Re: [Jack-Devel] Ethernet-based audio interface using (net)jack idea

PrevNext  Index
DateWed, 17 Aug 2011 10:08:15 +0100
From John Rigg <[hidden] at jrigg dot co dot uk>
Tojack-devel <[hidden] at lists dot jackaudio dot org>
In-Reply-ToFons Adriaensen Re: [Jack-Devel] Ethernet-based audio interface using (net)jack idea
Follow-UpJohn Rigg Re: [Jack-Devel] Ethernet-based audio interface using (net)jack idea
On Tue, Aug 16, 2011 at 06:27:03PM +0000, Fons Adriaensen wrote:
> I seriously doubt if you can get a sampling clock with the required
> low jitter using the FPGA's digital PLLs. The next thing to look at
> would be an external VCXO module to generate the master clock, with
> the loop closed by digital filtering in the FPGA followed by an
> analog lowpass driving the VCXO. Lock range for VCXOs is small but
> it should be enough to cover typical sound card frequency

This is a very important point. Quality of the clock signal at
the ADC and DAC clock pins is more critical than elsewhere in the
system by several orders of magnitude. For high audio quality every
device with converters in needs a good local clock recovery circuit
to remove jitter from the distributed system clock. 

Ideally a two stage PLL would be used, with a VCXO typically used
for the second stage. Good VCXOs aren't cheap, and it's not unusual
to use two to cover the common sample rates (one for 44.1kHz and
multiples, and another for 48kHz and multiples).

There are lower cost IC solutions available like the TC DICE chip,
which has a very effective low jitter PLL. It's one of the cheapest
ways I've seen to get clock recovery good enough for high end
professional audio, even if you don't use its other features like
ADAT in/out.

There's a good paper on PLL and clock basics here:
http://www.grimmaudio.com/whitepapers

John
PrevNext  Index

1313571913.23988_0.ltw:2,a <20110817090815.GA2554 at localhost0 dot localdomain>