[LarryB]

I am aware that there are different schools of thought regarding the threshold for jitter audiabilty. I have no interest in fanning those flames so for the purpose of this thread, let's assume that all jitter is audible, and that the lower its levels the better.

I am curious as to the "average" jitter levels of various devices, such as:

1. A highly respected name brand CD player, for example one from Ayre, Krell, Wadia or Mark Levinson.

2. Garden-variety CD players used as transports.

3. Well-engineered transports (e.g., Mark Levinson or dCS)

4. Well made DACs, for example the Overdrive.

5. Less expensive but respected NOS DACs, for example Scott Nixon or ack Dak.

6. Reclockers, such as the Pacecar

7. Media server-like devices such as SqueezeBox and Sonos.

Thanks-in-advance for your patience, and for not laughing.

Larry

[audioengr]

I am aware that there are different schools of thought regarding the threshold for jitter audiabilty. I have no interest in fanning those flames so for the purpose of this thread, let's assume that all jitter is audible, and that the lower its levels the better.

I am curious as to the "average" jitter levels of various devices, such as:

1. A highly respected name brand CD player, for example one from Ayre, Krell, Wadia or Mark Levinson.

2. Garden-variety CD players used as transports.

3. Well-engineered transports (e.g., Mark Levinson or dCS)

4. Well made DACs, for example the Overdrive.

5. Less expensive but respected NOS DACs, for example Scott Nixon or ack Dak.

6. Reclockers, such as the Pacecar

7. Media server-like devices such as SqueezeBox and Sonos.

Thanks-in-advance for your patience, and for not laughing.

Larry

Here is how I would attempt to rate them, lowest jitter first:

1. Reclockers, such as the Pacecar

2. Well made DACs, for example the Overdrive.

3. A highly respected name brand CD player, for example one from Ayre, Krell, Wadia or Mark Levinson.

4. Well-engineered transports (e.g., Mark Levinson or dCS)

5. Less expensive but respected NOS DACs, for example Scott Nixon or ack Dak.

6. Media server-like devices such as SqueezeBox and Sonos.  (SB3 andDuet would be better than #7 and stock Sonos will be worse than #7)

7. Garden-variety CD players used as transports.

This is a difficult thing to generically rate because the clock quality and design implementations vary so much from one product to the next. It is also difficult to rate digital-only devices against D/A devices.  But for purposes of jitter-only, this is probably a reasonable rating list.

Here is maybe a more useful list for computer audio purposes:

1) Pace-Car 2
2) Off-Ramp 3
3) Duet
4) AppleTV
5) AirPort Express
6) Sonos


Steve N.

[LarryB]

Thanks Steve.


I dug around the Stereophile website and found these measurements of jitter:
 
Wadia 861:  190 psec

Ayre CX-7:  159 psec

Benchmark DAC-1:  119 psec

Squeezebox:  321 psec

Squeezebox transporter:  235 psec

Sonos ZP80: 388 psec

Genesis Technology (PS Audio) Digital Lens:

Fig.2 represents the identical test conditions and signals, but with the Digital Lens between the Lambda transport and the Theta processor. The periodic jitter components are much lower in level (except the spike at 4kHz), and the spectrum is generally cleaner. The RMS jitter level also dropped from 230ps to 160ps.

Repeating the measurements with a Classé DAC-1 processor yielded less of an improvement in jitter performance—the DAC-1 had better jitter performance to start with. Fig.3 is the DAC-1's jitter spectrum when processing a 1kHz, -90dB sinewave without the Lens. The RMS jitter level was 135ps. Note the rather high levels of signal-correlated jitter. Fig.4 is the same measurement, but with the Lens in the signal path. The spectrum is only slightly cleaner, but the RMS level dropped to 105ps.

-----------------------------------------------------------------------------
While I don't doubt that lowering jitter is beneficial, it is obviously not the only relevant parameter as evidenced by the fact that the Benchmark DAC has the lowest jitter yet is, IMO, terrible sounding.

[LarryB]

Here is maybe a more useful list for computer audio purposes:

1) Pace-Car 2
2) Off-Ramp 3
3) Duet
4) AppleTV
5) AirPort Express
6) Sonos

Perhaps I should order a Pacecar. Oh wait, I already did!

[audioengr]

The thing to note here is that jitter measurements like this dont tell the whole story.  Even these are incomplete.  The jitter magnitude is all that is measured in these quoted numbers and it does not specify whether this is RMS or cycle-to-cycle or peak or what.

The other missing parameters are what is the spectrum of the jitter, what correlation it has to the music signal and if it is changing, what is the modulation spectrum (random or low-frequency, high-frequency etc..).

It is these missing parameters that makes two devices with similar magnitude of jitter sound entirely different.

The human brain, not the ear is what detects these things so well.  Once the human ear is tuned-in to the music waveform, I believe it is extremely sensitive to perturbations and noise in the waveform.  The brain obviously compares the music to stored information, for instance: what a church bell sounds like.  If the bell is somehow not quite right the brain can easily detect this by comparing the reproduction to the brain-stored version of the real McCoy.

This is why I take published jitter measurements with a grain of salt.  They are simply insufficient to fully characterize the jitter.  It is a multidimensional measurement, real-time and spectral that is need to fully characterize jitter, and the modulation waveform must be characterized as well.  A very complex and precise measurement is needed.

When customers ask me why I dont spend a lot of time doing jitter measurements, this is why.

Steve N.

[LarryB]

Interesting, to say the least.

But you must rely on some measurments when developing your products.  N'est ce pas?

[audioengr]

Interesting, to say the least.

But you must rely on some measurments when developing your products.  N'est ce pas?

Certainly, but my ears are the final filter.

This is no different from designing a speaker or an amplifier.  You can make a lot of classical steady-state and even some rudimentary transient measurements, but even if these are perfect, it may still sound horrible.  The same problems exist in analog measurements as digital. They are simply insufficient to tell you what it will sound like.

Steve N.

[LarryB]

You don't have to convince me; despite bieng a scientist (neuropharmacologist, to be exact) by training, I am a subjectivist when it comes to audio.