Could it be jitter is NOT the enemy of digital audio?

Welwynnick

Distinguished Member
I've been skirting round this issue for a while, but never have time to get the post together. This is what has been vexing me, though.

When discussion of digital audio turns to digital, people always talk about jitter as if it's THE evil to be avoided. Sound quality is seen as a function of jitter, so if you improve that, you improve SQ.

With a scientific and engineering backrground, I always wanted to explain sound quality in terms of things that could objectively assessed, but like Gary, I learned to trust my ears when they disagreed with any measurements. My thoughts are that jitter is a figure of merit, but is not the principal factor affecting sound quality of digital audio. How could it be? The timing variations we are talking about are very much smaller than the period of the highest frequency, but I get the impression that most people regard jitter as a factor that carries through the system, and applies a phase distortion to the analogue waveform.

Is that what people really think it does?

I don't. I think there is something else much more fundamental that affect sound quality, and jitter happens to influence it. Bit Error Rate. Digital systems are never perfect; that's not the point of them. They are simply pragmatic, cost effective solutions to getting as much information over a link as possible with adequate quality and integrity. If the link happened to be perfect, that is with no bit errors, then it would be over-engineered and in-efficient. It should be carrying more data to justify it's costly resource consumption (typically bandwidth). Digital systems generally use error correction, and often adaptive power control mechanisms, to control the bit error rate to within acceptable target limits. But they do not eliminate errors.

I could be barking up the wrong orchard here, but my thought is CDs are just the same. CDs are very primitive by modern standards, and there is a significant amount of redundancy in the error correction data stored on the disc. That is only needed because not every bit can be read correctly - a small proportion of them are read in error, though modern drives must be better than the early ones. But where error correction is applied, the result is generally only to reduce the bit error rate, not to eliminate it. I'm pretty sure that even the best CD players cannot read every bit accurately even after error correction, and although a player can guess what it should have been, that error will always be there. It may be the least significant bit, which is probably inaudible if wrong, but equally it may be the most significant bit, which would be bad news in a quiet passage.

I am beginning to think that the reason jitter is important is because a poor quality digital interface will increase the BER from the player. When jitter becomes significant relative to the bit period, the normal distribution of jitter will eventually cause a small proportion of bits to arrive at the wrong time, and cause a bit error. The added BER will then be a function of the jitter (when considered relative to the bit clock period).

I suspect this is the mechanism where jitter degrades SQ. If jitter is engineered to insignificant levels, though, the contribution to BER will also become insignificant, and no further improvement in SQ will be possible. That will still be limited by the native BER from the CD drive.

The potential for better SQ from computer audio may come from the different way the digital audio data is generated. I suggested before that PCs are not error tolerant like CD players, but I didn't explain myself very well. There are errors associated with both drives, but the mechanism to ensure accuracy ( and a low BER) is quite different. When a PC reads a CD, it will perform error checks, and will re-read a sector if there are errors. A CD player (with some honorable exceptions) cannot do this; it just streams the data off the disc in real time, as it is required. Where there are errors, it will just do it's best to correct them, but cannot go back and read again until the data is correct.

When bits are read off a CD in PC, ripped to a HDD, and subsequently played off that HDD, they are read, buffered, tested, and re-read until again until they are right. That does NOT guarantee that there are no bit errors at all, but I believe that this more robust reading mechanism produces a lower BER than a CD player. Obviously, you still have to get all the other digital processes right to capitalise on this head start, but if you CAN avoid subsequently adding too many bit errors, then the final output may be better than a CD player, nomatter HOW good the digital interface to the DAC.

Unfortunately it's difficult to prove this because the industry is obsessed with jitter instead of BER, which I think is the fundamental culprit, but is not assessed and compared in the same way. Of course CD players may still be better then PC audio, but when you compare the extraordinary effort that some manufacturers go to, with the value-for-money-at-all-costs approach to manufacturing PCs, it's not that hard to understand why. Flimsy and full of interference, compared with bank-vault mechanical and electrical integrity. If only HTPCs could be built with the same attention to quality and detail, the benefits of PC audio might be mor apparent.

Sorry I wasn't more concise, but I got it off my chest now!

Nick
 

deaf cat

Active Member
:)
So is there any way to actually measure bit errors?

I use EAC to rip cd's, but have not really got a clue as to all the ins and outs of how it works.

I hope to demo a couple of streamers in a few months time which have very low jitter ratings, if the EAC, and software, and wireless network do there job of minimal bit error, the end result should be pretty impressive.:smashin:
 

CJROSS

Well-known Member
welwynnick - jitter has never bothered me, not one iota in my DAC experiences. CD Transports were held aloft as being special when they measured at 200 spec of jitter (Theta Carmen for example), these days £200 DVD player can better that. But even then the audiophile biased industry will find ways of deriding even low spec jitter devices like £200 DVD players. I think back to the Paul Miller reviews of devices like the Theta and what it would be like with say a £200 Toshiba being measured with a Theta badge on it. :rotfl:
 
J

jimmy_b

Guest
welwynnick said:
When bits are read off a CD in PC, ripped to a HDD, and subsequently played off that HDD, they are read, buffered, tested, and re-read until again until they are right. That does NOT guarantee that there are no bit errors at all, but I believe that this more robust reading mechanism produces a lower BER than a CD player. Obviously, you still have to get all the other digital processes right to capitalise on this head start, but if you CAN avoid subsequently adding too many bit errors, then the final output may be better than a CD player, nomatter HOW good the digital interface to the DAC.
Nick
I don't have a great technical background but the above confuses me. When a PC reads from a CD I am sure it doesn't know/care whether it is reading a music track or a word file. However, I have NEVER had an error in any word files I have read from a CD (or any other file for that matter). Even if there is just a single bit error then a word file would be corrupted in some way. Given that this doesn't happen (unless the CD is damaged and a "1" can't be differentiated from a "0") then I would say that there is zero BER in this process. Am I missing something......?
Jimmy
 

Cable Monkey

Well-known Member
I think you are basically supporting what Nick said, that you don't get errors with pc because of the robust error correction. No such luck with audio cd players and as a direct consequence I have been able to rip (with EAC) and replay tracks unplayable on any audio disk spinner. However the same CD could not be played on the pc or ripped with a more conventional ripper (Itunes). Consequently some of the pc's superior capabilities are due to software used as well as their inherrently better error correction. This then ties in with reading a word document that has software to allow the contents to be viewed if the checksum is correct. If not the pc simply tells you the file is corrupt. Same with a FLAC file which plays but will stop and give an error message if the checks reveal an error so it is either perfect presentation or nothing. Anyone who has heard a CD skipping and screeching knows a CD player will manfully try to carry on regardless of the problems encountered!
 

Welwynnick

Distinguished Member
jimmy_b said:
When a PC reads from a CD I am sure it doesn't know/care whether it is reading a music track or a word file. However, I have NEVER had an error in any word files I have read from a CD (or any other file for that matter). Even if there is just a single bit error then a word file would be corrupted in some way. Given that this doesn't happen (unless the CD is damaged and a "1" can't be differentiated from a "0") then I would say that there is zero BER in this process.
Exactly!

Reading ANY data off discs or drives in a PC is more robust and reliable than playing a CD in a player. A player is a real time streaming source, where the drive and the DAC and everything in between are clocked together, and there is no opportunity to go back and re-read. Unlike a PC.

Nick
 
G

gallium

Guest
Most CD players (if not all) are designed to output zeros if they hit unrecoverable disk errors. Otherwise you'd get screaches and pops in the audio.

There is significant error correction coding on redbook and although raw BER is pretty high (dust and other foreign objects also cause errors), there is enough redundancy to render the result as a perfect bitstream.

Regarding jitter, its quite easy to eliminate with a small buffer and associated quartz clock, even if you use a $30 DVD player as transport.

All these miracle transports are nothing more than well built, overpriced paperweights with about $20 of mechanical parts and electronics for low-jitter output buffering. Nothing more than a $15 PC cd-rom can't do these days.
 

PeteD64

Active Member
Some interesting thoughts Nick.

I'm also of the opinion that a hard disk makes a much more robust storage medium & transport for digital music than a conventional CD & player. The ability of the PC to read the data from disk faster than real time & correct errors should be a big advantage over regular CD players. I think I remember reading that the Linn CD12 did this but it cost a small fortune compared to a PC.

I've finally ripped my CDs to my PC & hooked it up to my hi-fi this year. The next stage is to improve the quality of the playback which is limited by the on-board sound at the moment. I'm thinking of one of the RME cards at the moment but the new Slim Devices Transporter could be promising as well.

The Off Ramp Turbo box also looks interesting but seems to be very expensive for what it is. Must have been coated with some Hi-Fi "snake oil!What's with the "it sounds better when powered by a battery" stuff?

The other option that intrigues me is having network connectivity built into an AV amp, or even better a processor. That way you don't have to worry about the quality of the analogue or digital connection from the PC to the amp, it can just pick up the music over the network & play it itself.
 

stiben

Active Member
PeteD64 said:
I'm thinking of one of the RME cards at the moment but the new Slim Devices Transporter could be promising as well.

If you're thinking of the Transporter should should also consider the Lynx L22 - best soundcard out there.

I recently upgraded to a RME HDSP 9632 and I like listening to it a lot more than the E-mu 1212m I had for 2 years, even though the 1212m is superb.

Better bass, the soundstage and separation of instruments are also dramatically improved. Highs are not quite as detailed (sharp) but I plan on upgrading the capacitors to fix that, copying a mod on Head-Fi.

For £220 S/H it was a bargain considering how much an equivalent sounding CDP would cost. Same goes for the 1212m at £120.
 

Welwynnick

Distinguished Member
PeteD64 said:
The Off Ramp Turbo box also looks interesting but seems to be very expensive for what it is. Must have been coated with some Hi-Fi "snake oil! What's with the "it sounds better when powered by a battery" stuff?

The other option that intrigues me is having network connectivity built into an AV amp, or even better a processor.
The Off-Ramp Turbo wouldn't be the first hi-fi product that is supposed to sound better with battery power. There was a Mission pre-amp many years ago with batteries, and was supposed to sound better as a result. I know modern equipment often has good power supplies and regulation, but when the ultimate source of the sound is AC, you can't stop all interference getting in, and some of it will be added by the PSU itself. Batteries give a lovely clean smooth low-impedance supply, and I heard the ORT benefits quite significantly compared with the mains supply. They're just not very practical that's all.

Network connections may have their advantages too. Cables cause all sorts of problems, like noise, interference and ground loops. If wireless is reliable it may work better than cables. Not entirely sure, yet, though.

Nick
 

PeteD64

Active Member
welwynnick said:
Cables cause all sorts of problems, like noise, interference and ground loops. If wireless is reliable it may work better than cables. Not entirely sure, yet, though.
I'm not sure any of that matters too much with an ethernet or wireless connection between PC or server & an amp. The higher level network protocols, assuming it's TCP/IP based like on the Denon 4306, are designed to make sure all the bits get there & in the correct order & ask for re-transmission if not. The music is still just a bunch of data at this stage. It's only once the amp converts the music to an analogue signal & sends it to the speakers that any of these low level effects can alter the sound quality.
 

Nic Rhodes

Distinguished Member
Nick

I think agree with what you have written here.

Jitter needs to be contolled and at a certain point this largely eliminated the issues it's brings. We talk glibly about how easy it is to stick a PLL / buffer in to elimate jitter on £30 DVD player and we don't do this 'dedicated' transports but the relaity is somewhat different. True it is now relatively easy using modern chips to get a decent jitter figure nowdays(not previously) but many many players do not do this still. Just look at the HCC reviews and you will see many jitter figures in the 1000s. You will also see similarly well priced players that have been designed to be fit for purpose. These techniques used are part of the issues but are not a complete solutions by any means. PLL are for stability of lock onto the digital signal first and foremost, and only secondarily for jitter reasons that some hint at here but yes they could be used but it might need more the class dual PLL.....

Dedicated transport vary significantly from rebadges with clocking (Theta rebadged Pioneer!!) to dedicated machines. My CD has 56 power supplies for jitter related reasons...however for BER than physical stability of read syatems and crystals should also be considered. Some use multiple reads, other used lead boxes on leaf springs...and we still have the bad boy in all this SPDIF interface.

Jitter is certainly not the bee all and end all but it something that has to be taken seriously, especially with PCM signals. There are many other things that need to be considered and I think the Meridian approach under pins your agument here re BER. They do what they do for cost reasons, others however have been 'less accurate in the use of their engineering tools' that they utilise and have thrown SOTA OTT engineering at the transport. Both have merit and are differnt engineering 'styles'.
 

karkus30

Banned
welwynnick said:
The Off-Ramp Turbo wouldn't be the first hi-fi product that is supposed to sound better with battery power. There was a Mission pre-amp many years ago with batteries, and was supposed to sound better as a result.

I was always under the impression that batteries had a habit of causing white noise, maybe these newer cells are far better ?
 

The latest video from AVForums

Star Wars Andor, Woman King, more Star Trek 4K, Rings of Power & the latest TV, movies & 4K releases
Subscribe to our YouTube channel

Full fat HDMI teeshirts

Support AVForums with Patreon

Top Bottom