Digital Drive

Upsamplers, DACs, jitter, shakes and analogue withdrawals, this is it.

Return to Digital Drive


Message Sort: Post Order or Asylum Reverse Threaded

USB clocking for 44.1k?

129.21.194.20

Posted on March 2, 2017 at 17:43:49
dave789
Audiophile

Posts: 559
Joined: September 21, 2001
I like the idea of async USB, since the hi-fi DACs clock is usually better than a PCs clock.

However, it seems that USB chips use 12MHz clock, which is not an integer multiple of 44.1kHz.

So a 44.1kHz digital audio signal goes through USB (not S/PDIF, nor I2S), will there be some artifacts like those from asynchronous sample rate conversion?

 

Hide full thread outline!
    ...
RE: USB clocking for 44.1k?, posted on March 2, 2017 at 22:35:31
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

The best way of thinking about the whole situation for USB Audio is that the USB System is a conveyor belt that sends one Bit-bucket every millisecond (USB 2.0 Full Speed) or 125uS (USB 2.0 High Speed).

Now that Bit Bucket comes with an address where it goes to and 80% of all buckets are reserved for "isochronous" transfers, in other words audio & video. A Bit Bucket may be empty, may be halve full or may ven only contain 1 Bit of "payload", whatever is needed to get the correct amount of bits across the conveyor.

The USB conveyor belt runs on it's own clocks and timing. When sending data to a USB Audio Device the buckets are filled with a bunch of bits every time the USB Audio devices requests a transfer. The USB Audio Device contains a data buffer and requests enough bits in enough buckets from the PC side to keep the buffer from running empty.

On the USB Audio Device the local audio clock will then be used to transfer the data from the buffer to the DAC. So in this case you have two clocks, asynchronous and (with perfect systems) without cross contamination or interference.

The USB clock matters to the USB system (too much jitter and data corruption may happen or the receiving system works harder), the audio clock matters to the Audio system (too much jitter and audio performance is degraded).

In practice beat notes between these two clocks may appear and power supply contamination may cause cross modulation, non-zero ground impedance may cause cross modulation and so on. So for audio devices there are plenty of chances for things to wrong, even if all bits arrive safely (which should not be taken for granted).

What all the various USB repeaters (I prefer this term over the marketing stuff) do is to take a USB signal that has been degraded, re-synchronise the clocks, restore the waveform and otherwise make the signal "like new" and all prior degradation is cancelled. As we are dealing with "digital" signals, such a process can be 100% lossless and can be repeated many times (in theory ad infinitum).

For a fun demo we take 4pcs of fully USB standard compliant USB A-B cables, plug an iPURIFIER 2 "A-Type" at the "B" end of each cable (and one iPURIFIER 2 "B-Type" at the DAC) and stream 768kHz/32Bit Audio perfectly with zero dropped packets...

Remove just one of the iPURIFIER 2 and the systems drops back to USB 2.0 full speed (12Mbps) and thus 96kHz maximum sample rate. Remove more and even at full speed we see dropped data. Remove all and the system does not connect even in USB 2.0 Full Speed mode.

Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: USB clocking for 44.1k?, posted on March 4, 2017 at 08:47:39
Bit-Bucket...thats pretty catchy.

Perhaps a iFi Audio Bit-Bucket is coming to market?

Anyways, I think people have too many clocks in their chain of Buckets. Is it possible to slave the USB Bit-Bucket clock and the Audio Bit-Bucket clock to 'one' master clock?

 

RE: USB clocking for 44.1k?, posted on March 4, 2017 at 10:03:09
dave789
Audiophile

Posts: 559
Joined: September 21, 2001
"Is it possible to slave the USB Bit-Bucket clock and the Audio Bit-Bucket clock to 'one' master clock?"

Good question. A DAC that has two clocks for 44.1k times integer and for 48k times integer, respectively, had a third 12MHz clock near the USB chip.

 

RE: USB clocking for 44.1k?, posted on March 4, 2017 at 11:12:25
A DAC....you saw had those, not every Dac has the same number of clocks or the same Hz.

 

two clocks +1, posted on March 4, 2017 at 11:18:05
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
All decent usb cards do, with I2S output to the dac. The quality of the 44.1x and 48x clocks count a lot, nevermind all these gold coloured Chinese XOs that they sell at a premium price.

 

Just two clock, posted on March 4, 2017 at 14:18:34
As far as I can tell my gear only has two clocks.

1 on the computer side and one on the dac side.

Although the Audiogd Master 7 and many other Dacs have two clocks [or more inside]. As for the Audiogd one clock is at the Amanero module and another at the DSP module. If I use USB in I usually bypass the DSP clock.

 

RE: USB clocking for 44.1k?, posted on March 5, 2017 at 20:30:32
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> Is it possible to slave the USB Bit-Bucket clock and the Audio
> Bit-Bucket clock to 'one' master clock?

That is called "Asynchronous USB Audio".

It actually does not directly slave the clocks (that is not possible in that sense), but it makes sure the audio clocks determine audio timing.

Modern mixed signal system routinely have multiple clocks and the wide variety of clocks needed does not make synchronisation across all clocks feasible.

The important part then becomes to make sure each individual clock is correct for the job and performs adequately and that the audio conversion uses one high quality clock after a short memory buffer (to handle any data flow control).

Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: USB clocking for 44.1k?, posted on March 6, 2017 at 05:12:47



Modern mixed signal system routinely have multiple clocks and the wide variety of clocks needed does not make synchronisation across all clocks feasible.


Interesting, it was my understanding that recording studios routinely used one Master clock to sync everything.

Maybe I misunderstood.

 

RE: USB clocking for 44.1k?, posted on March 6, 2017 at 06:08:05



For some reason I was thinking someone could use something like this Black Lion Micro Clock MkIII XB, to sync up their digital from end [Rpi for exampe, their USB card and their Dac's internal DSP processor...or everything else. Provided, of course, that these items had the proper connection....likely a DIY project, but should be doable.

Not Feasible??

 

RE: USB clocking for 44.1k?, posted on March 6, 2017 at 08:21:56
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> Interesting, it was my understanding that recording studios routinely
> used one Master clock to sync everything.
>
> Maybe I misunderstood.

Maybe, instead of trying to understand how such systems work, you merely took away what you wanted and ignored most?

You cannot slave/sync the USB (or Wifi, or Ethernet etc. etc.) Clock directly to studio clocks, for the simple reason that the clock frequency is wrong.

The job of the "House Clock" is to synchronise the audio (and often video clocks) of multiple converters, so that multiple devices do not loose "sync".

You can see the need when handling Video & Audio using separate converters. I think you can also appreciate the need if we (say) use three stereo ADC's to make a six channel recording. Each individual Quarz Crystal in a given oscillator will have a frequency that is just a tiny bit different. So if we wait long enough, they will no longer be synchronised.

So while the need to synchronise clocks is obvious, there is a dirty little secret in that. Most Pro-Audio devices perform notably worse on external clocks, than when running on cheap internal crystals.

And of course on top of that any non-audio clock in the system (e.g. USB, Ethernet etc.) will not be synchronised to the audio clock.

You may find this article useful on the studio perspective:

http://www.soundonsound.com/techniques/does-your-studio-need-digital-master-clock

Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

Ahhhh Great Info THANKS!!!nt, posted on March 6, 2017 at 08:32:26
.

 

RE: USB clocking for 44.1k?, posted on March 6, 2017 at 08:42:21
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> Not Feasible??

Not really.

You can sync the audio clocks, but Async USB operates the audio section from a local clock anyway and the USB clock is needed for the USB operation and not directly linked to the audio side.

It probably is worthwhile for you to truly understand how Asynchronous USB works and what the implications are. This article may be of help:

http://www.edn.com/design/consumer/4376143/Fundamentals-of-USB-Audio

Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

Page processed in 0.025 seconds.