Jitter for Dummies

[hdomke]

Jitter is a term I hear a lot these days, but I don't really understand it.

 

Supposedly, a key reason Music Servers sound better than CD Players is that hard drives are less prone to jitter.

Expensive CD players advertise that they have less jitter.

The Digital-to-analog converter I just bought (Benchmark's DAC1) heavily promotes it's own way of reducing jitter.

 

• What is jitter?

• What does it sound like?

• How is it measured? Do all agree on these measurements?

• Do all digital sources of music suffer from Jitter?

 

[The Computer Audiophile]

Welcome to Computer Audiophile. Very good question. I think I have heard a million explanations of jitter, but still don't understand it. I'm not an engineer so it is one of those things that I just know is bad and leave it at that. This doesn't help you at all though. I will see what I can find and hopefully Steve Nugent or Gordon Rankin will lend a hand on this one.

 

[Wavelength]

Gang,

 

I am not an expert at jitter but have had to deal with it in many areas of digital design.

 

Basically any data that is synchronously clocked with data can have jitter. So in terms of this example there would a CLOCK and DATA seperated on the circuit. When the clock is not stable or varies from the data jitter induced errors can results.

 

In digital audio there are a minimum of 3 clock signals. These are BCLK (bit clock same as CLOCK above), WCLK (word clock determines R or L channel data) and DATA (same as above).

 

Any digital audio product can have some jitter.

 

Most of what happens in SPDIF is from the idea that the CLOCK is wound around the data in one stream and removed on the other side where the DATA and CLOCK are seperated. The SPDIF receiver then has to create BCLK, WCLK and DATA and clock it out. From this jitter can be caused by the sending device or the recieving chip. It can then be removed or reduced by external circuitry.

 

With USB and Firewire only DATA is sent over the link therefore USB/Firewire don't have any jitter because there is no clock associated with the DATA. But the receiving controller can add "intrinsic jitter" to the signals BCLK, WCLK before they leave the part. In USB and Adaptive mode the Computer tells the USB Controller to change the clock every 1ms or so. This then changes the clock and in it self causes jitter. On the programable parts like the TAS1020 this can be reduced but most code changes ever 4ms and still causes issues. Another key factor is that all the USB controllers use Frequency Synthesizers to create the BCLK and WCLK signals. USB uses internally a 12MHZ clock and it's kinda hard to create a Audio Master Clock (4xBCLK) from the 12MHZ USB Clock. For example 44.1K uses an 11.2896MHZ clock/4=2.8224MHZ for BCLK/64=WCLK=44.1K. In creating these clocks allot of jitter is added into the stream.

 

In Asynchronous USB and external Master Clock can be used bypassing the Frequency Synthesizer and resulting in a much better (100x) clock than the internal ones. In ASYNC mode the DAC actually controls the computer. Therefore the clock at the dac never changes and the results are much better.

 

~~~~~

 

Testing...

 

There are 3 companies making audio test gear. Audio Precision, Miller Research and Prism Sound. There are others but these are the main ones. Neither Prism nor Audio Precision feel there is a way to Numerialize the amount of jitter. Miller is the only company that gives a numerical output usually given in pico seconds (ps).

 

In SPDIF the JTEST is used, well Miller uses this for everything but it was developed by the late Juliene Dunn (who worked at both Prism and Audio Precision). This is what you see in Stereophile with the 11.025KHz signal at -6dB and 223Hz at 1 bit resolution. They take the FFT output and look for matched side band peaks. What this means is there will be a spike at 11.025K on the FFT indicating the carrier signal. Then there will be paired spikes on each side of equal value. Those can be from Juliene's work tabulated and equaled to the Numeralized value.

 

The problem is that the number (from Prism and Audio Precision) would never be rock consistent.

 

The engineers at Prism feel that this test is not applicable to USB/Fire/PCI etc audio testing because here there is no clock associated with data. In their analyzer you can specify output devices that are on the analyzer or any device hooked to that computer. Therefore Prism offers the only true native testing enviroment to date.

 

Nobody agree's with the data but then again none of use really agree on much so does it matter.

 

Jitter is only a testable barmometer for how well digitally clocked data reaches the dac. This is only one thing there is distortion, drive, technology, well.... just about everything effects the sound.

 

Best bet is always listen to something before you buy it. Believe me I heard a ton of really expensive stuff suck wind last week. Don't waste the cash if you don't need too. Contact a dealer you will be happier than buying something from Audiogon and trying to recover from that.

 

Thanks

Gordon

 

[Lord Chaos]

I never thought about jitter until I started listening to music through my brand-new Benchmark DAC1, which I bought because it was an all-in-one solution. I'd never considered buying a separate DAC before that. When I hooked it up and found out how good it sounded I naturally started wondering why. Benchmark makes a big deal out of their jitter immunity. Perhaps that's why their device sounds better than the other DAC-equipped devices I've used. I don't know. As with anything I take this with a healthy addition of salt.

 

[Hierge]

Jitter is a term I hear a lot these days, but I don't really understand it.

 

Supposedly, a key reason Music Servers sound better than CD Players is that hard drives are less prone to jitter.

Expensive CD players advertise that they have less jitter.

The Digital-to-analog converter I just bought (Benchmark's DAC1) heavily promotes it's own way of reducing jitter.

 

• What is jitter?

• What does it sound like?

• How is it measured? Do all agree on these measurements?

• Do all digital sources of music suffer from Jitter?

 

I will take a shot at the second question. I found an online guide that answers the other three questions, but I don't have the link right now.

 

I used to listen to Winamp and MP3 audio for the last ten years until I found CA. I am a fan of the band Coldplay, particularly an album called A Rush of Blood to the Head. There is a track called In My Place on this album that is a beautiful song, but the overbearing cymbals are a disaster of what I now know is Jitter.

 

On my old Winamp setup with a straight soundcard, these cymbals sound like MUSH and pretty much ruin a nice song. After I made the discovery of computer audiophile technology, I read a discussion of Dragonfly designer Gordon Rankin's discussion of how a computer clock is the culprit on jitter.

 

Mainly by using an external USB DAC with asynchronous technology you can use the external DAC circuits to fix jitter to a level below audible limits.

 

I experimented with this and found this to be the single greatest sound improvement of a DAC. It's like looking through a clean window. That dramatic.

 

The "common wisdom" on CA is pretty much panning out regarding DAC technology. The most dramatic sound improvements I have heard were when I ditched Winamp though. I am addicted to JRiver and the exotic players like HQPlayer and Bughead which sound truly amazing.

 

I can even listen to my computer sound card with HQPlayer with jitter and it is still amazing. Once you go low jitter though, you can never go back. That's my opinion anyway. Oh, and lossless audio really sounds much better than MP3. I still listen to a lot of MP3, but I'm slowly replacing all these with lossless recordings which sound amazing.

 

I'm very happy. This is the golden age of computer audio.

[firedog]

1. Jitter means there are timing errors when producing analog from the digital stream. The digital info for audio consists of clock info as well as the "bits"/data of the file itself.

 

2. What does it sound like? It's easiest to hear with percussion instruments, female voices, and plucked/strummed strings, IMO. The timing errors make the music sound less clear, less precise, less detail, edgy: more "digital". Less natural. Enough jitter will also make the playback lose a lot of the spatial clues, so the soundstage and instrument placement placement are less well defined, more restricted.

 

3. Good question. There isn't a standard for measuring it, so pretty much any jitter measurements you read are useless for comparison purposes, unless you know they were done the same way.

 

4. All digital sources have some jitter. The question is how much and what kind. For instance, if the jitter is random you probably won't hear it; if it occurs at a, or several, specific frequencies it can alter your perception of the playback. In spite of what Benchmark and other manufacturers say, no DAC is immune to jitter.

 

Audiophiles tend to call everything in digital playback that they don't like "jitter". Often what they are hearing is not jitter, but some other problem in the playback caused by electrical noise or who knows what, but not by timing errors.

[audiventory]

• What is jitter?

 

Digital signal’s samples clock instability.

 

• What does it sound like?

 

It’s additional noise. More jitter - more noise.

 

• How is it measured? Do all agree on these measurements?

 

Need pro measurement hardware. Method based on comparison with reference clock source.

Reference clock source must have precision 3 … 10 times better measured device stability.

 

If measurement developed and performed correctly from technical point of view, why not agree?

 

As any measurement we can use different methods. And get different results. Even for all correct methods.

 

Here need consider measurement conditions, method and result as solid piece.

 

• Do all digital sources of music suffer from Jitter?

 

Jitter present absolutely for all digital hardware due there use clock sources anyway.

These clock sources non-ideal, of course.

 

Jitter value defined by clock source stability. It's price matter.

 

In pro applications used thermostat decissions. There clock source work in constant auto-adjusted high temperature environment. It allow provide higher clock stability.

[Jud]

http://www.thewelltemperedcomputer.com/Lib/Troisi.pdf

 

The effects of jitter are to raise Total Harmonic Distortion and overall noise of the analog result. Of the two of these, the first is thought to be more audible and more detrimental to the sound.

 

As with nearly everything in digital audio, it seems, there is debate/discussion regarding how audible jitter is, particularly at low levels. I've seen jitter measured in terms of time error - usually picoseconds, ps - and loudness, dB. The former measurement tends to be used by people who are arguing for audibility, the latter by people arguing against.

 

Jitter is also known among more technical folks as "phase noise."

[4est]

Jitter is a measurement of phase noise in time. As with other noise it manifests itself differently depending upon what that noise is. The most obvious examples of sonic attributes when jitter is lessened in PCM are: better PRaT, tighter bass and a more clearly defined 3d spacial image (especially in the highs- you should be able to "see" the roundness of a cymbal).