J-Test on APX can be tricky. I recall stereophile's J-Test measurements for Sonnet Morpheus being less than ideal and was called out by Cees
https://www.stereophile.com/content/sonnet-morpheus-da-processor-manufacturers-comment
Sonnet's comment about the J-Test here is completely valid, and you can see it demonstrated on other DACs. The J-Test requires a steep filter to be valid, and so doing it on a NOS dac for example will seem to show very poor jitter even if that isn't the case.
On the Rockna wavedream as an example from a different manufacturer:
J-Test in NOS/'Filter Off':
But with steep filter ('Linear'):
However this is for a J-Test, which is not the test I did on the U18 (or other DDCs).
The J-Test is used to test jitter at the analog output of a DAC, but with a DDC you can simply measure the jitter directly on the digital connection's clock signal itself, which is what my DDC measurements show. And this is not susceptible to the same issue as it is not dependent on reconstruction at all.
I didn't expect this much deviation from Gustard's own published measurements. Assuming the unit you evaluated wasn't bad, can you tally your J-Test parameters with Gustard's depicted here in this pic.
There's a few odd things about Gustard's own published measurements.
1) Not really sure why they're posting a THD+N test. This is a digital device, it's either bitperfect or not. Harmonic distortion could be added in DSP but assuming a DDC is bit-perfect a THD+N test is irrelevant. (The U18 is bit perfect, I've not encountered any DDC which wasn't so far other than the Chord MScaler which is of course intended to alter/upsample audio passed through it)
2) If I use exactly the same settings as they've chosen and play real music through the device (note that they do not show what waveform they were playing), I get the following:
Very very similar so there doesn't seem to be any discrepancy.
The RMS Jitter value though has no context for settings so I can't repeat it.
But, in the full screenshot they do show, there are a couple things to note about how they've configured things:
- They have their filters configured to the 'audible range' between 50hz and 22khz, which is not really what you should be looking at for jitter.
Jitter does not create a 1:1 result, ie: jitter at 1khz does not mean that you'll have unwanted noise at 1khz on the output, it means you'll have unwanted content 1khz either side of the fundamental frequency. So if you were playing a 15khz tone and your source had a lot of 1khz jitter, you'd have unwanted content at 14khz and 16khz.
For this reason, you cannot consider jitter in terms of a straight 'audible range'.
The AES3 standard specifies a corner frequency at 700hz and the APx555 defaults to a High-Pass 700hz and Low-Pass 100khz which is what I personally use for my measurements.
AP has a pretty thorough document on jitter as a subject which may be interesting to some:
http://www.audiophilleo.com/zh_hk/docs/Dunn-AP-tn23.pdf
- They are running the U18 at 192khz. This is fine, as some music is going to be at 192khz, but it is not representative of MOST music, and performance at 44.1khz and 192khz is going to be different. Additionally 48khz base rates typically perform better than 44.1khz base rates due to it being easier to clock divide between the 48mhz USB rate and 48khz sample rate than it is to go from 48mhz to 44.1khz.
This occurs on many DACs and DDCs, and is why I include J-Tests for both rates on DACs and jitter tests at both rates for DDCs.
In any case, you cannot compare 192khz performance to redbook performance.
Repeating the above test but at 44.1khz looks like this: