[LAU] Sync and digital audio transport - was Sample rate vs. SNR

Fons Adriaensen fons at linuxaudio.org
Sat Jan 26 12:52:32 UTC 2013

On Fri, Jan 25, 2013 at 05:44:32PM -0800, Len Ovens wrote:
> The AES seems to recommend AES3 (like s/pdif... almost) as the sync signal
> in a digital facility. There are a lot of things that will sync to AES3 or
> s/pdif, but there are a lot that will not too, while they will sync to
> word clock or ADAT... even if they output aes3. Not only that, but word
> clock seems to max out at 48k for many units while those that do lock to
> AES3 will lock at higher clock rates. (The D1010 for example has word
> clock in but only locks to 48K yet will lock to 96k AES3 or s/pdif) The
> few master clocks I have seen (not that I have looked at a lot either) all
> seem to output word clock only. What is standard practice for master
> clocking?

Word clock only at 48 kHz seems to be a limitation of some particular 
equipment. It's not clear to me why things should be that way, I can't
see a good technical reason for it.

AES3 inputs (at least the balanced ones) are rare or semi-pro equipment
anyway. Those that have them will be capable of syncing to them, the
inputs would be useless otherwise. Syncing to AES may have some advantages
as you sync to the data clock which has a much higher frequency than the
sample rate. The inverse, regenerating a higher frequency from a lower one,
is usually a bad idea, phase noise is mutiplied by the square of the ratio.

I don't think there is a 'standard practice' for audio clock distribution,
it will very much depend on the nature of the installation. If there's
anything video in house as well the primary source is likely to be video,
with audio derived from it. I a large facility with many studios the 
system will be layered - a GPS driven master with an atomic standard as
backup driving a second level master generator in each studio. In such 
a case the signal between the primary and secondaries could be video or
just 1, 5, or 10 MHz if no video sync is needed. 

A lot of master clocks only provide word clock, these tend to be the
'audiophile' ones used to 'make the sound more transparent' or some such
nonsense. Pro equipment will always have AES3 outputs in addition to word

> Question two: I read a comment that a studio ( I am not sure if this was a
> recording studio or broadcast but got the idea it was recording) had
> removed all their MADI equipment and replaced it with aes3 as this allowed
> them to set up physical patch bays per channel pair. Seems like a lot of
> work and extra wire compared to the benefits. Does this make sense? or
> these guys just odd balls?

A few years ago that could have made sense as there were few MADI routers.
But e.g. <http://www.directout.eu/en/products/m.1k2.html> will allow per
channel routing between 16 MADI ins and outs. Of course a wall of AES3
connectors looks cool... In practice the choice may be driven by any
existing wiring and in-house standards if e.g. a studio is upgraded.

> I thought it odd that MADI cards were so expensive when AES3 equipment is
> so cheap... but I have since noticed that a lot of AES3 stuff is called
> "s/pdif professional" and does not use balanced lines. I guess the s/pdif
> standard is free to use?

Probably yes, AES3 (known as AES/EBU here) surely is free. AES 3 allows
unbalanced connections using coax and BNC, this may be a good option for 
installations that have existing video wiring. But if the connector is
XLR the signal MUST be transformer balanced.

S/PDIF pro is a variation of the consumer S/PDIF format, some of the
extra bits take another meaning. This is also the AES3 format.



A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)

More information about the Linux-audio-user mailing list