Getting the best out of a Rega DAC (or any Dac)?

[treb]

Well, the suggestions on this thread have been helpful. The Saturn going into my Rega DAC still sounds supreme and I have limited options with alternatives. I wanted to avoid using a CD player as a transport for a host of reasons (I prefer listening to entire albums but haven't found the time lately for long listening sessions, plus my headphone system is on my desk, away from my amp and speakers, which the Saturn feeds).

I've gone back and given toslink another go. The last time I tried it was when the Rega DAC first came out and I still had my old MacBook Pro. I really didn't give it a fair chance. But with my MacBook Air, I don't have optical out. So, I tried going toslink out of my Apple TV 3 through AirPlay (using a silflex 3' cable).

I have to say that the ATV3 to the Rega Dac sounded a whole lot better than my MBA and USB/Vlink. It at least didn't sound 'flat' anymore and sounded more neutral. I compared the toslink into the DAC with coax from my Saturn (switching by using the input selector on the DAC) and the sound was more comparable between the two than between the Saturn transport and MBA/USB/Vlink. In fact, it was a bit hard to distinguish between the two. To me, the toslink in from the ATV sounded a touch less 'edgy', but that was all (in fact, I came to prefer this sound).

To stream via AirPlay, I used iTunes and Audirvana Plus. I preferred A+ (imaging was more solid) but was getting occasional dropouts (no such dropouts on iTunes - may try Airfoil with A+). Files were ripped in .aiff.

This is what I heard on my system. I'm open to trying other things that were suggested like an SBT or a Mac Mini (or even something like a QA-550 SD card transport, which seems interesting).

Treb

 

[treb]

I forgot to say, thanks to all for the suggestions and the lively debate thus far. I do agree and see 'these forums as a way of getting information about things people have tried' but do appreciate a healthy form of scepticism to inform observations/conclusions, when those observations occur.

 

[Daniel.]

Fans make (acoustic) noise. If you have a pc in the room with your audio then it should be quiet.
Of course years and years and years ago the people who came up with the squeezebox thought that it was obvious that since pcs make elctrical and acoustic noise, the senisble thing was to keep them in a different room and connect via ethernet or wireless.

I can't see why one would want to use a pc in the same room as the hifi, but it greatly tickled me when people started making foo computers with fancy psus and SSD's to solve this problem. It's IMHO about as smart as keeping your fire on throughout summer and getting an airconditioner to keep the room at a nice temperature.

We were talking about audio output not acoustics, everything you said here was irrelevant so please read the posts properly.

 

[Daniel.]

As far as jitter it is a very real part of audio and is focused on a lot by manufacturers because it does affect audio fidelity.

Everything else i have agreed with you on adam

 

[Paul R]

As far as jitter it is a very real part of audio and is focused on a lot by manufacturers because it does affect audio fidelity.

At what level? Show your working.

Paul

 

[Daniel.]

At what level? Show your working.

Paul

http://m.youtube.com/watch?v=JYjHKv2_OqQ&desktop_uri=/watch?v=JYjHKv2_OqQ

And perhaps i could refer you to a textbook by john watkinson, The art of digital audio. He speaks a bit about the importance of jitter especially with oversampling dacs. Not so much a issue with nos.

 

[Daniel.]

At what level is what we are sort of talking about, whether jitter correction systems like phase lock loop or asrc remove all jitter. Without the error correction systems sound would definitely be affected, its whether source errors like that from transports or medium type can be compensated for.

 

[Paul R]

And perhaps i could refer you to a textbook by john watkinson, The art of digital audio. He speaks a bit about the importance of jitter especially with oversampling dacs. Not so much a issue with nos.

I have that book, I bought it at IBC in 1990.

But it doesn't help you with the relevance of your assertion about jitter and audio fidelity.

Paul

 

[Daniel.]

I have that book, I bought it at IBC in 1990.

But it doesn't help you with the relevance of your assertion about jitter and audio fidelity.

Paul

I have 2000, not sure 1990 would have the same context. None the less it does and if you cant grasp the concept of jitter then you must be missing something. The youtube link should have at least open some ideas.

 

[adamdea]

As far as jitter it is a very real part of audio and is focused on a lot by manufacturers because it does affect audio fidelity.

Everything else i have agreed with you on adam

it can affect fidelity, but as for whether it really explains what people hear, I'm not so sure.

1) When considering jitter you have to bear in mind that only jitter at the conversion stage actually matters. Interface jitter is irrelevant if the conversion clock is able to work independently of it.

Dplls and ASRC can give you huge amounts of jitter rejection above a very low corner frequency. Look at the j test figures for most modern dacs on s/pdif transmission and they are vanishingly low (try looking at an mf v DAC ii for example - which is or was very modestly priced.) In fact if you look at the stereophile measurements you'll find that many dacs measure just as well spdif as on asych USB.

The real problem with spdif is not interface or transmission jitter, ie hf fluctuatiions in the timing of each bit. It's long term rate differences between the transmission of the sending clock and the conversion clock. The hf stuff just evens out once you have a small buffer and can be filtered out by dpll. The difference in clock rate is a problem though as the buffer will eventually fill up or run out unless you either
A. Match the conversion clock to the incoming stream. This might mean using a less stable clock. Maybe.
B use ASRC to encode the low frequency drift as a change in data value. This like very low frequency jitter, but this should not be audible because of masking effects.
C have a big buffer. Provided you are just steaming audio and aren't worried about latency this works just fine. Frankly I don't know why's most dacs don't use this - chord have for years.

Maybe all if this is a problem but try finding some proper analysis of jitter on two transports spdif out being traced to audible spuriae. By contrast benchmark have published figures showing that the jitter on the dac's output after hugely long spdif cables (can't remember-100m?) is still negligible

2) how much jitter at the conversion stage is required to be audible? Well you'd be surprised how little evidence there is on this. If its broadband white noise type jitter, probably loads and loads because it will just raise the quantisation noise floor, and you could raise that by quite a lot without bothering anyone at normal listening levels. There is nice paper by Julian Dunn estimating audibility of different frequencies of sinusoidal jitter at high listening levels taking into account the audibility threshold at different frequencies. Iirc he came up with 75 ps. This was at frequencies though at which jitter attenuation should be easy, and ignored the masking effect of program material.

It may be that you can measure a tiny tiny tiny fluctuation in the conversion clock when (eg) the pll has to work a bit harder, but is it audible?

3) the argument that audio fidelity is down to jitter is plausible, but there's loads of speculation and guesswork involved. Once it's below 200 ps or so I'm not sure that any real evidence of audibility.

4) when people say that know what jitter sounds like, take it with a bit of a pinch of salt. In very very very many cases they mean that they think they can hear a difference and they think this might be caused by jitter.

 

[Daniel.]

Id agree with most of that. Jitter is still a very real thing its just hard to know when the error is low enough to be inaudible.

 

[adamdea]

I compared the toslink into the DAC with coax from my Saturn (switching by using the input selector on the DAC) and the sound was more comparable between the two than between the Saturn transport and MBA/USB/Vlink. In fact, it was a bit hard to distinguish between the two. To me, the toslink in from the ATV sounded a touch less 'edgy', but that was all (in fact, I came to prefer this sound).

To stream via AirPlay, I used iTunes and Audirvana Plus. I preferred A+ (imaging was more solid) but was getting occasional dropouts (no such dropouts on iTunes - may try Airfoil with A+). Files were ripped in .aiff.

This is what I heard on my system. I'm open to trying other things that were suggested like an SBT or a Mac Mini (or even something like a QA-550 SD card transport, which seems interesting).

Treb

I'm glad you found a solution you enjoyed. If you want to experiment, do try an sbt. I love mine- it's my favourite gadget of all time. The only reason I hesitate to recommend it is because it's out of production. If you don't mind taking a punt on something which might not be supported for long then try it. It sounds as good into any DAC I've tried as any other transport (cdp, pc, laptop) I've tried.

More to the point it works brilliantly with an iPhone or iPad any allows you to keep your data and server anywhere in the house you like

 

[Paul R]

Id agree with most of that. Jitter is still a very real thing its just hard to know when the error is low enough to be inaudible.

You wrote on the previous page

As far as jitter it is a very real part of audio and is focused on a lot by manufacturers because it does affect audio fidelity.

You seem a little inconsistent.

Jitter is focussed on by some manufacturers and many equipment enthusiasts because it's a word that sounds like it must be important, because you can imagine how jitter will affect the sound, and because it can be measured to give a rather meaningless number, so you can compare goodness.

But nobody knows how much is audible, other than that when formal experiments were done the level required to be heard was so great that equipment producing that much by accident would have to be faulty.

Paul

 

[Daniel.]

I see no inconsistency.
jitter is focused on still with equipment design, its whether how much error correction is needed before it stops making a difference. You seem to confuse in what i think jitter makes a difference with. If timing correction is not needed then lets not use pll or any kind of reclocking system. Lets just have 2 second jitter between each sample as jitter is not a issue.
There needs to be some level of reclocking but not perhaps to the extent that some people claim but like you have said there is not much evidence except testimonies and experiences.

 

[darrylfunk]

bizarre.... Jitter doesn't guarantee better sound between different systems...

read up on the super clocks review in sound on sound and you will find that clocking with reduced jitter didn't always result in expected gains to sound quality.

 

[Daniel.]

Maybe all if this is a problem but try finding some proper analysis of jitter on two transports spdif out being traced to audible spuriae. By contrast benchmark have published figures showing that the jitter on the dac's output after hugely long spdif cables (can't remember-100m?) is still negligible

Iv heard that a system using pll causes audible distortions unrelated to jitter, could this be a possible reason why optical could sound inferior to async usb.

There is nice paper by Julian Dunn estimating audibility of different frequencies of sinusoidal jitter at high listening levels taking into account the audibility threshold at different frequencies. Iirc he came up with 75 ps. This was at frequencies though at which jitter attenuation should be easy, and ignored the masking effect of program material.

Jitter becomes more important at higher frequency's so wouldn't the jitter approximation be different for oversampling dacs, especially ones with the highest oversampling frequency's..

3) the argument that audio fidelity is down to jitter is plausible, but there's loads of speculation and guesswork involved. Once it's below 200 ps or so I'm not sure that any real evidence of audibility.

Yea its hard to say, best always to leave room for doubt and keep researching.

 

[Paul R]

jitter is focused on still with equipment design,

Mostly for marketing reasons to impressionable 'audiophiles'.

its whether how much error correction is needed before it stops making a difference.

I don't understand what 'error correction' has to do with it.

You seem to confuse in what i think jitter makes a difference with.

I want you to say what you think makes a difference, how much, and how you know.

If timing correction is not needed then lets not use pll or any kind of reclocking system. Lets just have 2 second jitter between each sample as jitter is not a issue.

That would be a 'drop out'.

Paul

 

[Daniel.]

Mostly for marketing reasons to impressionable 'audiophiles'.


I don't understand what 'error correction' has to do with it.

Jitter is a type of error, the more error there is the more correction may be needed.

I want you to say what you think makes a difference, how much, and how you know.

I think jitter correction is needed that's why we have phase locked loops and ASRC but i am not sure that any further jitter reduction is needed nor does jitter may still be a problem but i still leave room for doubt.

That would be a 'drop out'.
Paul

Yea exactly jitter that high has caused massive impact on the digital to analog conversion so jitter is a impact if it causes this to happen.

Paul I don't seem to be getting through to you or you don't seem to understand my point, you just seem to have a big idea that jitter is nothing at all important or you believe i think jitter is still important to the point of audible improvement, which i actually don't. All i tried to point out that jitter is focused on because engineers believe it does. Now please stop nit picking.

 

[Paul R]

It's not nit-picking to ask you to back up your claims about audibility. Which you appear to not be making any more.

The PLL in most S/PDIF receivers isn't there to reduce jitter per se, it's to sync the clock.

Whatever type of USB interface you are using there is no direct connection between the data transfer on the bus and the data rate at the converter, the data is always sent in bursts, the average needs to match the nominal sample rate. It's just the way USB works. There is no reason in principle why 'synchronous' USB audio should be worse than 'asynchronous' other than the engineering difficulty of matching the DAC clock to the USB data rate for 'synchronous' connections. And there's always a buffer...

Paul

 

[Daniel.]

I wasn't making any claims about audibility paul, only that jitter is important.