Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
How odd. I'd guess that the vast majority of Asylum inmates could immediately describe and identify hiss from a tape machine or hum from a ground loop or the rushing sound a stylus makes as it dances in an undulating vinyl groove. Yet when it comes to jitter, the great bugaboo of digital audio, there doesn't seem to be a clear consensus of what it sounds like or what it does.
Tape hiss is obvious and constant. Any time distortion (digital jitter, wow, flutter) can be perceived only as an interaction between the recording and the playback timing. This makes recognition of low level time distortion difficult and inconsistent. The situation is not unlike that with wow and flutter in LP and tape. There are some subtle tape distortions such as scrape flutter that many audiophiles won't recognize and identify unless they are at high levels.
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Ah, good point. After a little thought I see that every example I gave could be grouped under the heading of "Noise," and that jitter should be classed as "Distortion."
Is this a good way of thinking about it? "Noise" is inherent to the playback chain and is always present, while "Distortion" is error created by how the signal chain handles the music itself. I'm struggling to figure out how frequency response deviations fit into that paradigm, perhaps that's why they are specified separately.
Anyhow, being constant, Noise must be far easier to recognize than Distortion, which is only present as a shading to the music itself. I guess one of the benefits of Digital is a dropping of the noise floor to levels Analog recorders are incapable of. Certainly it's easy to tell my "AAD" recordings from my "DDD" recordings simply by the presence of the tape hiss in the former. I suspect, though I don't know this for sure, that the Digital recorders also have a flatter frequency response than Analog recorders. I do know that the specifications for CD playback show a far flatter frequency response than that of any phono cartridge I've ever seen.
I'm still left wondering what "Jitter" sounds like. As it is a Distortion, I guess the question is tantamount to wondering what "Harmonic Distortion" sounds like. As you point out, the vary nature of Distortion makes it difficult to hear unless it's sufficiently high in level. I'm not sure I've ever heard Harmonic Distortion, perhaps because all the amps I've ever listened to have sufficiently competent circuits as to reduce it to the level of inaudibility. Certainly I don't sit up nights worrying about ways to reduce the Harmonic Distortion in my system!
If Jitter is just Distortion, then doesn't it follow that like any other Distortion it too could be reduced to the level of inaudibility? I guess to paraphrase the old question, "If a tree falls inaudibly, does any one care?" How soft does Jitter have to be before we can stop worrying about it? How loud does Jitter have to be before it becomes intrusive?
Jitter is a very non-linear form of distortion, and not harmonic at all. There is also a non-linear relation with signal amplitude in PCM encoding, because there are large numbers of bits changing at different signal levels. In fact, the worst jitter often occurs around the 0 level with the PCM twos-complement encoding because all of the bits change from 0 to 1. These changing bits on the data line are easily coupled to the clock in the form of data-correlated jitter (often thought to be the worst form) because of return path impedances (among many other mechanisms).
Whoa! I think I threw out my neck watching your post go over my head!
I can't say I "read" your post, as reading supposes some sort of comprehension, but after examining it several times I conclude that Jitter, like Frequency Response, doesn't fall neatly into the category of Distortion. That is, while Jitter is a Distortion, it would be best to treat it as distinct from the types of Distortions normally associated with audio gear. For example, Jitter is not normally an issue with amplifiers.
So timing errors are distinct from other types of distortion. Fair enough. It strikes me that timing errors have always been part of audio. We just didn't call them Jitter back then. Turntables had wow, tape decks had wow and flutter. As you point out with digital Jitter, this "analog Jitter" would also be non-linear and not harmonic at all. And yet even today the golden ears would never describe analog playback as hard, or glaring, or boring and uninvolving, or if they did, they wouldn't blame analog timing errors for the problem.
What is it about digital Jitter that makes it so pernicious? I find it incredible that timing errors within a competently designed electrical circuit could be greater than timing errors created by a moving mechanical device. To the contrary, I suspect the timing errors in an electrical circuit to be much smaller than the ones in analog devices, perhaps orders of magnitude smaller. For example, it's common knowledge that even a cheap electrical watch can have greater accuracy than the nicest mechanical watches. If listeners can dismiss "analog Jitter" why are we so concerned about the much smaller "digital Jitter?" How does the digital domain turn common sense on its head so that a much smaller cause has a much larger effect?
Sorry, the definition of jitter is fairly simple, but the causes can be much more complicated to explain. Note that there is jitter always present that is somewhat analogous to the wow and flutter example, and most people may not be bothered by it, or even recognize its presence. It can be caused by ground loops, so you may have a power line frequency on the conversion clock that becomes a jitter component. Many digital components have very large amounts of low frequency jitter and users aren't bothered by it all that much. Clocks drift with temperature and power supply levels and are easily influenced by other digital and analog operations in the device (and even outside the device).
But the type of jitter I was describing as data-correlated is mostly caused by highly non-linear interactions between the data and clock while still in the digital domain, so doesn't really sound like something that happens naturally in the analog world. Maybe something like a distorted ghost of the music in the background, but also displaced in time. Music is repetitive by nature, and the repeating data bit patterns can be coupled to the conversion clock through a variety of means because the data waveform is not symmetric like a clock. Even a simple operation like transferring the data and clock from a digital filter output to the D/A convertor input can add data-correlated jitter to the clock since they share the same return, and the data bit pattern will modulate that return.
Post a Followup:
Post a Message!
This post is made possible by the generous support of people like you and our sponsors: