[hellenic vanagon]

Quote:

 
Is the 24 bit jitter still better if you use the 16 bit test signal converted to 24 bits ?
 
 
I meant there is a possibility in theory that jitter can be subtracted by the ADC, since it is not uncorrelated. Although it may or may not be an issue in practice.
 

1)I really don't know. Talking about ST, it will be very enlightening to see measurements before and after modification for 16 bits jitter.
 
2)There cannot be subtraction. Just "masking" which equals, somehow, to adding. 
 
Please have a look here:
 

 
 
 
 

 

[stv014]

Quote:

Originally Posted by hellenic vanagon /img/forum/go_quote.gif
 
2)There cannot be subtraction. Just "masking" which equals, somehow, to adding. 

 
Well, whatever you call it, it is possible for the ADC to reduce the measured jitter, since it is run from the same clock, and low frequency variations in the clock will affect both at the same time. By the way, this is from a cheap Xonar D1 card at 48 kHz/16 bit (the various peaks are either part of the test signal (which includes a low amplitude 250 Hz square wave), or are just interference and still present even if nothing is played):
 

 
Edit: for comparison, also the original test signal:
 

 

[hellenic vanagon]

Quote:

 
Well, whatever you call it, it is possible for the ADC to reduce the measured jitter, since it is run from the same clock, and low frequency variations in the clock will affect both at the same time. By the way, this is from a cheap Xonar D1 card at 48 kHz/16 bit (the various peaks are either part of the test signal (which includes a low amplitude 250 Hz square wave), or are just interference and still present even if nothing is played):
 

Do you mean "increase" instead of "reduce"?
 
 

 

 

[stv014]

Quote:

 
Do you mean "increase" instead of "reduce"?

 
If I have two sound cards, one with a real sample rate of 47999 Hz (both DAC and ADC) instead of 48000 Hz, and another with 48001 Hz, and do a loopback test on both, which recorded signal do you think will be "faster" ?

 

 

[hellenic vanagon]

Quote:

 
If I have two sound cards, one with a real sample rate of 47999 Hz (both DAC and ADC) instead of 48000 Hz, and another with 48001 Hz, and do a loopback test on both, which recorded signal do you think will be "faster" ?

 

I don't know.
 
Can you explain, please?
 
 

 

[anttipi]

Bump. Is the test wav still up for download somewhere? 

 

[stv014]

Try these files: jtest.zip

 

[anttipi]

Thanks! 
 
I want to see if there's any difference between my Realtek ALC892 mobo vs. Cambridge Audio 651BD when fed via Toslink to CIAudio VDA-2 DAC. The Cambridge player supposedly measures pretty well. ALC892 specs state S/PDIF output jitter max 4 ns. In any case, I can hear no difference between them in my modest setup so this is just out of curiosity.

 

[stv014]

By the way, the utility I used for generating the signals is available here. It is a command line program, and can be used to generate various other test tones as well. For the jitter test in particular, write something like this to a simple text file:

 JTest 0 30 s96

That means generating the JTest signal, starting at time 0 (immediately at the beginning of the file) for 30 seconds, with the least significant bits toggled at 96 sample intervals (so a low level square wave with Fs/192 frequency is added). Then run the program as follows:

 testgen.exe jtest.txt jtest.wav 44100 16

jtest.txt is the name of the above text file, jtest.wav is the output file, and 44100 and 16 are the sample rate and resolution.

 

[anttipi]

For some reason, I get significantly worse results using the 44.1 vs. 48k test files (max difference is hundreds of picoseconds judging by one or two high spikes that are missing when using 48k). EDIT: After some further reading, it seems those spurious unmirrored spikes are ADC artifacts.

For the results below I used the 48k file (mainly because Dunn mentioned 48k in his paper).

CIAudio VDA-2 DAC: From any somewhat decently clocking source it is ~115dB. Sources used were Toslink out from Asus P9X79 motherboard, RME HDSPE AIO PCIe soundcard and Cambridge Audio 651BD Blu-Ray player. Interestingly, no difference whatsoever when fed via coax out from CA.

Cambridge Audio 651BD: From analog stereo outs ~125dB.

I was very disappointed to find out that the Portta/Monoprice HDMI de-embedder I've used to back up my SACD's failed miserably. ~85dB with a shockingly wide skirt around the 12kHz signal. Plus the "baseline" was up there at -100dB. :rolleyes: l'll be getting my hi-rez music exclusively as downloads from now on. The RME AIO has a reclocking DAC but I'm not sure if it can clean up a mess like that.

On a positive note, I was pleased to find the noisefloor on the VDA-2 was very nice and stable in my presumably somewhat noisy surroundings. However, the same can't be said of the Cambridge Audio which showed some unpleasant tendencies: strange pulsations, asymmetry and instability in the noisefloor.