Head-Fi.org › Forums › Equipment Forums › Sound Science › A question about jitter
New Posts  All Forums:Forum Nav:

A question about jitter - Page 2

post #16 of 22
mike1127, perhaps I misunderstood the original question? I think the OP was asking about jitter with respect to digital signals. Your answer didn't seem to make any reference to PCM, I2S, S/PDIF, or other digital formats used for communication of audio data. Jitter on a digital signal can indeed result in the receiver incorrectly interpreting the value.

If clock compensation for jitter on an incoming digital signal affects the clock used to create the analog signal in DAC, then I usually refer to the "corrupted" analog signal as distorted rather than as jittery. I don't think I've ever heard of a jitter value being used to describe an analog signal.

Re-reading your post, were you trying to illustrate jitter in the ADC resulting in recording the incorrect information in the first place? You mentioned both a recorder and a CD player.
post #17 of 22
Quote:
Originally Posted by WesMiaw View Post
mike1127, perhaps I misunderstood the original question? I think the OP was asking about jitter with respect to digital signals. Your answer didn't seem to make any reference to PCM, I2S, S/PDIF, or other digital formats used for communication of audio data. Jitter on a digital signal can indeed result in the receiver incorrectly interpreting the value.

If clock compensation for jitter on an incoming digital signal affects the clock used to create the analog signal in DAC, then I usually refer to the "corrupted" analog signal as distorted rather than as jittery. I don't think I've ever heard of a jitter value being used to describe an analog signal.

Re-reading your post, were you trying to illustrate jitter in the ADC resulting in recording the incorrect information in the first place? You mentioned both a recorder and a CD player.
I don't know every way the term "jitter" is used, but generally it refers to timing errors in the conversion from digital to analog, not to bit errors.

This is related to digital signal theory in the most general sense.

S/PDIF, PCM, etc are not part of digital signal theory. They are part of practice.

There is some overlap between the notion of jitter at the DAC chip and jitter in a digital signal. Jitter in the digital signal will probably harm the DAC's circuitry ability to convert to analog jitter-free. This may depend on the DACs particular circuit.

Jitter does happen in the recorder, that is, the A/D chip. It creates distortion there (in the form of incorrect values in the digital data). Then more jitter happens in the DAC.

As I said, digital signal theory requires that the samples be perfectly evenly spaced in time. That is a prerequisite for the mathematical magic to happen. To the extent this doesn't happen in practice, digital is not a distortion-free method of recordings and playback.

Again, for those people who find it hard to believe a finite set of samples can reproduce an infinitely smooth waveform, it really can happen in theory. The prerequisites are (1) the input signal has to be band-limited, and (2) the samples have to be perfectly evenly spaced in time. Neither of these things are perfectly true in practice.
post #18 of 22
Quote:
Originally Posted by mike1127 View Post
I don't know every way the term "jitter" is used, but generally it refers to timing errors in the conversion from digital to analog, not to bit errors.
Ah, okay. This is what led to our debate. The definition of jitter I use is "the time variation of a periodic signal". Doesn't matter what the signal is carrying or what signal processing is happening. (If it is not periodic, i.e. tied to some sort of clock, I don't know how you would characterize timing errors.)

Jitter on an incoming digital signal could result in a one being interpreted as a zero and vice versa. This is more likely when the receiving clock is independent of the sender clock. Less likely when the clock is being derived from the signal itself (e.g. S/PDIF). And even less likely if the receiving clock and sending clock are the same (e.g. I2S).

I think I understand what you were/are saying now. If the master clock (a digital signal) of the DAC chip has jitter, then even though you are interpreting all incoming ones and zeros correctly, you are flipping your transistors with the jittery clock. And so your analog voltage might hit voltage V at time T+e instead of at time T.

But you were also saying that even if your master clock is perfect, your analog voltage might hit voltage V at time T+e instead of at time T, just because you can't flip transistors at perfect intervals. And this also falls under your definition of jitter.
post #19 of 22
Quote:
Originally Posted by WesMiaw View Post

Jitter on an incoming digital signal could result in a one being interpreted as a zero and vice versa. ....
I think I understand what you were/are saying now. If the master clock (a digital signal) of the DAC chip has jitter, then even though you are interpreting all incoming ones and zeros correctly, you are flipping your transistors with the jittery clock. And so your analog voltage might hit voltage V at time T+e instead of at time T.
The reason I focused on my definition was that it is a direct compromise of digital signal theory. Getting bits wrong is just a detail. But jitter in the A/D and D/A conversion directly attacks the integrity of the theory.
post #20 of 22
Quote:
Originally Posted by Sherwood View Post
it just outputs it. The data was already sampled when it was converted to digital. Jitter is the problems with the accuracy with which CD players output ditial data to a DAC.
The source jitter is probably the largest contributor of jitter in directly driven DACs, reclocking ones don't really need to care about the timing, as long as the data rate does not fluctuate too much. Assigning blame for jitter to the cable should be done very carefully, as the quality of the clock at the source and the (in)ability to drive a capacitive load by it are far more likely causes.
post #21 of 22
Quote:
Originally Posted by WesMiaw View Post
The definition of jitter I use is "the time variation of a periodic signal". Doesn't matter what the signal is carrying or what signal processing is happening. (If it is not periodic, i.e. tied to some sort of clock, I don't know how you would characterize timing errors.)

Jitter on an incoming digital signal could result in a one being interpreted as a zero and vice versa. This is more likely when the receiving clock is independent of the sender clock. Less likely when the clock is being derived from the signal itself (e.g. S/PDIF). And even less likely if the receiving clock and sending clock are the same (e.g. I2S).

I think I understand what you were/are saying now. If the master clock (a digital signal) of the DAC chip has jitter, then even though you are interpreting all incoming ones and zeros correctly, you are flipping your transistors with the jittery clock. And so your analog voltage might hit voltage V at time T+e instead of at time T.

But you were also saying that even if your master clock is perfect, your analog voltage might hit voltage V at time T+e instead of at time T, just because you can't flip transistors at perfect intervals. And this also falls under your definition of jitter.
I think you are talking about setup and hold time. There is a bit clock which is derived from the transmit bit stream. So if the data has jitter, the clock will still be able to track it. You can not have two clock sources to clock the data.
Then there is also the matter of the sampling clock which is derived from the byte clock. This has to be synchronous to the original sampling clock or you will have a buffer unferflow/overflow problem. Normally jitter in audio refers to this clock (44.1KHz).

A simple way to look at this is. The transmission line carried all the information to build a frame of picture. Then you play this frame at 30 frames/s. If you have jitter at the playing clock, you'll have a jittery picture.
post #22 of 22
http://www.jstage.jst.go.jp/article/ast/26/1/50/_pdf

Good article about jitter, there was no audible effects until 750ns or 750,000ps in a blind test and 30ns or 30,000ps in a sighted test where the user had control and knew how much jitter was being applied. This is much higher the what is in equipment today.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › A question about jitter