Synergy?
- Thread starter Ethan Winer
- Start date
Ah. That's more issues. He is using a generic oscillator for the source and not S/PDIF signal from a real device. It is typical to do that as it solves the sync problem as the frequency/phase is then predictable but does deviate from what we want to measure.
I think his test nets out to changing impedance does increase jitter. I am working on characterizing the same but will be keeping the impedance the same, and the source a real digital audio signal.
I think his test nets out to changing impedance does increase jitter. I am working on characterizing the same but will be keeping the impedance the same, and the source a real digital audio signal.
Sorry, I'm not following you? The Legato is a DAC (one of Jock Homos from AR-T). I'm not sure why you think that there is no SPDIF from a real device being measured??
??
If you give me the link I can read the context better. But based on what you post there was a mention of a "osc." Osc is short for Oscillator. I take that to be an instrument, not a transport. And I am talking about a source which won't be a DAC.
Just to put some context on this - the first graph I posted is of an ART Legato DAC outputting SPDIF & it shows jitter of "4.9psec rms, ~28psec peak to peak." He reckons that this is so clean an actual device that "The results are so clean, that .. maybe.. it would even be possible to look at cable / connector differences?
Here I would like to show some shots - maybe- interesting?"
Which are the graphs I posted.
Firstly take note that graphs 2 & 3 are zoomed in to show what's happening in the audio freq band - the frequency range on the graph is from 0 - 50KHz whereas the first graph is showing a zoomed out frequency range of 0 - 2MHz
So graph 2 shows the RCA connector & his notes on it that the data dependent jitter (ddj) is very evident as spikes in the audio freq range (the blue trace)
Graph 3 is an add-on graph that shows the reduction in these spikes with the addition of an RF attenuator.
It's a pity that graph 1 (all BNC connectors) & graph 2 (RCA connector) are different scales but the jitter has risen from 4.9pS to 9.7pS.
Here I would like to show some shots - maybe- interesting?"
Which are the graphs I posted.
Firstly take note that graphs 2 & 3 are zoomed in to show what's happening in the audio freq band - the frequency range on the graph is from 0 - 50KHz whereas the first graph is showing a zoomed out frequency range of 0 - 2MHz
So graph 2 shows the RCA connector & his notes on it that the data dependent jitter (ddj) is very evident as spikes in the audio freq range (the blue trace)
Graph 3 is an add-on graph that shows the reduction in these spikes with the addition of an RF attenuator.
It's a pity that graph 1 (all BNC connectors) & graph 2 (RCA connector) are different scales but the jitter has risen from 4.9pS to 9.7pS.
Last edited:
The link is at the start of my post!
Edit: I was trying to find the "osc" reference before I replied to you & see it now & see why you were confused by it. Yes osc would normally refer to oscillator but in this case it refers to oscilloscope input being at the 1Mohm setting
Last edited:
Again for anyone interested there is another such detailed treatment of SPDIF input stages with measurements & graphs here http://www.audiocircle.com/index.php?topic=90454.0
This time done by Jocko Homo of AR-T
Again, a similar style of test set-up
He then connects various terminations to the end of the cable, DAC, RF attenuators, etc. & shows the scope shots & analysis.
In case anyone is unfamiliar with Jocko Homo he is a hugely experienced RF engineer of high repute who has been working in this field >40 years (I believe)
This time done by Jocko Homo of AR-T
Again, a similar style of test set-up
He then connects various terminations to the end of the cable, DAC, RF attenuators, etc. & shows the scope shots & analysis.
In case anyone is unfamiliar with Jocko Homo he is a hugely experienced RF engineer of high repute who has been working in this field >40 years (I believe)
Last edited:
My head is spinning from the latest conversations. I'm SO glad I just trust my ears! They haven't failed me yet........and they aren't so daggum complicated to follow.
That said, now back to our normally scheduled programming.....*violently shakes head*
That said, now back to our normally scheduled programming.....*violently shakes head*
I didn't mean to take over this thread & go so this technical on it but it just seemed to evolve into where we are at the moment. Feel free to branch this thread & hive off the posts to a separate thread, if you like!
I see that now but it is pretty uninteresting differential. These are very small numbers and takes away the reason to have BNC over RCA if the data holds. 9.7 picoseconds is still very low number even though he used an RCA cable.
Just translating for others: the standard defines an impedance of 75 ohms for either end. The issue is that due to its design, RCA connectors cannot maintain that so in theory any cable with that either end, which is typical a lot of S/PDIF equipment, is not optimal. BNC locking connectors are designed for this use and do achieve 75 ohm. In this test he used a very clean async USB to S/PDIF device and measured that with BNC end to end and got the 4.9ps. Then he put in a cable with RCA at one end then an adapter to BNC so in theory, our impedance is now different. Yet he finds that jitter only rose to 10 picoseconds. TO put things in perspective, you need 500 picoseconds to cause enough jitter distortion to match one bit of a 16 bit audio sample at 20 Khz. So in theory, you want your jitter to be lower than this. At 10 picoseconds, we are easily achieving this.
BTW, for my testing I am trying to determine the effective impedance of the different cables.
Yes, glad you picked up on this as I was about to post something about these numbers.
Agreed, they are very low jitter numbers BUT:
- remember that is RMS jitter, & not peak to peak jitter 9.7rms is about 60p-p
- The Legato is a very good DAC which absorbs almost all of the reflections that get bounced back from the RCA impedance mismatch.
- So in a more "normal" type of DAC the jitter figure would probably be very much higher due to the multiple reflection bounces from end to end. See the Audiocircle thread that I linked to for a typical commercial DAC & look at the scope shots from that.
- Maybe these levels of jitter are considered inaudible but if you look back a bit on that thread you will see a graph of a stock Hiface giving 28pS RMS jitter & a Hiface with external power supply (ie.e no longer powered from USB) with 15pS RMS jitter. He states that he can reliably identify these apart. I agree with him as this is what I did in my first modifications to the Hiface & people who have both stock & external powered Hiface say that the sound is so different that it is hard to conceive that these are the same devices.
- I think the point is that even at these small numbers, depending on the type of jitter, it's spectrum & amplitude, it can easily be audibly identified.
Your explanation is concise. Are the jitter figures you quote RMS or P-P? Are they for random or deterministic jitter?
I can attest to hearing the effect of putting a BNC/RCA adapter on a cable (I.e changing it's impedance).
I can attest to hearing the effect of putting RF attenuators on a SPDIF cable. Many others also hear this difference - I can give you links
I can attest to (many others also) changing only the PS of a Hiface giving an easily identified difference to the sound - so according to his measurements going from 28pS RMS (164p-p) to 15pS RMS (88p-p) jitter is very noticeable
So maybe we underestimate the detrimental effect of even very small amounts of data dependent jitter - if that is what we are dealing with here?
Anyway, all these numbers are fine but I urge everyone to try one or more of these RF attenuators & see what they think. Trust your ears
EDIT: I'm interested in the results of your cable impedance measurements
Last edited:
I noted the peak to peek values in the post and they are still pretty small numbers. That is a distortion floor of -115 dB.
Well, we can only go by the data in front of us and this one doesn't show significant difference in connectors. In addition, some of the mismatch here may be due to the adapter and not just the RCA connector.
I don't doubt that he heard the difference. I doubt that the difference exists
.The margin error here is well within placebo effect. I have done similar tests and I can convince myself the difference that exists doesn't. And just like that it vanishes. I like us to stay in the domain of measurement for now as that argument can go on forever.
I don't agree that it is easy or remotely so. If it were the case, we would be able to put them in blind test and have the reliably identify the change. Do you think they can?
I always use Peak to Peak as these as jitter spectrum is unknown and hence RMS values don't make sense in my book.
If they are sighted tests, then I don't need to see them. I believe that they think they heard the difference. Problem is, I can't tell if they did or did not.
Maybe but the case is weak if it is just subjective sighted tests.
I will report them soon.
Amir,
Easier to reply without quoting you.
Yes, I have done some of these tests I quoted blind but without rigorous scientifically controlled DBTs there will always be doubt. However, I am happy that this is not placebo just as confident about it as I am about hearing the difference between vinyl playback & CD playback even though it is sighted. I guess the easiest way to test any of this is to try RF attenuators & do a blind test yourself. The RF attenuators, if reducing reflections is truly their mode of operation, will only effect the jitter numbers by something like the small values you see here.
EDIT: Just for others reading this - we are talking about hearing the consequences of jitter & not jitter itself. When talking about jitter that is correlated to data I believe we are very sensitive to it. Much like there are only a small number of people with absolute pitch but a large number of people who can easily identify modulations in pitch! My point being if we played the two slightly different pitches separately & independently most would not be able to tell the difference between them but put the same slight pitch differences in the same playback & most will notice the slight pitch change. In some ways this is analogous to jitter that is correlated with the data - such jitter varies with the music & does not produce a constant noise like that produced by random jitter.
Easier to reply without quoting you.
Yes, I have done some of these tests I quoted blind but without rigorous scientifically controlled DBTs there will always be doubt. However, I am happy that this is not placebo just as confident about it as I am about hearing the difference between vinyl playback & CD playback even though it is sighted. I guess the easiest way to test any of this is to try RF attenuators & do a blind test yourself. The RF attenuators, if reducing reflections is truly their mode of operation, will only effect the jitter numbers by something like the small values you see here.
EDIT: Just for others reading this - we are talking about hearing the consequences of jitter & not jitter itself. When talking about jitter that is correlated to data I believe we are very sensitive to it. Much like there are only a small number of people with absolute pitch but a large number of people who can easily identify modulations in pitch! My point being if we played the two slightly different pitches separately & independently most would not be able to tell the difference between them but put the same slight pitch differences in the same playback & most will notice the slight pitch change. In some ways this is analogous to jitter that is correlated with the data - such jitter varies with the music & does not produce a constant noise like that produced by random jitter.
Last edited:
---You two guys, John & Amir, are doing a fantastic job for our community.
We are going somewhere concrete. :b
We are going somewhere concrete. :b
Well, I don't know about that - Amir, based on the jitter numbers of this particular example, says that it's inconsequential a difference; I, based on my experience in this claim that it is audible.
Once some people try these minicircuits RF attenuators on the SPDIF cables of a reasonable system i.e. one on which you normally would be able to hear differences between SPDIF cables!
And report back on what they hear (blind) then maybe we will get somewhere concrete. At the moment it is just theoretical discussion, like so many other things!!
It wouldn't end there either, John. Let's imagine an even more robust test: Several high end and pro monitoring systems, experienced hifi and pro listeners among the participants, dozens of DBT trials in different locations, with different systems, in a well-designed study carried to statistically sound conclusions. Let's say it concludes that in none of the trials was anyone able to differentiate between the two more often than they would choose by pure chance. Would you stop believing your ears? I can't recall a single believer in hi-res changing his mind in the face of Meyer and Moran. I haven't seen anyone here change their mind about the correlation between measurement and choice in the face of the Harman study.
Tim
Tim,
Well it would end there for the people who heard the differences (blind) in their systems - doesn't matter what any studies say, does it?
Have you tried the RF attenuators yourself yet? Are you going to or has Amir's comments on the figures proven to you that it can't possibly be audible?
Well it would end there for the people who heard the differences (blind) in their systems - doesn't matter what any studies say, does it?
Have you tried the RF attenuators yourself yet? Are you going to or has Amir's comments on the figures proven to you that it can't possibly be audible?
Last edited:
Tim,
Well it would end there for the people who heard the differences (blind) in their systems - doesn't matter what any studies say, does it?
Have you tried the RF attenuators yourself yet? Are you going to or has Amir's comments on the figures proven to you that it can't possibly be audible?
Proven? No. As I often say, proof is a very big word. Delayed my interest? Yes. I haven't completely lost interest, but it's not a priority. I need a fret dress and set up on an electric guitar, a pair of floor monitors, some boom stands and a couple of good mics and an ADC... I'm afraid it would be impossible to conduct a valid blind test of an RF attenuator in my system, with me, anyway. I already believe there is no audible difference. I'm a well of negative expectation bias. And it would be very easy for me to hear my bias, given that someone would have to unplug the cable, remove the attenuator, re-plug the cable, etc.....you'd have a gap of at least a few seconds between samples. Not ideal at all. Testing me could only show that I could ID a difference. If I did not hear a difference, we would have very good reasons to question the results.
Half a test at best.
Tim
Let's be clear. I didn't say "it can't possibly be audible." I said that I can't go by a subjective sighted experience someone has as evidence of some measurement mattering. You probably don't know but I have done a ton of blind tests of such things and the differences I hear are very small when they exist. And it took a critical ear and knowledge of what content to use to get there. So when I hear the differences when measured in much narrower circumstances is readily audible and obvious, my antenna goes up, way up!Tim,
Well it would end there for the people who heard the differences (blind) in their systems - doesn't matter what any studies say, does it?
Have you tried the RF attenuators yourself yet? Are you going to or has Amir's comments on the figures proven to you that it can't possibly be audible?
I don't think science and objectivity supports a stance that says 10 picoseconds difference is audible. At least I don't know how to get there. The difference between 50 and 60 psec peak to peak jitter is -114 dB and -116 dB distortion products. If I constructed such an experience artificially, I don't think people will be able to hear that difference.
As far as I am concerned, once someone gets the jitter down to tens of picoseconds the job is done. For this, I mean what comes out of the DAC. For that reason, these digital measurements are not that informative since they are devoid of the jitter filtering that goes on after PLL filters out what it can. To be sure, PLLs in general aren't going to filter out low frequency jitter but without that measurement, we don't know where we stand. It is entirely possible in the scenarios that you mentioned where people heard a difference, what came out of the DAC would have shown little to no difference in jitter.
There are also other factors here which is RF and electrical coupling of devices.
Similar threads
