Jitter in firewire/USB connection?

[nopietns]

Hi, still looking for my ultimate DAC, and have got one question on which I can't seem to find answer.

Is the jitter in USB/Firewire connection of a DAC the same concern as in connection through Toslink/Coax or is it somehow different?

 

[jimmyjames8]

I just did a quick google on the subject and the consensus seems to be that jitter is not an issue with USB interface as it is asynchonous.

 

[nopietns]

I'm not that much into electronics/computers, so "asynchonous" didn't mean anything to me when I googled.
And how about FireWire?

So that would mean that a dac with USB connection would reduce the jitter impact on the quality of the sound?

I guess the main advantage is that they both carry an encrypted signal, which (I guess) more reliable than Coax/Toslink considering the jitter.

Quote:

Originally Posted by slwiser Firewire is also considered a bus whereas the USB is considered only a signal path where data is push out to anyone and anything able to capture it. Being a bus you have two way communication available. Sending and receiving units talk to each other to ensure communication integrity. In other words, no lost bits if the receiver is not ready. This is the specific reason it is used in video equipment.

 

[thomaspf]

While a USB audio connection is usually isochronous, which is asynchronous but with a constant average bit rate, that does not automatically guarantee lower or no jitter.

The most common USB audio mode still has the master clock of the playback chain at the PC. The USB connected DAC must estimate the clock by averaging over the incoming signal rate.

The clock output of the common TI USB audio chips have been measured to contain quite a bit of jitter. Higher end USB audio components either use addtional dejitter stages similar to what you would do for a S/PDIF connection or use USB async mode which allows the DAC to be the master of the playback chain.

USB audio async mode still uses isochronous connections but by establishing a back channel from the DAC to the PC, the clock in the DAC can pace the data flow from the PC. In that case the jitter is determined by the quality of the clock in the DAC and the implementation of the conversion driven by that local clock. There are only a few solutions on the market that use async USB audio. Gordon who also posts on this forum has become quite the expert since he managed to implement async in his devices by handcrafting the firmware for a standard USB audio chip.

Cheers

Thomas

P.S.: for completeness

USB audio async mode is not unlike a DAC with a word clock output that you feed back to a source so the S/PDIF input is slaved to that clock. The benefit of USB is that you only need a single standard cable and no expensive sound card.

 

[nopietns]

Oh, thanks. In that case the USB connection doesn't seem the ultimate one.

 

[bigshot]

Jitter, as it occurs in home audio components, is 100 times below the threshold of audibility. It isn't an issue. Coax, USB, optical... it's all the same.

See ya
Steve

 

[nopietns]

FWIK, the jitter just changes/colours the signal.
Stereophile: Jitter & the Digital Interface

 

[jimmyjames8]

Quote:

Originally Posted by bigshot /img/forum/go_quote.gif Jitter, as it occurs in home audio components, is 100 times below the threshold of audibility. It isn't an issue. Coax, USB, optical... it's all the same. See ya Steve

Gosh. I wish I could be as **** sure of myself about everything as you are. It must be a burden.

 

[DefectiveAudioComponent]

If you are worried about jitter then just select a DAC that transfers audio in bulk mode.

 

[bigshot]

Quote:

Originally Posted by jimmyjames8 /img/forum/go_quote.gif Gosh. I wish I could be as **** sure of myself about everything as you are. It must be a burden.

Just take about 20 years of your life to do your homework- ask people who know a lot of questions and listen to the answers carefully. I'm happy to bear the burden for you.

There are things that *do* have an impact on sound quality. Jitter isn't one of them.

See ya
Steve

 

[nick_charles]

Models of jitter audibility place it in the low 10s of ps (peak to peak) certainly effective bit-depth of a 16 bit system is degraded by 1 lsb at that level, less for 20 bit systems.

However the emprical evidence for audibility of both deterministic (signal-correlated) and random jitter at any single frequency or within musical signals places audibility in the 10s of ns for deterministic jitter (Benjamin and Gannon, Dolby labs, 1998) and in the 100s of ns for random jitter (Ashihara et al , 2005). These figures arrived at after fairly rigorous testing and measurements.

Evidence for the audibility of really low levels of jitter is (so far) anecdotal , not backed up by rigorous measurements and under dubiously controlled test conditions.

You pays your money and you takes your choice.

I think it is time for jitter solution vendors to back up their claims with serious tests.

 

[bigshot]

Who needs serious tests when you can get by anecdotal evidence not backed up by rigorous measurements and under dubiously controlled test conditions?!

The concept of jitter is designed to differentiate expensive high end audio components from inexpensive mid-range ones that sound exactly the same. It's an inaudible upsell. Stereo salesmen play on people's fears of theoretical problems to get them to spend more on their equipment than they need to. Jitter is a hoodoo.

See ya
Steve

 

[thomaspf]

I guess what you guys are saying is that the clearly measurable impact of jitter on the sound is way below the hearing threshold.

Those studies have indeed been done and it seems that our ears are not very sensitive to this specific form of distortion.

However who can hear the difference in a DAC or amplifier that provide -110db vs. -115db of THD+N in a controlled blind test? So these arguments are not just applied to jitter.

The degradation in hearing that comes with age is at least an order of magnitude higher in any case...

Cheers

Thomas

 

[rei_yano]

Quote:

Originally Posted by bigshot /img/forum/go_quote.gif Who needs serious tests when you can get by anecdotal evidence not backed up by rigorous measurements and under dubiously controlled test conditions?! The concept of jitter is designed to differentiate expensive high end audio components from inexpensive mid-range ones that sound exactly the same. It's an inaudible upsell. Stereo salesmen play on people's fears of theoretical problems to get them to spend more on their equipment than they need to. Jitter is a hoodoo. See ya Steve

thank you. its good to hear this kind of information since everywhere else it seems like they just want your wallet to be empty.

 

[thomaspf]

Well, this is clearly overstating the issue. Jitter is not being designed but it is a reality in all synchronously clocked circuits.

So all such circuits do have jitter some just have somewhat less. The real question here is how little is enough to move it below the threshold of hearing and there you have two shools of thought.

When the original work on jitter was conducted people used and verified a theoretical model to analyze how much jitter would introduce a distortion that equals an error in changing the lowest significant bit in a 16 bit sample. It turns out that is a pretty small number <20ps at 20Khz.

The other school of thought is conducting hearing expriments and it turns out people can not discern any difference in sound even if the distortion is a lot more than flipping the lowest significant bit.

The bottom line is that you can buy a low jitter DAC for a lot less money than lets say a high end amplifier with 0.0025% of THD+N across the frequency range.

In audio your signal quality is limited by the weakest link in the chain and that is usually the speakers. If you want to spend a lot of money speakers are a good choice.

Cheers

Thomas