Calling all clocks!

[Nicholas Bedworth]

Some new phase noise measurment gear finally arrived, so we are willing to take a look at any and all digital audio clocks that members would like to send in. Please PM for the details... and see example below. Now you'll know... What we're finding is that some clocks have good high-frequency phase noise, which is relatively easy to do, but the low frequency phase noise isn't so hot. The lower offset regime is apparently critical for realism, presence, soundstaging and the like, but we don't have any engineering model yet for why this would be so.

The clock needs to have a 256 Fs output (meaning, typically, a bit clock or "Super Clock" output. The phase noise device can only go down to about 1 MHz, which is substantially above even a 384 kHz word clock.

If you send me the model of your product, we can double-check to see if our present gear can handle it.

And in a few weeks, we should be able to do elaborate jitter measurements as well.

 

[amirm]

Congratulations Nick. Phase measurement devices can indeed to way lower than audio measurement devices can. The problem with them is that they need a clock to measure. Embedded clock in a S/PDIF stream for example that is changing with data becomes impossible to measure with them. Have you found a work-around here or are you just aiming for measurement of independent clocks?

 

[Nicholas Bedworth]

Thanks! The digitization capability of the device is capable of grabbing the raw data that we can then run through elaborate software-based clock extraction technology to get all the jitter information. That isn't ready yet, but, famous last words, "it should work."

But the phase noise measurements we're getting with John Miles' Time Pod protoype are already as good as, or better, that the Symmetricom 5120a with the internal clock option. The Time Pod uses an external PC for user interface, data processing, etc., so that keeps the cost down, and it also allows one to do all kinds of things with the raw data, such as what we are contemplating for the jitter analysis.

 

[DonH50]

For random noise, close-in phase noise (LF noise) is indeed the issue. Especially since the slopes of the various 1/f^x products go to infinity at d.c...

Most digital measurement systems (e.g. JBERTs) extract and follow the clock to measure data jitter, or (alternatively) use a reference clock input to look at jitter/noise on the data clock.

Given the thread a while back on how small a time difference people can localize, I would expect LF noise/jitter to impact imaging and smear edges, dulling the sound and affecting the perceived sound stage (e.g. depth).

 

[ar-t]

But the phase noise measurements we're getting with John Miles' Time Pod protoype are already as good as, or better, that the Symmetricom 5120a with the internal clock option. The Time Pod uses an external PC for user interface, data processing, etc., so that keeps the cost down, and it also allows one to do all kinds of things with the raw data, such as what we are contemplating for the jitter analysis.

How much does it keep the cost down? Does it have the same "Can't measure a square wave" problem, the 5120 is supposed to have?

 

[Nicholas Bedworth]

The Time Pod is expected to retail for about $6000, depending upon accessories. The current price of the Symmetricom is $37000, so the delta is $31000. Of course the Time Pod uses a laptop for control, computation and display, whereas the Symmetricom is self-contained. The Time Pod is a frequency-domain analyzer, works down to about 1 MHz. But it's also fundamentally a very high speed A/D converter, and given the raw sample data, one could use time-domain analysis techniques as well.

 

[DonH50]

1 MHz, or 1 mHz?

 

[Nicholas Bedworth]

@DonH50... that's 1 MHz input signal... the measurement offset goes down to 0.1 Hz usually. With the Symmetricom, the lowest input carrier in practice seemed to be more like 2 MHz. Of course most people using this gear are more concerned with GHz...

 

[DonH50]

Got it, thanks. I didn't do a search for the gear... And yeah, one of the problems with microwave test equipment for this is that it rarely extends down to audio frequencies.

 

[ar-t]

Only $6000? Gotta get one of those, assuming it works, as you hope.

Does it need a reference oscillator?

Speaking of which, any chance of using your connections to get us one of those fancy $500 TAD units, as a lab reference?