DIY Atomic Clock

[Hammer8]

Hi, for fun I purchased an used SRS PRS10 10Mhz Rubidium clock and would like to connect that to a TEAC 503 DAC. The PRS10 seems to be the cheapest 10Mhz clock out there and luckily, it outputs a 50ohm signal and that is what the TEAC accepts.

Has anyone had any experience with the PRS10? I still need to get a power supply, heatsink and case for it. Any words of wisdom on what parts I should use and how to hook everything up?

Thanks! Hammer8

 

[billshurv]

You do realise that the atomic clocks are often don't have good short term (jitter) performance and are generally used as a long term reference to the low jitter clocks?

 

[Hammer8]

You do realise that the atomic clocks are often don't have good short term (jitter) performance and are generally used as a long term reference to the low jitter clocks?

I read that somewhere as well, but srs also makes a clock made for audio, the PERF10, with is also based on the PRS10 I purchased and so I figured it must work! The PERF10 was recently reviewed on ultra audio which is how I got the idea for a diy unit.

 

[billshurv]

"condition" that was the word I was looking for. You need to use the atomic reference to condition the (preferably oven controlled) local oscillator.

 

[Ron E]

"condition" that was the word I was looking for. You need to use the atomic reference to condition the (preferably oven controlled) local oscillator.

To save others the search, that is exactly what the PRS10 does.
http://www.thinksrs.com/downloads/PDFs/Manuals/PRS10m.pdf

 

[DF96]

There is no point whatsoever in using an atomic clock for digital audio - except for impressing the ignorant. The very act of putting a crystal clock in a phase lock loop to adjust its long term frequency stability will degrade its short term frequency stability (jitter) so the best thing to do is buy or build a decent crystal oscillator and use that. Or buy an 'atomic locked' oscillator and remove all the locking components.

 

[billshurv]

To save others the search, that is exactly what the PRS10 does.
http://www.thinksrs.com/downloads/PDFs/Manuals/PRS10m.pdf

OK that is a lot better than I thought, as designed for cellular, which has very high short term stability requirements, hence the Oven controlled oscillator. But that is to keep carriers in place on narrowband systems in the GHz area and been 15 years so cannot remember the exact requirements.

So it would work as a 10MHz source with low jitter, but like DF96 said, doubtful benefit other than cool factor and a good stand alone OXCO will still be better.

 

[Hammer8]

There is no point whatsoever in using an atomic clock for digital audio - except for impressing the ignorant. The very act of putting a crystal clock in a phase lock loop to adjust its long term frequency stability will degrade its short term frequency stability (jitter) so the best thing to do is buy or build a decent crystal oscillator and use that. Or buy an 'atomic locked' oscillator and remove all the locking components.

Thanks for the advice - wouldn't one think the PRS10 will be better than the standard crystal in the Teac? In their Esoteric line, they sell separate clocks which I think is also based on the PRS10.

Also, does anyone know why some clocks output/inputs are 50ohm and some 75ohm?

For example, my Teac 503 is 50ohm and Teac's own Esoteric line and most other DACs (I think) accept 75ohm signals.

Thank you, Hammer8

 

[jean-paul]

Consumer grade SPDIF is 75 Ohm, clocks and their signal distribution are 50 Ohm in general. Different stuff.

 

[modmix]

wouldn't one think the PRS10 will be better than the standard crystal in the Teac?

Wouldn't surprised

A while ago I've done some experiments with a PRS10 based DIY clock feeding a Mutec MC-3+. Intriguing.
Much better than a Antelope clock based on LCR-900.
Stability and Noise Performance of Various Rubidium Standards has some measurments.
Experiments using OCXOs with even better phase noise gave better results.
KS-24361 is currently the best I ever had.

Also, does anyone know why some clocks output/inputs are 50ohm and some 75ohm?

Clock specs rate signal level with a specific load (most often 50ohm). That does not necessarily mean that the clock output impedance is 50ohm.
Much more important is to avoid reflections at the cable end - means: cable impedance should match input impedance as close as possible.
BTW: RG400 is a good 50ohm cable for this application.

hth
Ulli

 

[DF96]

wouldn't one think the PRS10 will be better than the standard crystal in the Teac?

Using a good quality crystal is an excellent idea, as the crystal is one of the main things determining jitter. Just don't bother to lock it to an atom, or GPS or anything else. The extra circuitry to add the ability to lock will degrade the oscillator, so only worth doing when long term stability is needed. Why degrade something important (jitter) in order to improve something unimportant (long term stability)?

 

[Monte McGuire]

Why is 10MHz supposed to do anything useful with digital audio? Unless you're using it with some fractional N synthesizer, it isn't a multiple of any modern sample rate.