I've recently acquired a new pair of headphones (Sony SA5000) that have opened my eyes (and ears) on my music collection and i'm now enjoying listening to music almost like listening to it for the first time... Its probably placebo lead although these headphones are seemingly detailed and definately brighter in the top end than my old HD600s ... anyways.

I've started trying to maximise my enjoyment by listening to good old fashioned CDs through an old samsung dvd player (709, one of the first released) via my Tag Mclaren DAC20. The same DAC is used for my computer based listening, though sadly driven by a soundblaster live complete with its 48khz lock, audible or not i use foobar's SSRC resample to 48khz to hopefully minimise any problems.

The main reasons i'm listening the old fashioned way is

i) true 44.1khz playback
ii) complete silence in my room, nothing mixing in with low level details with my semi-open headphones
iii) remote control (something i had on my PC but haven't setup recently)

Its all about reason ii) really.. anyways i've started considering buying a more "hifi" transport (an audiolab or tag mclaren one perhaps to match) and am just wondering if its going to make any difference.. I know HA usually straddles the line that digitally you dont need to spend too much money to sound good, the rest should be down to placebo and deliberate colourings of the sound to appeal to a particular customer. I was just wondering if those that understand everything could pick this over and reason whether a transport can be important at all.. and if so whether that importance could result in an audible improvement.

To me the transport just has to get as good a data off a disk as it can... so long as the laser is ok and the disc is clean this should be ok.. I've been reading tho and apparently jitter can be a real problem with seperate dacs and all digital in general i guess... now how much of this is hifi bull or not is what i'd like to find out.

If this jitter is audible it makes me wonder why the digital specs didn't allow for DACs etc to have buffers in them to recieve 0.25second of the signal and then start playing it, only having to sync its own buffer playback to its own clock.. with it being realtime throughput i guess this could have an effect?!?

I know some people do digital out, to digital in and compare a recording made in this fashion to the original source.. if its identical its all well.. does this test suffer from jitter or is it not as vulnerable as its never taken out of the digital domain into analogue??

I swear that things sound nice when i'm listening the old fashioned way but scientifically the lowered noisefloor around me is going to have such a huge impact compared to any other changes.. at least that's what i'd expect, and i'm fully aware of placebo imposing itself on me..

If jitter is a problem i'm not sure how much the transport could help.. if the problem is cheaper transports not actually sending out the data with a good timing in the first place then obviously having a better transport outputing this more ideally could help... i'd wonder if the cable between the two could have some effect.. too late for me to find right now but i'm sure i read somewhere of digital transmission requiring some other capabilities not typically measured when used for analogue transmission.

Anyways enough rambling.. if anyone could help me discuss this a bit it'd help me know whether to not bother and save my money or possibly consider getting a better transport for reasons of obtaining that last 0.5% potential out of my setup.

You are spending too much time on this forum! I highly doubt that your perceived differences between the HD600s and the SA5000 are mainly placebo! HD580s/600s have a very definite character (to my taste, an unpleasant one).
Anyway, those headphones look nice.

Are you canadian?

Different DACs cope differently. Some reclock and so remove the effects of jitter; others have circuitry to reduce jitter (I don't know much about the technology). I don't know about yours.

Firstly spdif was designed with price in mind, rather than audiophiles.
Secondly DACs can do this anyway. Some do, but they are more expensive.

No, it should not. Jitter has an effect at the stage when digital is converted to analog. Digital - digital processing should not suffer from it, unless the jitter is absolutely enormous.


1. Check your DACs susceptibility to jitter.
2. If it is get a transport with a good clock (cables are secondary I think; but the spdif is said to be better than optical) if you want to stick with your DAC.
There are jitter reducers that you can put between transport and DAC. I don't know how effective they are.

There is an excellent paper presented at the 93rd AES convention in 1992 by Dunn and Hawksford titled "Is the SPDIF Digital Audio Interface Flawed?". I highly recommend you get a copy (Google for it) and read the paper - it will answer most of you questions much better than a forum post can. If you aren't technically inclined then you can limit your reading to Section 5 (Audibility of Jitter Errors).

However, be extremely careful when buying digital audio equipment. There is a disturbingly low level of correlation between "good" and "expensive". As I am sure you know, it's always a good policy to read audiophile and hifi magazines and websites with a finely tuned bs meter close to hand.

Let's start with some basic truths:

1. The SPDIF transmission protocol is vulnerable to jitter. What this means is that the timing of the digital pulses arriving at the DAC will be slightly "smeared". One of Creek's engineers once summed it up nicely to me: "the right sample playing at the wrong time is as much a problem as the wrong sample playing at the right time".

2. Good transports (which are not necessarily the expensive ones) will have lower jitter on their SPDIF outputs. And good interconnects (which are not necessarily the expesive ones) will protect the signal from additional noise-induced jitter.

3. Any half-decent DAC uses a long time-base PLL to reclock the incoming signal, in order to reduce the effect of the jitter. It is impossible to completely eliminate it, but it is possible to reduce it to inaudible levels. But not all DACs implement reclocking sufficiently well (and many of them are so-called "high end"), and therefore may sound different with different transports and/or interconnects. (Your suggestion of using buffering could work, and there have been a few cases of DACs with massive buffers so that they could be completely isolated from the transport's clock, but in practice the PLL method is good enough, so this kind of buffering is massive overkill).

4. I do not know if the Tag McLaren DAC20 is a good or bad DAC in the reclocking department.

So, you're thinking of buying a Tag McLaren transport. Go to your dealer and explain why you think it may improve your listening experience. He will wax lyrical about how much better the Tag transport will sound. So you can then explain that he'll be perfectly happy about bringing it to your house to allow you to do a blind comparison against your Samsung DVD player. If he refuses, walk away and tell him he's full of it. If he agrees, do the blind comparison and you'll have your answer.

Incidentally, if you find that your Tag DAC is susceptible to jitter, perhaps a better solution would be to replace the DAC rather than the transport. The Benchmark DAC1 is widely regarded as one of the finest DACs on the planet, and is know to be supremely jitter immune. It costs about $1000, which I'd guess may well be cheaper than a TAG transport.

Thanks for the replies guys..

I've managed to find a white paper on my DAC, it seems to make it sound like quite a good one.. perhaps you guys could see if you agree

http://www.iaguk.com/tma/dac20_tech1.htm

It also describes one of the transport options i've got.. in short it looks like it does do the PLL stuff you described



As such it should be decent i guess.. point being that the transport should be less of an issue than it could be for a DAC that doesn't take these precautions.. I'll go and read the entire thing, seems quite nice to read tho a good few bits go a bit over my head.

The DAC20 and its transport arent really current anymore, dont think so anyways.. hifi shops aren't really an option.. so only ebay and similar online auctions are options.



Heh, well it is a TOS somewhere... i spend time on headfi where things are much looser and hifi oriented (placebo is rarely discussed, its always about the equipment, wires, burn in or whatever), this question and asking it on HA is in the grey area between HA blind tests and hifi for me.. I'm still not happy to completely abandon spending some money on audio equipment, even if i haven't or can't blind test it.. but i am trying to be more strict on this as it tends to save me money.. hence asking this question..

There are big differences that I hear between the two headphones.. you are right that the HD600s can sometimes be described as rolled off or veiled.. an amp is supposed to help on this front (and getting a WNA amp really did appear to make them a bit brighter than they were before, with more power and details, again not blind tested against my old amp ).. all i'm saying is in the harsh bright of day on some music its not always night and day differences.. then with other music and in a more peaceful mood i will hear new sounds on recordings.. or rather the presentation will emphasise a particular sound more so.. I dunno.. one things for sure the extra brightness is palpable with these new cans as bad music played too loud is painful with its treble!

I've got some pics of those cans when i first got them

http://www.jimtreats.com/sa5000/small/

I really do like 'em looks-wise and sound-wise so far. Bass and treble sounds are really nice and the soundstage to me is different to the HD600s.. it does really good percussion... everything really tends to sound as good as or better than my HD600s, although being new i'm only really listening via my SA5000s so far.



Nope, though i guess my lazy online words are a mix of styles.. why is that a canadian thing, i'd thought it'd be a general american/canadian thing.. i dunno.. i have a habit of always using it

Just to clear it up alittle. Its not the spdif protocol that is vulnerable to jitter, but _only_ clock recovery from spdif that is vulnerable to signal modulated jitter. And that cannot be helped by any better transport.
You either reclock, do multistage PLL, or use external means to synchronise.
Jitter on the SPDIF link is irrelevant by itself. It causes no bit errors. SPDIF buffering is not needed, its buffered enough. The main and only issue is how to recover jitter-free master clock.

It looks like your dac20 is good enough to stop worrying about it or transport. You might get your .5% from looking into dedicated headphone amp.

But surely it's because the clock is embedded into the data stream that it is susceptible to jitter? And the clock can already have been slightly affected by the quality of the transport's transmitter circuit.


Surely jitter on the link *can* be relevant if the receiving circuit doesn't recover the clock well? (In the early days of two-box players, many DACs appeared which were quite badly affected by jitter. I would hope that this is no longer the case).

If you want a real and proven piece of hardware that makes an audible difference, I recommend buying a device such as a Behringer DEQ2496 or DCX2496 and use their powerful/precision E.Q.s to manipulate the sound subtely or significantly, to your preference. Don't jump the gun and compare this to the crude graphic E.Q.s that are coommon. These devices I suggest can be used with zero audible noise or distortion addition to the signal(when used properly). The DEQ is a strict E.Q. device, where as the DCX is a DSP crossover with E.Q. functions. The DCX has an advantage, though, of having a computer interfance and software GUI to make very easy/convenient WYSIWYG modifications to the frequency response.

-Chris

???
Eq is a quite separate matter. Unless those boxes have good jitter specs at their outputs - quite possible.

The grand opening! Thanks for sharing!

Off topic, do you know much about equalization WmAx? I want to do software equalization from PC. The foobar equalizer isn't very fine (not many frequency points, and 1db intervals). Is there another equalizer or does one have to use convolution?

Sorry, I don't know of specific software E.Q.s for the PC for normal playback purpose(s). When I require adjustment of a specific file, I tpyically use Cool Edit Pro/Adobe Audition. When I require specific adjustments/manipulations for standard playback or acoustical testing/experiments, I will occasionally connect a DCX2496 to the computer.

-Chris

OK, thanks anyway!

cliveb, this is all nice and dandy, but you must not loose from sight the core issue. And its only the matter of jitter-free clock during DA convesion. Fact that cheap stuff does a subpar job in recovering the clock directly from spdif is not a flaw of protocol, but flaw of that stuff.
Spdif is purely bit transport protocol. No part of this protocol *requires* to use clock recovered directly from the spdif stream. To decode the stream, you only need to have *synchronized* clocks. How you do that, is up to you. Within a device, you simply route master clock by independant wires. It became simply such a convenience to use just 2 wires to connect different equipment by spdif. You could instead have had 1 device that generated master clock and distribute it via special wires, and ignore any jitter that gets added to spdif stream. You could have independent atomic clocks in every device - sufficient synchronization.

There are ways to synchronise master clock with spdif stream, and there are ways to get away without synchronisation. It makes it more difficult than if spdif had separate clock wires, but separate clock wires have their own bunch of jitter problems, and infact it might be fortunate that they didn't take that path. We'd be now discussing how stupid they were and how little it solved the jitter problem and be stuck with just more useless wires. In any case, spdif does perfectly what it was designed for. Technically, it doesn't even have clock in it.

Better transport gives very little simply because spdif stream jitter occurs *after* the transport, so you still have to deal with the core of the problem - get the master clock right without *depending* on the spdif stream.

Ideal solution would be to have jitter-free master clock right next to DACs, and send a copy of it to the transport. It can have jitter as hell there, and spdif data from transport even worse. Simply because data is synchronized with such master clock, all that jitter is simply cut off in the DAC then.
Anything can be relevant when receiving end isn't doing well. Power supply ripple due to inadequate filtering - I believe you wouldn't call AC power supply protocol flawed because of that?
Of course that jitter still happens, because its cheap. But I bet that you can't think of anything better than spdif that woudn't introduce its own list of problems. Its *extremely* difficult to distribute master clock without introducing jitter. You simply have to find ways to tolerate that.

Have you read the Dunn and Hawksford paper I referred to above? One of their conclusions is that SPDIF is flawed precisely because it is difficult to accurately recover the clock from the stream.

If THAT is their conclusion then with all due respect to Hawksford they are pulling bullshit. But more likely, you misread something.

There is not a single place in AES/EBU spdif specification that tells you HOW to exactly recover the clock. It says that biphase mark coding makes it *easier* to recover the bit clock, but it and the sync bits are intended for a single purpose - to *synchronize* receiver and transmitter so that *data* could be recovered, *not* to use some crappy and noisy PLL as a master clock in DA conversion.
If you think that purpose of that paper was to prove that you cannot recover jitterfree clock from NRZ passed through a noisy medium, then guess what - that was *well* known some 50 years earlier. Every single engineer who has ever dealt with precise timing knows what the jitter is. And so did those who designed the spdif protocol.

What Hawksford *may* have tried was to bring the issue of audibility of jitter to the focus of a bunch that in '92 were stuck in blind belief that as soon as you have your audio in bits that nirvana is guaranteed, discarding DA conversion issues too.

I am wrapping up, I don't want to argue about this. I just pointed for you the important points that matter so you can look for right things.
spdif defines how to transmit data, not how to transmit jitterfree clock.
if you go for picosecs of jitter, it is *impossible* to do with any means but by *generating* jitterfree clock right next to the DA conversion unit, no matter is it spdif, jesus or whatnot.
PLL clock recovery is meant for errorfree *data* recovery from the serial transmission - synchronization.
You do not *have to* push a noisy PLL down the throat of a DAC. But you can, and if you do, you get what you pay for. Which, of course, is what its all about.
There is nothing to change in serial digital audio transmission to make it less "flawed" than spdif is. Jitter, like shit, just happens. You have to deal with it, anyway.

We appear to be in violent agreement. What I was trying to point out is that it is possible to recover a clock that's good enough for human listening from the SPDIF signal, but that there are some (many?) DACs out there that don't. Sure, this is a flaw in implementation, but it's an implementation flaw that is encouraged by the SPDIF protocol, because it makes it easy to recover a clock that turns out to be jittery.

As you have pointed out, the correct place for a low-jitter clock is in the DAC. And indeed a number of enterprising companies (Arcam and Linn are two that spring to mind) did indeed implement this, with a feed back to the transport so that the transport's clock was slaved from the DAC's. But guess what? They didn't *sound* any better. Why? Because other good quality DACs were bothering to reclock properly, and were already achieving jitter levels beneath the threshold of audibility (without any need for proprietory hardware add-ons).


Where else is the DAC supposed to get its clock from? Given that the overwhelming majority of transports don't have any means of accepting an external clock, the DAC can't have its own free-running clock, because it won't stay synchronised with the transport.

I know exactly what you feel. But nevertheless, it is not eventually right to think so. Spdif cannot be held responsible for misuse of its clock, simply because it is impossible to have jitterfree serial data transfer. If you read the Hawksford paper you'd get the gist of it. Bits that vary in combinations create a jitter. Your only option is to use clock that is separate from the data. In spdif this is implicitly assumed. It makes little sense to send same clock with each spdif link when you need only one or zero clock lines for any number of spdif lines. Therefore issues of DA jitter are left intentionally at the discretion of equipment designers. It is completely irrelevant for transfer between purely digital equipment. It matters only in the final destination - DA conversion.

Still, if you think about it alittle, you have to agree that spdif was one of the best things that happened to digital audio. If it were made more sophisticated and cluttered, it'd never take off into consumer market, or would cost needlessly much. And most probably DA jitter issues would have been still present, only much more obfuscated by "the better" protocol encouraging using its "better" clock. In other words, it would be only worse.

In fact, initial spdif protocol allowed such a huge jitter tolerance that it was like shouting "don't even think to *use* me in DA conversion for chrissake". It was later updated because equipment from different vendors were actually getting bit errors due to stupid implementations!
This is arguable. Reclocking (actually resampling) is straightforward way to get rid of any spdif jitter, but it requires digital filtering. This is not a simple act when you have to reclock by 0.0x% the clock differences. In reclocking approach there are different issues. *Sound* of the DAC is not defined only by its jitter.

In academic sense, transport slaved to the DAC is the only right way. But it is so much inconvenient. And right here you start to realise that spdif is sufficient and elegant as it is.

I'm surprised that you ask this, after in previous paragraph showed yourself one of the options.

There is *sea* of possibilities to get line-jitter independant clock that is synchronised with the transport. I think most have been already mentioned. Even from spdif with PLL you can recover clock in at least dozen different ways. With pll, its best to make clock corrections as rarely as possible, not at every bit. There are special long synchronization patterns in the stream every 192 samples. Its possible to make PLL that is separated from the line jitter and have lock-in and tracking time constant in seconds, as seems to be the case with the mentioned DAC20. You can buffer few thousand bits and digitally initiate slight corrections when the buffer is departing from exactly 50% fill few times a minute, having complete control over how much correlation and spectrum such added jitter has. This has actually been done in a consumer chip. This works well, and only requires low-jitter master clock that can be gently steered towards right timing. And very low jitter clocks can cost a little fortune.

It would be strange if Sony engineering staff (who have quite a reputation at headphone development at reasonable pricing) couldn't produce a better pair of headphones for more than twice the price of the HD600's (which I do own and like very much) ...

My apologies for missing the fact that you were talking about recovering the clock *directly* from the SPDIF, which you have now helpfully placed in bold, so that even I (with my -10 eyesight) can't miss it Of course one cannot expect to use such a clock.

I was under the misapprehension that you were suggesting that a DAC should completely ignore the SPDIF stream for clock purposes. When using a transport which cannot accept an external clock, then short of buffering a *vast* number of samples to cope with any conceivable clock rate differences with the transport (which would in any case cause undesirable latency), ultimately a DAC does have to derive its long-term clock rate from the SPDIF stream. Again we appear to be in agreement, while managing to misinterpret what each other is saying.

Well my DAC20 seems to be a reasonable DAC, it has some things in place to work around jitter, and has reviewed reasonably.. didn't stop me from getting the audio bug for a bit there..

The afore mentioned Benchmark DAC1 seems to be world class with its jitter tolerance.. it can handle the signal going through 1000feet of cable.. impressive!! I've seen some people talk about loving it and a few others being a little underwhelmed by it (hifi-wise here not AB tested) ..

Someone mentioned Dodson DACs as being superior to DAC1's .. I looked around and found this

http://www.dodsonaudio.com/

http://www.positive-feedback.com/Issue18/dodson_263.htm







The audiophile in me (yes there is still a little bit in there somewhere ) would love to hear this DAC and its big brother the DA-218 (a bargain at $8000) but it really does dissapoint me that its supposedly dependant on transport somehow.. aside from read errors which shouldn't be significant on decent condition discs the transport can only contribute to jitter, and so i guess its saying that its just not jitter tolerant at all.

Benchmark DAC1 is resampling. Thats a reasonable thing to do when you oversample anyway, and there are practical reasons why this is right thing to do in any case (eg. you have several independant inputs to select from - you can't have them all in sync and spend time on slow PLL clock recovery). Benchmark DAC1 is considered to be one of the best dacs in the world. Its made for professional no-bullshit types.
Right thinking. I find it ironic when $8000 DAC (what the hell costs there so much anyway?) is missing fundamental elements to make it insensitive to jitter, and, umm, power cords! Snake oil alarms are ringing in me at levels approaching hearing damage..

Ahh. There is nothing else to talk about...

This is actually funny.
http://www.dodsonaudio.com/UltimateAudio_DA218_Review.htm

This results either total immunity to transport and digital cable induced jitter, or this is blatant lie. 2ps jitter clock is serious.

And that thing is sensitive to power cords?

ROFL.

Yeah, right. bybee quantum purifies needed break in...

Glad you liked it..

I'd still like to hear one but i aint about to spend that much money on something Its nice seeing your angle on things as my tech knowledge in this area is fueled more by hifi-blurb than a rigurous understanding of the fundamentals.

The Cryogenic treatment made me smile.. I'd like to see the measured improvements it brings.

That review you found is saying that the DAC sounded the same as using the transport/player on its own.. ie. $8000 DAC in or out made little audible difference to the sound to those testing it.. and yet this is linked on their site?!?