• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

What is the point of a J-test

fatoldgit

Active Member
Joined
Feb 29, 2020
Messages
279
Likes
326
I (think I) understand the underlying principle of the test but my simple analogy is it appears to be like testing cars by dropping them three stories up and using that as some real world test metric (along with acceleration, handling, skid tests, braking tests etc)

If our source into the DAC isnt jittery then, in my mind, a DAC that has poor jitter rejection and one that does wont (all other things being equal) sound measurably different (i.e. the one with poor rejection wont be placed into a stressful situation)

I am thinking mainly of an async USB input where the DAC controls the flow so input jitter is in theory low (not considering noise in this question).

So what weighting should be placed on the J-test when selecting a DAC, in comparision to other available metrics.

Wise/knowledgeable comments appreciated.

Thanks,

Peter
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,507
Location
Seattle Area
It is true that the test signal started for one purpose and we are using it "off label." While it sometimes unearths jitter over Toslink/Coax, it is generally used to see internal interference in the DAC. Take USB. It has a sub-packet timing that runs at 1 kHz. Often we see this in the j-test measurements:

index.php


You see those pulses at 1 kHz intervals? That is cause by interference between USB input and DAC output. We can also see odd issues like that rising noise floor. Or the general noise floor being high (both true here).

Now we could use a standard 12 kHz tone. But the J-test signal is clever in that it is a perfect sine wave that toggles all of its bits 250 times a second (in the 48 kHz version I use). That 250 Hz component must not be seen in the output of the DAC outside of what it represents. But often it too bleeds. You can see it here:

index.php


Net, net, the signal/test has proven itself in showing interference between digital and analog sections of the DAC. Sometimes these are interface jitters but most of the time are just poor design hygiene. No other test I run is able to show this effect as well as J-test.
 
OP
F

fatoldgit

Active Member
Joined
Feb 29, 2020
Messages
279
Likes
326
It is true that the test signal started for one purpose and we are using it "off label."

So your first sentence summarizes it, which in my reading states that the J-Test has been repurposed from its original intent to highlight anomalies in other parts of the DAC's topology.

Thanks for that.

So one other question, if I may.

Lets assume we have say a standalone DDC/reclocker and on the DDC input side we have all sorts of nice buffering, reclocking, accurate clocks, low distortion power etc (i.e. the signal being sent OUT is as good as it gets), then what is considered good intrinsic jitter on the output side (as measured at the SP/DIF outputs).

Thanks,

Peter
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,507
Location
Seattle Area
Lets assume we have say a standalone DDC/reclocker and on the DDC input side we have all sorts of nice buffering, reclocking, accurate clocks, low distortion power etc (i.e. the signal being sent OUT is as good as it gets), then what is considered good intrinsic jitter on the output side (as measured at the SP/DIF outputs).
If I understand your question, the spectrum of jitter determines its audibility and not any specific number. If you wanted a number, 250 picoseconds is enough to have equal weight to 1 bit of 16 bit audio sample. So you want it much lower than that.
 
OP
F

fatoldgit

Active Member
Joined
Feb 29, 2020
Messages
279
Likes
326
"the spectrum of jitter determines its audibility and not any specific number."

Thats what I have read in several places.

The DDC/reclocker I am looking at has AP measurements at around 60ps at the coaxial output so seems all is good.

Many thanks again.

Peter
 
Top Bottom