audio source. That's not what was being measured, so
any audio signal would work for this test. It could just as well be audio from an AM radio, and the result would still be just as valid.