how can digital be different

Originally posted by wadia-miester
Pete, more than a smattering of truth in that statement there sir :)
Click to expand...

Hey, there's nothing wrong with a bit of blagging your way through things you don't really understand - some of us even make our livings by it :D :p Seriously though, julian has already been good enough in this thread to correct my misconception that CD players were unable to interpolate through missing data packets - if there's anything I've said that's erroneous, please do enlighten me. :)
 
Pete, you cannot 'add or derive' what isn't there in the first place, the 'Guess' if you like is only as accurate is the incoming information, so shite in shite out.
Not the point I was trying to make though

"And in any case, it seems to me that the interpolation is more likely to be performed in the analogue domain as by its nature you can't interpolate digital information, in which case it would be a job for the DAC and again have nothing to do with the transport"

why do you reason it has to be done in the anaolgue section Pete?

could you also explain how upsamping/SRC transports do this Interpolation with out the aid of a dac ?, as I'm bit lost this evening cheers Wm
 
Actually, thinking about it a bit further I'm talking through my proverbial orifice WRT the analogue-domain interpolation business, or at least typing through it (which sounds painful), although I'd need to know a bit more about exactly how the information is coded digitally in terms of the relationship between the numerical value of the bytes and the waveform they encode to be able to have a discussion about it. I stand corrected :) And I have no idea how upsampling works, that's something else I need to find out about.

Anyway, I'd like someone to follow up the rest of the stuff I said in my last post, as I at least had a vague idea that the rest of it was grounded in reality :) Julian, my understanding from what I've read today is that the 'net' error-rate, in terms of actual audio information lost, is zero in the vast majority of practical situations, and hence it follows that a transport's ability to cope with frame errors cannot contribute to its perceived sound quality. Is it your experience that this is not the case?
 
Last edited by a moderator:
Pete, I would also say there are upsampling dac's too M/F, MSB etc. but the over/upsampling is done in the digital domain, after the filter, before than the dac section.
Transport errors, now there's a pandora's box of fun :)
 
Originally posted by wadia-miester
Pete, more than a smattering of truth in that statement there sir :)
Click to expand...
Well, you'd know all about that!

Let's not get sidetracked by upsampling for the moment.

Interpolation can be done either in analogue domain - where it's called a low-pass filter, although implementing such is not really compatible with frequency response up to 20 kHz - or digital. The advantage of doing it in digital domain is that you can use the error-detect signal to invoke it only for the few points for which it's necessary - rather more problematic to do in the analog domain.

Does Julian yet accept that his re-read theory concern is irrelevant to the typical sound of a drive playing a typcial CD? I made a statement earlier that it would need 10s to 100s of errors to get past the CIRC to affect the sound produced by the replay system. Does anyone want to challenge this? If not, this would require read-errors of approximately 1 every 1-10kB of data read from the disc. Has Julian ever seen anything this bad? PeteH has cited evidence that CDs typically play with Zero such errors. In the absence of any counter-evidence can we discount this theory?

If WM actually has any relevant information on transport errors, then it would seem appropriate to give it here, rather than make oblique gnomic statements.
 
graham,
as i have little evidence other than my experience that re-reading does occur quite frequently in 'data' circumstances. also i would say that if this were a freak occurrence that hardly ever happened then the interpolation hardware wouldn't be added to cd players for cost reasons - especially early ones.
i still think that errors and interpolation play a part in the sound of a transport, especially as most people who try different transports say that there is a difference in their sound.

http://www.discdupe.org/i/evaluations.htm
the above site has eveluated a number of cd-r disks although this is not what we are discussing and they only look for a certain type of error it does show that E32 errors (unrecoverable ones) CAN occur. for a definition of the test and errors see
http://www.discdupe.org/i/bler.htm

cheers


julian
 
Read the articles on this site. I know that Altmann is trying to plug his own jitter reduction devices to some extent but what he writes about the subject is excellent.

Particularly relevant is the following paragraph in the "How does jitter sound?" section:

In order to be sure that we hear jitter effects and not reading errors or interpolations from the CD player we had to monitor the correction flag (cflg) output of the CD-player's decoder chip SAA7376 (go to the datasheet.pdf).

This was accomplished by an external microcontroller that was programmed to increment the number on a 4 digit display on every interpolation or hold. With this tool we had the accurate count of all uncorrected samples that were output during the track or the entire CD.

The funny thing is, that the low cost CD723 player is able to read any CD or CD-R that is not severely scratched without any interpolation or hold. You can even put it upside down, there will be zero interpolation /hold.

Uncorrected samples could only be generated by dropping the player (5cm) or by extremely scratched CDs.

This gave us the confidence that our experiment setup is valid, and we knew that the jitter phenomenon really deserves our full attention.
Click to expand...

So, the $99 Philips CD723 they used for their jitter experiments was able to read even mildly scratched CDs absolutely flawlessly, without interpolation or hold

Altmann contends (and I agree) that the answer to all "how can digital be different" questions (transport, cables etc) are all down to jitter due to various causes.

Michael.
 
Thanks for that Julian.

However I think it backs up my argument rather than yours. The total E32 errors for all (except a couple of obviously crap) brandss ZERO! I also don't seem to be able to get you to say what "quite frequently" means - I would consider 1 error per 10MB pretty horrendous for a data disc, but I contend that an audible effect (other than discrete clicks) requires rates 1000 times this.

Again - the point is not (and never has been) whether discs are read 100% perfectly in all possible circumstances, but whether bit errors routinely contribute to the sound of a transport. This report says an unequivocal NO.
 
graham,
i still say that if there were as few read problems as you say there would be no need for interpolation and definately no need for meridians re-read system as used on their players. i guess the impact of the errors is a lot less than i initially thought however i'm still convinced that it can impact the sound of a transport.
cheers


julian
 
Originally posted by GrahamN

If WM actually has any relevant information on transport errors, then it would seem appropriate to give it here, rather than make oblique gnomic statements.
Click to expand...

Yes I do, why should I? & what-ever.
There are only 5 maybe 6 people who post here that could have that conversation accurately, and the other 30 or so to de-reason it :rolleyes: as you said graham, it's terrible being unitelligent, for which I make no appologies :)
 
Originally posted by julian2002
graham,
i i guess the impact of the errors is a lot less than i initially thought however i'm still convinced that it can impact the sound of a transport.
Click to expand...

Based on the evidence above I really don't see how you can possibly still hold that position. In the absolute worst-case example in the link you posted there are 30 uncorrectable errors, which AFAICT would correspond to something in the region of 600 microseconds of interpolation being necessary if the CD is to continue playing. I'd be extremely surprised if the most keen-eared observer could even pick up a total of 0.6ms of interpolation spread across the entire test length. And to reiterate, this is the worst-case scenario they could find with a cheap computer transport reading the dodgiest-recorded CD-Rs in the test - in all other cases, there were no net errors at all. So how on earth can error compensation impact the everyday sound of a transport when - as all of the data we've dug up between us demonstrates conclusively - there are no errors to compensate for?

And wadia-miester, if you have any relevant data that sheds new light on the subject, I'd presume it'll stand up to investigation so I don't see why you're not willing to share it. And let me be the first to apologise on behalf of the Zerogain forum members for not measuring up to the standards of intelligence you require before you'll enter into a semi-technical discussion with us. Raises the question of why you bother with a hifi forum at all though...
 
Originally posted by PeteH

And wadia-miester, if you have any relevant data that sheds new light on the subject, I'd presume it'll stand up to investigation so I don't see why you're not willing to share it. And let me be the first to apologise on behalf of the Zerogain forum members for not measuring up to the standards of intelligence you require before you'll enter into a semi-technical discussion with us. Raises the question of why you bother with a hifi forum at all though...
Click to expand...

Pete, I take it sarcasm and ironry have been lost on you ?, why should you appologise ?, the fact you decided to do this well quite entertaining if nothing else, but don't let me stop you from carrying on. (Joke Wm humour alert)
The intelligence jibe was a reference to Graham (N), who the other day felt, that IQ had dropped sharply on another thread.
The people than can discuss this subject not only with accuracy and genuine knowledge, don't tend to get too deeply involved with these interplays, of the 5 or 6 people I feel could have more than a bearing on this I don't count myself with enough total knowledge on this area to be classified as one.
As for extra data on this subject, thats' our little niche' as I've said before Coke don't give out there recipe', so please don't be offended if we don't go too much into detail.
Zerogain is a wide and diverse place, with many ideas and passions, and strength of debate, however humour plays a part too, unlike audiolab kit, it's not too dry, so Pete don't take it to heart.
Why not take the bull by the horns, construct some relevent modules with the various different theories, and preform some 'detailed testing fully recorded' and present it to the forum, Stebbo & Paul R maybe able to help here,
Prehaps mike could put this in FAQ's, when it's finished, so you can go back and refer to it.
I'm off back to my humble cd transport for a listen. Wm
 
Last edited by a moderator:
Originally posted by PeteH
And wadia-miester, if you have any relevant data that sheds new light on the subject, I'd presume it'll stand up to investigation so I don't see why you're not willing to share it. And let me be the first to apologise on behalf of the Zerogain forum members for not measuring up to the standards of intelligence you require before you'll enter into a semi-technical discussion with us.
Click to expand...

Perhaps he knows the Reed-Solomon is a BLOCK error code (think about what that means for interpolation).

Also, he also knows that the data is reformatted by the SPDIF output transmitter for transmission between the transport and the DAC. [Recall, the original posting was about transports and cables..] Now that's digital manipulation and transmission, so why am I mentioning it? :)

Based on the evidence above I really don't see how you can possibly still hold that position. In the absolute worst-case example in the link you posted there are 30 uncorrectable errors, which AFAICT would correspond to something in the region of 600 microseconds of interpolation being necessary if the CD is to continue playing.
Click to expand...

When it comes to data correction, the average error rate is not the whole story. In fact the variance of the error rate is significant in BLOCK codes.
 
Qutie possible I'm missing something here dat19, but I don't quite get what you're on about. The papers I, pete, michael and julian pulled up showed NO errors per disc getting through the CIRC other than in pathological cases - so whether average/variances or block/bit seems irrelevant. Care to elaborate? And yes, I'm fully aware that the SPDIF transmitter does a load of reformatting and the clock there is the crucial component - but I wasn't trying to complicate matters there for the moment.

For the record, I too believe, and have many times said here, in this thread, and elsewhere (but am quite prepared to be proved wrong) that any differences in sound due to transport and cables are due to susceptibility to jitter (and the arguments have been about making jitter-insensitive receivers or communications schemes at the DAC).
 
Originally posted by GrahamN
For the record, I too believe, and have many times said here, in this thread, and elsewhere (but am quite prepared to be proved wrong) that any differences in sound due to transport and cables are due to susceptibility to jitter (and the arguments have been about making jitter-insensitive receivers or communications schemes at the DAC).
Click to expand...

Does this validate the case more for a one box :eek: , or is a one box MORE susceptable to digital signal corruptional influencies ?
IME, Jitter is caused mainly by contamination, by internal power supplies, from ether main board or the rails that feed the the clock/Filter/dac chips, in adquate coupling/decoupling of said chips, rfi by emptied over the unit from clocks & dsp arrays, has anyone consided that a clock outputting a square wave form will generate more RFI than a sine wave variant?, cables and reflections big issue here.
However I will, the best sound I've ever heard from a red book/sacd is a 2 box unit, transport/dac, at least with two units, the power suppiles for the sevro drive and transport mech/clock & digital output boards are seperate from the dac section, yes the transmission of the signal is susecptable to 'attack' however, I personal feel this isn't such a critial 'area' as shoving it in one box (says he with a one box :D )
Just some musings Wm
 
Originally posted by dat19
Perhaps he knows the Reed-Solomon is a BLOCK error code (think about what that means for interpolation).
Click to expand...

I'm all ears :)

Originally posted by dat19
When it comes to data correction, the average error rate is not the whole story. In fact the variance of the error rate is significant in BLOCK codes.
Click to expand...

True but absolutely irrelevant in the context of a discussion wherein we've established that there are, in total, zero errors in recovering data. Noone ever said the average error rate is the whole story, but as the peak error rate under practically all circumstances is insufficient to cause any failure of the error correction I don't really see what point you're trying to make.

Wm, your post looked like an attack on the general level of intelligence of the ZG forum members, which I found a little insulting as I'm sure you'd imagine :) If it's an in-joke though clearly that's different.

And in case anyone didn't bother to read my earlier posts I'm not trying to say that all transports are the same, rather that I believe any differences between transports are due to jitter issues and categorically not bit error considerations.
 
Originally posted by PeteH
And in case anyone didn't bother to read my earlier posts I'm not trying to say that all transports are the same, rather that I believe any differences between transports are due to jitter issues and categorically not bit error considerations.
Click to expand...

Has anyone done a analysis in the area of Analogue V's digital sevro drive control and it's possible effects on 'Read only errors' not 'In transit' errors ?
Slight diversing here, sorry it's wadia again, by changing the transport mechainsm & software drive control on a 270 transport, (remember the older 3.2 mech is no light weight either), the difference in sound (precieved ;) ) reproduction, isn't small, now both units are digital sevro control, same psu's, damping & weight are greater on the P0 mech, now is this due you feel due to 'no possiblitity of read only errors' being eliminmated :confused: ' or jitter reduced ? or less microphony inside the unit, possibly having an effect?, or possiblity of less feed back through the psu's?
or simply a mechanical effect, we have done some reasearch on this, so I'm interested in some thoughts here
 
Originally posted by wadia-miester
now is this due you feel due to 'no possiblitity of read only errors' being eliminmated :confused: ' or jitter reduced ? or less microphony inside the unit, possibly having an effect?, or possiblity of less feed back through the psu's?
or simply a mechanical effect, we have done some reasearch on this, so I'm interested in some thoughts here
Click to expand...

Playing nicely now are we Tony? ;)

Simple test to do - the one I suggested (albeit with a slightly different objective) about 3 pages back.

I guess there's a simple experiment: connect the digi-out from a CDP to a PC input, dump a track to disk several times and compare. Obviously you'd have to take a bit of care to synchronise the bytestream for the comparison to work, but that's not difficult (I did something similar when trying to see any difference between a production CD and CD-ROM: of course no differences seen). If there are differences, but they only appear at the level of the LSB or thereabouts it would seem to me that there's some DAC-ADC process going on inside the capture card, and that's just acoustic noise getting into the measurement through the sound-card.
Click to expand...
Use your most accurate cable, minimise any mechanical motion of that (in case the physical geometry affects reflections etc) and make repeated measurements. Do this for both transports - look for differences between successive measurements on the same transport and for differences between measurements taken on either transport.
So the possible results are:
a) no differences between any of the measurements on either transport - any sonic differences definitely not bit error related
b) differences seen, but in LSB (or thereabouts) only. Either your capture card has some analogue stage, and noise is getting in, or there are read errors and there's some fairly decent interpolation going on. Significant differences in the difference rates for each transport (if you see what I mean) would indicate changes between the two transports in the handling of errors.
c) differences occur up into higher order bits, but fairly isolated. Probably no interpolation going on, but read-errors occurring. It's not clear to me ATM what does happen when the drive encounters an E32 error - does it blow out the whole block, or just say "OK - I got it mostly right, here it is and BTW there's a couple of bad bytes in there somewhere".
d) data actually rightish, but some extra datapoints appear from time to time causing misalignment of the data -> spikes getting into the cable/capture card causing intermittent loss of sync
e) complete blowout - your capture card is probably not locking onto the transport output. Need to work a bit harder on the experimental technique

I would expect a) - and then you're back to looking at timing errors, and a high frequency sampling scope (possibly with FFT analyser for jitter spectrum).
 
Graham I' don't play nicely :) I cheat fairly :D

You should know by now................

"Use your most accurate cable", they make no differerce what so ever, so how can this help :D
Timpy wrote a software package/(we built the rig) for testing fibre optic cables for the telecomunications industry, we've adapted it some what for our own devious purposes quite useful on occations, so we're not just totaly subjective.
But so far in this thread a the conclution seems to be that read bit errors are almost improbable?
Me I've a subjectively open mind, that some times needs a blue curtain to'help me see things more clearly'
I would err with you on prodiction 'a' initialy, for the 'stock' mech is still one the best available, however that still doesn't mean it's not falable, one thing though, all the sony mechs we have, are more susceptable to picking up disc imperfections, where are the teac isn't bothered in the slightest, never skips or hear scratches, yetr the same discs on arcam/cdfix/rotel are 'well in the mix' :rolleyes:
We shall see in the fulliness of time (well after the show) T.
 

Latest posts

Back
Top