It has been discussed on various forums here and elsewhere that absolute frequency accuracy is not important in digital audio. What is important is low phase noise in the whole audio band. This parameter is best specified as a spectrum plot and the values are referred to the level of the carrier, hence dBc. Manufacturers specify the phase noise at some deviation from the carrier, like 10 Hz, 100 Hz, 1kHz, etc. If it is not specified, it has a reason 
The phase noise (jitter) is usually worst at closest to the carrier, i. e. at 10 Hz distance. I think -100 dBc @ 10Hz is the minimum value for a DAC clock.
The phase noise (jitter) is usually worst at closest to the carrier, i. e. at 10 Hz distance. I think -100 dBc @ 10Hz is the minimum value for a DAC clock.
I also have looked at using a Rubidium AFS for audio. I think they may have been dismissed as unsuitable too quickly, based on the assumption 'it's designed for accuracy, therefore it must be imprecise'.
The particular model I am looking at, the FEI FE5680A, specifies its phase noise performance (unlike many clocks), and it is not unimpressive. It's at about -110dB at 10Hz, -130dB at 100Hz and plateaus of to about -150dB after 1kHz, all specified at a carrier of 10MHz. See attached plot.
However, I do still have questions about the clock. The output frequency is generated in an interesting way. A VCXO is locked to the divided output of a Rubidium physics package, producing a 50.225MHz clock. This is then fed to a DDS (an Analog Devices AD9830A) which then outputs the frequency desired, 10MHz by default*.
The phase noise plot is specified for an output frequency of 10MHz, would this still hold for an output of 11.2896MHz, the frequency I intend to use it for?
I suspect it will, because 10MHz doesn't seem to be a special number, like a divisor of 50.225MHz.
What will happen when the Rubidium lamp dies?
There are some cheap used RVOs on eBay, and I suspect they're cheap because they are toward the end of their lives. The MTBF of one of these is only a few years of constant use (and I'd guess that they are subjected to constant use). Will they stop outputting a clock, or will they just become less accurate when the VCXO can't lock (and maybe still good for audio purposes)?
If anyone knows, I'd love to be enlightened.....
* The desired output frequency can be programmed via RS232 for an output between about 1.5 and 15MHz. It'd be useful to have such an adaptable clock.
The particular model I am looking at, the FEI FE5680A, specifies its phase noise performance (unlike many clocks), and it is not unimpressive. It's at about -110dB at 10Hz, -130dB at 100Hz and plateaus of to about -150dB after 1kHz, all specified at a carrier of 10MHz. See attached plot.
However, I do still have questions about the clock. The output frequency is generated in an interesting way. A VCXO is locked to the divided output of a Rubidium physics package, producing a 50.225MHz clock. This is then fed to a DDS (an Analog Devices AD9830A) which then outputs the frequency desired, 10MHz by default*.
The phase noise plot is specified for an output frequency of 10MHz, would this still hold for an output of 11.2896MHz, the frequency I intend to use it for?
I suspect it will, because 10MHz doesn't seem to be a special number, like a divisor of 50.225MHz.
What will happen when the Rubidium lamp dies?
There are some cheap used RVOs on eBay, and I suspect they're cheap because they are toward the end of their lives. The MTBF of one of these is only a few years of constant use (and I'd guess that they are subjected to constant use). Will they stop outputting a clock, or will they just become less accurate when the VCXO can't lock (and maybe still good for audio purposes)?
If anyone knows, I'd love to be enlightened.....
* The desired output frequency can be programmed via RS232 for an output between about 1.5 and 15MHz. It'd be useful to have such an adaptable clock.
Attachments
I think I know the answer to your question. I used to work for an institute that provided calibration service, and we had a Rubidium frequency standard. At one point it was not accurate any more, but only by a few ppm error. The Rubidium lamp was defect, but one could not tell it visually, it glowed just like before. I think there was some locked indicator on the unit, I can't remember. After having replaced the lamp, the device worked again up to specs.
Personally, I don't trust in a PLL-based master clock as a low jitter source, be it Rubidium or whatever.
Personally, I don't trust in a PLL-based master clock as a low jitter source, be it Rubidium or whatever.
I know of "audiophille grade" clocks that are PLL. It just need to have 2 time constants - one fast and one slow.
I suppose the exact value of the frequency is not as important (to a few PPM), but when the lamp is really weak, the PLL loop will not be able to lock, so the freqv will just go off the grid, unless is there an electronic switch to cut the output when the lamp signal is too low.
I suppose the exact value of the frequency is not as important (to a few PPM), but when the lamp is really weak, the PLL loop will not be able to lock, so the freqv will just go off the grid, unless is there an electronic switch to cut the output when the lamp signal is too low.
The absolute frequency (the MHz) is not crucial for audio performance, as long as it is within a reasonable range of the exact freq, like 16.934400 MHz.
Therefore many cheaper ref. clocks skip the adjustment process alltogether.
However in my opinion frequency accuracy is a quality factor. If you use a car as an analogy, a leather interior is not strictly necessary in your new Porsche. It will drive the same with chinese plastic seats. However the leather make the driving experience somewhat better.
So the best clocks on the market have both frequency accuracy and low phase noise.
Therefore many cheaper ref. clocks skip the adjustment process alltogether.
However in my opinion frequency accuracy is a quality factor. If you use a car as an analogy, a leather interior is not strictly necessary in your new Porsche. It will drive the same with chinese plastic seats. However the leather make the driving experience somewhat better.
So the best clocks on the market have both frequency accuracy and low phase noise.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.