Intrigued, I tracked this down: The second jitter test is online. (Page 1) - Reference - AVI HiFi Forum. This is real music, unfortunately, girl and guitar, 
- but, it's Joni Mitchell ...
- but, it's Joni Mitchell ...
love the dressing down Nugent got in that thread
but they made a mistake, despite the improved instrumentation and much lower jitter these days, he just keeps on inventing new and varied black 'arts' to flog his wares.
despite being adamant, he got the test wrong with 2 different sets of tracks; then invented some noise that was masking the effects, after being given the opportunity to chose a track that would show the difference.
but they made a mistake, despite the improved instrumentation and much lower jitter these days, he just keeps on inventing new and varied black 'arts' to flog his wares.
despite being adamant, he got the test wrong with 2 different sets of tracks; then invented some noise that was masking the effects, after being given the opportunity to chose a track that would show the difference.
Last edited:
The "whole system approach", or real world, in other words, is not that simple as amplifying output of one component by gain of second component. Interconnection of several components adds interference currents that flow through power supplies of both components (coupled through power supply stray capacitances). These currents have spectrum up to MHz range and they create voltage drop on interconnect cables ground impedance. This voltage drop is added to signal input and is simply amplified if in audio band or can create interferences, demodulations if HF. That is the real issue that is not revealed in simple numbers.
Hi Pano, thanks, I tried to order but Stereophile's webshop is out of order, and the file will be copyright protected so unfit for posting.
Can you hear it on your system?
Edit: Frank, thanks for the link
Intrigued, I tracked this down: The second jitter test is online. (Page 1) - Reference - AVI HiFi Forum. This is real music, unfortunately, girl and guitar,
Skimmed down the thread - looks flawed to me in that it assumes everyone's using a PCM (i.e. multibit) DAC in estimating how to render the jitter into the download. Yet nowadays the vast majority use an S-D one. With these you'd need to know some tech details of the particular DAC before you could incorporate the effect of jitter into the waveform.
You're asking me for a reference to the flawed assumption or talking about something else? No reference needed - if the guy doing the signal processing algorithm to factor in the jitter realized that S-D converters behaved differently he'd have given a choice of files where the jitter was calculated based on specific parameters of particular DAC chips. Then someone who was listening via a (say) CS4398 would get to play a different file from someone with an AD1955 in their soundcard. Only then would the '10nS' and '100nS' jitter designations be meaningful to users of non-PCM DACs.
I would suggest you try it and see but that might be too much to ask. Or you could try reading the DS, figure 12. The actual DC values for AOL, PSRR, CMRR are essentially meaningless and due to fine scale matching, i.e. I can find parts with 160db DC AOL.
Last edited:
I will have to say this whole business of PSRR as specified for opamps is very confusing, I have yet to find any material which clearly describes what will actually happen in the real world, for units used in a normal fashion. This conversation, When are ferrite beads the most approproate component to use? | Sci.Electronics.Design | Electronics-Related.com, to me typifies the lack of clarity on the matter ...
its not meant to emulate what jitter would sound like to each individual person in their DAC, it doesnt induce jitter, nor is it designed to test jitter rejection. its jitter/noise embedded the file the same for everyone, just as it might be if there was jitter in the ADC (its uncorrelated though, so thats less than totally realistic). your objection would be more meaningful if they were hearing something, but the vast majority couldnt hear anything reliably at all. perfect test? no, but still pretty revealing IMO.
that being said I dont agree with everything, all dac manufacturers always manufacturing the perfect design for their eval board, that cannot possibly be improved on? is pretty hopeful.
Last edited:
Yeah I figured it was the same for everyone, that's the weakness of the approach.
That would make more sense (given only one set of files) but its not really relevant to the audiophile brigade then is it? Jitter once introduced at the sampling stage can't be got rid of. Besides then why did the guy who set it up go to the trouble of trying to get something close to S/PDIF correlation to make it 'real world'? I'd kinda hope that a CD I'd bought didn't have >10nS of sampling jitter.
Yet I reckon they would hear something with 10nS of jitter applied to the average S-D DAC because that kind are potentially of the order of 100X more sensitive to it than multibit DACs (depending on internal architectural aspects like the presence of SC filters).
Revealing of what? That jitter doesn't matter if you're a multibit fiend (like myself) - well I knew that already. Film at 11.
Last edited:
Guilty I was looking At the graph on my cellphone and misread it. The correct loss of CMRR due to frequency is only 50 db not 60.
I posted a while back my PSSR & CMRR test circuit that is variable frequency and gain. (Two opamp version)
The issue as has been pointed out is that the datasheet value for any parameter is given under very specific conditions. As parameters such as frequency or circuit gain change so do SOME of the numbers.
When you design from a data sheet if you take worst case deratings for most production needs that will increase costs but probably not result in failures.
As a DIY project you can make actual measurements and design accordingly. Those parameters are not guaranteed, so in production if the non specified parameter changes, a problem may show up.
As the manufacturer did not guarantee or even specify the parameter, this is a risk that the designer should have at least considered.
I strongly suspect you are aware of the issues of typical versus guaranteed performance!
But the point remains if you look at a circuit such as JENSEN TRANSFORMERS, INC. - APPLICATION SCHEMATICS AS019 you will see a circuit where the AD797 is called on to provide 45 dB of gain (the rest is from the input transformer.) This circuit has poor PSSR. It also uses typical spec.s for the servos and does not always work with those values!
ES
http://ezphysics.nchu.edu.tw/prophys/ael/File/Datasheet/SSM2017.pdf
Note figures 11 and 12. Is the same graph available for the 797?
how else would you propose he did it?, how could you do anything but embedded a particular amount in a file, otherwise its different for everyone, it would be more or less attenuated depending on your gear. making it a test of the gear, not of the audibility of jitter
youve lost me, its not designed to be gotten rid of, its designed to be heard, or not, in the file, not in the clock, its not a test designed to see how well your equipment deals with jitter, its a test to see whether or not you can hear a specific amount of jitter in playback of a file in a way that ISNT gotten rid of by your DAC, otherwise that would introduce a heap of new and unpredictable variables into the equation, making the test 100% pointless.
I'm probably more anal than most when it comes to being jitter free, but I realize its a sort of sport for myself and somewhat academic, I shoot for less than 1ps, but i'm not pretending I can hear 10ps.
this is just a strange kind of chest beating rant, it has no bearing on the test whatsoever, since its in the file clocked correctly, not an error in the timing. as a stored file, the file itself doesnt have any absolute timing.
haha, yeah tell that to your multibit mates, who are just as swept up in jitter fever as everyone else.
Last edited:
I would propose he didn't bother. Or leaving that aside, that he pointed out that the figures (10nS and up) bore no resemblance whatsoever to your own DAC's jitter requirements for lack of audibility if you chose to listen through an S-D DAC. In effect the experiment is telling people with S-D DACs that jitter of 10nS is of no consequence, when it may well be in fact.
You equally well lost me here. Since we're talking at crossed purposes, let's let it lie eh?
From last time we got into discussion you were an ESS fan. So you have cause for concern over jitter, though I reckon 10pS is probably overkill.
- Status
- Not open for further replies.