This is way out of my comfort zone, but it seemed intriguing idea to me. Crossover of technology between fields of endeavour is always interesting.
I was at the NAS near Silverstone at the weekend and I got talking to a telecoms engineer. He wanted to know if anyone had used a GPS ( Don't laugh!) derived master clock in a disc player.
After all, quartz is by no means perfect and the GPS has incredible timing stability and accuracy. He said that all telecom networks are timed this way. I found AD do something that could be within reach of the DIY-er, but I'm too ill educated to do something with it or assess its worth. Over to you really clever guys....
http://www.analog.com/en/press-release/6_11_09_Clock_IC_First_to_Harness_GPS/press.html
I was at the NAS near Silverstone at the weekend and I got talking to a telecoms engineer. He wanted to know if anyone had used a GPS ( Don't laugh!) derived master clock in a disc player.
After all, quartz is by no means perfect and the GPS has incredible timing stability and accuracy. He said that all telecom networks are timed this way. I found AD do something that could be within reach of the DIY-er, but I'm too ill educated to do something with it or assess its worth. Over to you really clever guys....
http://www.analog.com/en/press-release/6_11_09_Clock_IC_First_to_Harness_GPS/press.html
That's right, GPS master clock is an alternative to Rubidium clock in telecom (namely SONET/SDH) world. Accuracy is referred to as Stratum 1 (best) to Stratum 16 (free running).
However, absolute accuracy is not an issue in audio, and I don't know what are the jitter specs of a GPS clock. It is PLL-based I think, so it could be nothing very special.
However, absolute accuracy is not an issue in audio, and I don't know what are the jitter specs of a GPS clock. It is PLL-based I think, so it could be nothing very special.
What matters is the phase noise spectrum: the jitter is obtained by integrating over the frequency range of interest. For digital audio reproduction, we care about jitter down to 10Hz or so, whereas most jitter specs are integrated from 12kHz-20MHz. Interestingly the AD9548 datasheet has data down to 100Hz.
No matter how clean the GPS clock is, it still needs to be recovered at the receiver. This is done by a PLL, which by its nature sets a limit on the phase noise level achievable.
Look at the spectrum charts on pages 18 and 19. The performance @ 100Hz is somewhere from -90 to -100 dBc/Hz. Now look at Figure 13 on page 20. The gray line is the phase noise of an OCXO, which is basically a crystal in a fancy climate-controlled oven. That's pretty much the best a crystal can do. OCXOs are pricey though, so compare to -110 dBc/Hz for an off-the-shelf VCXO:
http://www.crystekcrystals.com/crystal/spec-sheets/vcxo/CVHD-930.pdf
No matter how clean the GPS clock is, it still needs to be recovered at the receiver. This is done by a PLL, which by its nature sets a limit on the phase noise level achievable.
Look at the spectrum charts on pages 18 and 19. The performance @ 100Hz is somewhere from -90 to -100 dBc/Hz. Now look at Figure 13 on page 20. The gray line is the phase noise of an OCXO, which is basically a crystal in a fancy climate-controlled oven. That's pretty much the best a crystal can do. OCXOs are pricey though, so compare to -110 dBc/Hz for an off-the-shelf VCXO:
http://www.crystekcrystals.com/crystal/spec-sheets/vcxo/CVHD-930.pdf
Low phase noise OCXO oscillators have really come down in price over the last few years. Previously quite expensive GPS disciplined oscillators are now available at a reasonable price. There is quite a detailed article here on GPS master clocks that you may find useful:
Master Clocks: A Guide to Master-Slave Time Clock Systems
Master Clocks: A Guide to Master-Slave Time Clock Systems
For digital audio what is needed is good short term stability i.e. low jitter. Other uses of clocks, including some telecommunications, need good long term stability and may be less concerned about short term stability as long as it isn't too bad.
Take a good crystal clock. Put it in a phase locked loop to lock it to an external source (e.g. GPS, rubidium). Then you have greatly improved the long term stability, but degraded the short term stability. This is because the very act of adding the components to allow the frequency to be adjusted (needed for the PLL) degrades stability.
So, an excellent idea for marketing; a poor idea for genuine audio engineering. Hence you will see people proposing it from time to time.
Take a good crystal clock. Put it in a phase locked loop to lock it to an external source (e.g. GPS, rubidium). Then you have greatly improved the long term stability, but degraded the short term stability. This is because the very act of adding the components to allow the frequency to be adjusted (needed for the PLL) degrades stability.
So, an excellent idea for marketing; a poor idea for genuine audio engineering. Hence you will see people proposing it from time to time.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.