USB a problem with jitter?

Folsom · « **on:** 14 Feb 2006, 05:36 am »

http://www.diyhifi.org/forums/viewtopic.php?t=453&postdays=0&postorder=asc&start=15&sid=14eb31b5d00edafaec900ee2d8ae52d4

I am curious..... If Jitter is SO BAD on USB then how is PC audio working?

Folsom · « **Reply #1 on:** 14 Feb 2006, 09:51 pm »

Aparantly some people disagree with Cat5 and networking like for SB's as well.

I thought that raw data, unprocessed is accountible bit by bit...... Jitter should not have an effect on any thing unless you are feeding a SPDIF signal through some thing, instead of data.

ctviggen · « **Reply #2 on:** 14 Feb 2006, 10:09 pm »

I can't really understand what they are testing. Jitter means some type of (typically random) deviation from proper clocking times. The reason a CD player has jitter is the DAC follows the signal of the data. The clock is "buried" (i.e., part of) the data. Any wavering in the edges that define the data will become jitter. On the other hand, a network doesn't get its clock from the data. The data itself arrives, and the SB generates its own clock. The jitter comes solely from the clock and should be very small.

In the site, they're showing a peak and what appear to be harmonics. But without a description of testing methodology, I can't tell if they're really measuring jitter or not. For one thing, since jitter is basically random, it shouldn't appear as a harmonic (it's not related to the underlying frequency; it's a random deviation from the frequency, but not multiplied by an integer).

ctviggen · « **Reply #3 on:** 14 Feb 2006, 10:15 pm »

Here's one definition of jitter:

Jitter is a generally undesired variation or fluctuation in an otherwise stable physical quantity.

Clock jitter is the variation in timing of a critical instant in a periodic waveform with respect to a jitter free reference

from: www.ieee-uffc.org/freqcontrol/tutorials/ FCS%20Tutorials%2000/Temple_files/slide0031.htm

This also shows how to determine jitter.

Folsom · « **Reply #4 on:** 15 Feb 2006, 12:19 am »

Yeah..... Well as long as bit by bit is happening it is all upon the DAC...

With CD players jitter seems to have a, in my experience, perverse effect on the actual sound quality, a real interpherance.... What else could you expect from an "on the fly" mechanical device? Do CD players buffers hold information say like RAM does for a computer, or do they hold streaming stair step waves? I think that might make a large difference. I am assuming they have a buffer?

Those USB things they did where at different hz, with a SPDIF signal which is nothing like raw data. If raw data, like a file transfer, is bit by bit and accounted for, any jitter would be benine.

I think some people might just be scared of the new wave

.

Jocko Homo · « **Reply #5 on:** 16 Feb 2006, 06:14 pm »

Did you really read the thread? Where do you get the idea that we are scared of anything?

The point is: USB has rotten close-in phase noise. Anyone who thinks it is jitter free is just plain wrong. I see lots of self-appointed experts who try to claim it is wonderful, with no data to back it up.

So, we post data and you think that we are scared. Give me a break.

No we aren't scared. Rather, we are the people who will design the products to solve the problem. Especially if there is a $ in it. And there will be.

Jocko

Folsom · « **Reply #6 on:** 16 Feb 2006, 06:36 pm »

Quote from: Jocko Homo

Did you really read the thread? Where do you get the idea that we are scared of anything?

The point is: USB has rotten close-in phase noise. Anyone who thinks it is jitter free is just plain wrong. I see lots of self-appointed experts who try to claim it is wonderful, with no data to back it up.

So, we post data and you think that we are scared. Give me a break.

No we aren't scared. Rather, we are the people who will design the products to solve the problem. Especially if there is a $ in it. And there will be.

Jocko

Yeah when I said some people I meant you Jocko and every one in the link, exclusively.

All I have to really talk about with USB and network, is bit by bit accountance. I can just not see how jitter could possibly affect any thing if some thing is done in a bit by bit process. I am not saying current stuff is, but we know it can be.

ricmon · « **Reply #7 on:** 16 Feb 2006, 07:21 pm »

So wahts the best way to get data from the pc to the preamp? spdif to dac, sb3, usb to dac?

chadh · « **Reply #8 on:** 16 Feb 2006, 08:15 pm »

I'm no electrical engineer. And much of that thread may as well have been in German to me. But I had the impression that their critique was not really a USB critique, but a problem with SPDIF. The data they were using seems to have come from audio signals sent via USB to a USB>SPDIF chip, and the SPDIF signal was then analyzed, no? And amazingly, they identified some noise that consistently appeared in the SPDIF signal.

If that is right then I don't know what the big deal is. I thought we already knew that USB->SPDIF conversion was suboptimal.

Again, I might be all wrong, but I also thought that they showed there were fewer sources of noise taking USB->SPDIF than using a soundcard that simply put out an SPDIF signal.

So, what is the best method of getting the signal from PC to pre-amp?

I think taking a USB signal straight to I2S might be a very good option. Most USB DACS take a normal SPDIF>analog DAC chip and stick an extra stage on the front that converts USB to SPDIF. Others are popping up that will take USB to I2S instead, and I2S apparently has fewer problems than SPDIF. I think the newest offering from Empirical Audio is a USB>I2S dac (is it the I2S offramp or something?). And Scott Nixon is now offering both the tube-dac and the chibi dac in USB>I2S form. I'm sure there are others.

Then there's the squeezebox. It has an SPDIF digital out, so I presume that it converts something to SPDIF and then uses a conventional SPDIF> analog DAC. But I'm not sure in which form the data is transmitted to the squeezebox. TCP/IP maybe? You can tell I'm making half of this up! Anyway, I'm not sure how the wireless transmission of data to the chip that converts to SPDIF in the squeezebox compares to USB->I2S. But people seem pretty happy with the squeezebox option when well modded.

I'm planning on a modified squeezebox for my home listening pleasure, and am hoping for a USB>I2S dac to use at the office.

Chad

ricmon · « **Reply #9 on:** 16 Feb 2006, 08:30 pm »

Just as I thought. Much to do about nothing. I am going to stick with the SPDIF to dac. No SB or any other divice or circuit between signal and dac. Plus I can use any media palyer of codec to manage my play list and play around (digital correction) with the signal to my hearts content. By the way I use onboard audio via a SPDIF header on the mobo.

JeffB · « **Reply #10 on:** 17 Feb 2006, 01:22 am »

This jitter stuff has me confused. I have the feeling that nobody is building reduced jitter circuits even if they think they are.

The seemingly logical thing to do is to read the digital data into a buffer and then have a high performance clock remove the data from the buffer and feed it to the DAC.

Using the timing of the arriving SPDIF data seems stupid. I don't see how USB is any better. Both of these formats just get the data transferred, but not at high precision timing.

Then there is I2S. I only read enough about this to know that there is a path for data and a path for timing. Why on earth try to encode the timing with the data transfer. You know the data is suppossed to be clocked at 22kHz.

If you go USB to I2S and the timing is off at the USB how do you correct the timing going to I2S? How does the timing signal correlate to the data?