Since no one is replying I will jump in. I guess the issue is that you don't have a lot of choice of frequency. What frequency range are these units generally operating at? Possibly you could use them to drive a DDS chip (Direct Digital Synthesis) to create a different frequency. You will still have system jitter issues, so I am not sure if their will be overall benefit. If you do buy one and use it, let us know how it goes, I am curious.
Alvaius
Alvaius
Yes, the issue of a useful frequency is a big problem.
Are there any DDS chips that have jitter tolerances worth looking at? I figure benchmark for a quartz oscillator is 1ps RMS and for an OXCO it should be better. Most of the synthesizers I've seen have been more in the 50ps range.
Something to chew on though: my DAC board has an asyncronous reclocker and a circuit to generate the audio clocks (MCK, SCK, LRCK) from the master oscillator. Since the ASRC is happy to clock the data out at any reasonable rate, there is no reason not to use a nonstandard rate, provided the DAC (AD1853) doesn't have a heart attack. I don't see why it should, but...maybe ask Analog?
Another route to go (that I am definitely going down) is ovenizing an integrated oscillator, or building a discrete oscillator and ovenizing the crystal.
Are there any DDS chips that have jitter tolerances worth looking at? I figure benchmark for a quartz oscillator is 1ps RMS and for an OXCO it should be better. Most of the synthesizers I've seen have been more in the 50ps range.
Something to chew on though: my DAC board has an asyncronous reclocker and a circuit to generate the audio clocks (MCK, SCK, LRCK) from the master oscillator. Since the ASRC is happy to clock the data out at any reasonable rate, there is no reason not to use a nonstandard rate, provided the DAC (AD1853) doesn't have a heart attack. I don't see why it should, but...maybe ask Analog?
Another route to go (that I am definitely going down) is ovenizing an integrated oscillator, or building a discrete oscillator and ovenizing the crystal.
PLL ideas
Hi Mark,
As alvaius has noted the output frequency is a problem. Most frequency standards output 1, 5 or 10 MHz. The jitter associated with a DDS chip is too high. A much better approach would be to design a frequency synthesiser based on a phase locked loop (PLL).
The output of the standard can be divided down to a much lower frequency e.g. 1 Hz and then used as a controlling input to a PLL. See http://www.rt66.com/~shera/ for an interesting article on the problems of PLL design. The standard reference work on PLLs is Phaselock Techniques by Floyd Gardner published by Wiley Interscience. More modern books are also useful.
This is not an easy task and the problem will remain of obtaining a high specification voltage controlled crystal oscillator (VCXO). For best stability this could be a temperature compensated TCVCXO or ovenized type OCVCXO. For the frequencies associated with CD replay these will have be custom ordered and will be expensive.
At least one crystal manufacturer does VCXOs as stardard or near-standard parts. Forssell and King describe a suitable 12.288 MHz VCXO in their Audio Electronics Five/1997 article 'Jitter Detector/Modulator Circuit Design'. See www.audioxpress.com for back issues. It was manufactured by Champion Technologies, distributed by Brookdale Electronic and the part number is K1525AE.
For improved stability the VCXO could be temperature controlled as tiroth has noted or indeed the whole PLL could be temperature controlled. Radio amateurs have been homebrewing OXCOs for years and back issues of the ARRL Handbook in the US and the Radio Communication Handbook in the UK will provide some ideas.
James
Hi Mark,
As alvaius has noted the output frequency is a problem. Most frequency standards output 1, 5 or 10 MHz. The jitter associated with a DDS chip is too high. A much better approach would be to design a frequency synthesiser based on a phase locked loop (PLL).
The output of the standard can be divided down to a much lower frequency e.g. 1 Hz and then used as a controlling input to a PLL. See http://www.rt66.com/~shera/ for an interesting article on the problems of PLL design. The standard reference work on PLLs is Phaselock Techniques by Floyd Gardner published by Wiley Interscience. More modern books are also useful.
This is not an easy task and the problem will remain of obtaining a high specification voltage controlled crystal oscillator (VCXO). For best stability this could be a temperature compensated TCVCXO or ovenized type OCVCXO. For the frequencies associated with CD replay these will have be custom ordered and will be expensive.
At least one crystal manufacturer does VCXOs as stardard or near-standard parts. Forssell and King describe a suitable 12.288 MHz VCXO in their Audio Electronics Five/1997 article 'Jitter Detector/Modulator Circuit Design'. See www.audioxpress.com for back issues. It was manufactured by Champion Technologies, distributed by Brookdale Electronic and the part number is K1525AE.
For improved stability the VCXO could be temperature controlled as tiroth has noted or indeed the whole PLL could be temperature controlled. Radio amateurs have been homebrewing OXCOs for years and back issues of the ARRL Handbook in the US and the Radio Communication Handbook in the UK will provide some ideas.
James
James,
Does a PLL design exist that can match the jitter performance of a simple quartz oscillator? Even Fox's JITO-2 PLL-based clock oscillators are over an order of magnitude worse than basic Valpey-Fisher quartz oscillators that are available for $1.30 each/100. It just seems like a lot of work to use a complicated circuit with a stable reference and then end up with something worse than your $1 crystal.
There are a number of companies that will custom-make OXCO with excellant jitter spec, but these are apt to be quite expensive.
Does a PLL design exist that can match the jitter performance of a simple quartz oscillator? Even Fox's JITO-2 PLL-based clock oscillators are over an order of magnitude worse than basic Valpey-Fisher quartz oscillators that are available for $1.30 each/100. It just seems like a lot of work to use a complicated circuit with a stable reference and then end up with something worse than your $1 crystal.
There are a number of companies that will custom-make OXCO with excellant jitter spec, but these are apt to be quite expensive.
Tiroth,
there are a large number of variables to trade off in any PLL design. It is possible to achieve 1ps RMS jitter figures from a suitably designed PLL.
To back up a bit, yes it is going to be possible to generate a clock for a CD transport or DAC using the rubidium oscillator as a master reference. Is is going to be easy - no. Its going to take blood sweat and tears and thats for someone who has done some oscillator or PLL design before. It going to require some time, some calculation and access to lots of fancy test equipment. Care will be required in layout, in power supply, in component selection etc.
A rubidium reference will give you long term frequency stability - if you periodically calibrate it against a known standard - cesium or passive/active maser. Is an absolute frequency accuracy of 1 part in 10^9 or 10^10 range important for CD replay - I don't think so. As many have highlighted the jitter / phase noise specs of the CD clock are much more important than the absolute frequency.
Is all of the above design and development work worth doing. Yes if designs like this interest you in themselves and you get a kick out of PLL design. At the end will you achieve better playback from CD than a good standard oscillator ( discrete, canned or ovenised ) ? Probably not. It depends on how much time you have available, I suspect.
James
there are a large number of variables to trade off in any PLL design. It is possible to achieve 1ps RMS jitter figures from a suitably designed PLL.
To back up a bit, yes it is going to be possible to generate a clock for a CD transport or DAC using the rubidium oscillator as a master reference. Is is going to be easy - no. Its going to take blood sweat and tears and thats for someone who has done some oscillator or PLL design before. It going to require some time, some calculation and access to lots of fancy test equipment. Care will be required in layout, in power supply, in component selection etc.
A rubidium reference will give you long term frequency stability - if you periodically calibrate it against a known standard - cesium or passive/active maser. Is an absolute frequency accuracy of 1 part in 10^9 or 10^10 range important for CD replay - I don't think so. As many have highlighted the jitter / phase noise specs of the CD clock are much more important than the absolute frequency.
Is all of the above design and development work worth doing. Yes if designs like this interest you in themselves and you get a kick out of PLL design. At the end will you achieve better playback from CD than a good standard oscillator ( discrete, canned or ovenised ) ? Probably not. It depends on how much time you have available, I suspect.
James
The most important feature for a clock in digital audio is the lowest jitter. At present time, best oscillators are built with high quality quartz crystals, specially designed for this purpose. Rubidium oscillators are more stable over long periods, but exhibit higher jitter, and are not really suitable.
Regards, Pierre Lacombe.
Regards, Pierre Lacombe.
TCXOs and OCXOs are designed for long term stability, not short terms jitter performance. An OCXO will actually have higher jitter due to the added noise of the higher heat.
I have played around with cooling PLL circuits with good results in increasing jitter, but they will never be as good as a good crystal circuit.
Nemestra, are you sure that the DDS will be worse in terms of jitter than a PLL. That was true in the past, but with the new high speed products with 14 bit DACS, I am not sure. Just checked the AD9954 data sheet. It can achieve up to -125db phase noise. That sounds better to me than most PLLs, but I am going off memory. What are your thoughts on this?
FYI, for the rest reading, a good quartz oscillator can have phase noise down in the -150db or more range.
One avenue I want to explore is a SAW stabilized oscillator circuit. These can achieve phenominally low jitter performance and may not be overly expensive.
Alvaius
I have played around with cooling PLL circuits with good results in increasing jitter, but they will never be as good as a good crystal circuit.
Nemestra, are you sure that the DDS will be worse in terms of jitter than a PLL. That was true in the past, but with the new high speed products with 14 bit DACS, I am not sure. Just checked the AD9954 data sheet. It can achieve up to -125db phase noise. That sounds better to me than most PLLs, but I am going off memory. What are your thoughts on this?
FYI, for the rest reading, a good quartz oscillator can have phase noise down in the -150db or more range.
One avenue I want to explore is a SAW stabilized oscillator circuit. These can achieve phenominally low jitter performance and may not be overly expensive.
Alvaius
oscillators
Hello Alvaius,
it is true that TCXO's are designed for long term stability but they can have very good phase noise performance. IMHO more than adaquate for clocks in audio equipment.
TCXO's are used in gps units which have very tight phase noise reqirements.
As the fundamental frequency increases the close in phase noise 1 Hz to 100 Hz increases this is because the Q of the quartz xtal decreases. You quoted -150 db at what frequency away from the carrier was this measured at.
Blair
Hello Alvaius,
it is true that TCXO's are designed for long term stability but they can have very good phase noise performance. IMHO more than adaquate for clocks in audio equipment.
TCXO's are used in gps units which have very tight phase noise reqirements.
As the fundamental frequency increases the close in phase noise 1 Hz to 100 Hz increases this is because the Q of the quartz xtal decreases. You quoted -150 db at what frequency away from the carrier was this measured at.
Blair
Osillators
1.TCXO and OCXO are NOT specialy designed for long term stability. The only diference is in working temperature range of the oscillator , frequency in temperature (the diference is in temperature compensation).
2.Long term stability depends of crystal unit (angle of cutting , working overtone , diametar of crystal plate , orientation....) , and the main thing AGING.
3.I think that best thing will be to use "0"angle cut crystal unit without any temperature conpenzation , ofcourse with god oscillator unit , low phase noise design (with special power supply).
REGARDS
Sasa
1.TCXO and OCXO are NOT specialy designed for long term stability. The only diference is in working temperature range of the oscillator , frequency in temperature (the diference is in temperature compensation).
2.Long term stability depends of crystal unit (angle of cutting , working overtone , diametar of crystal plate , orientation....) , and the main thing AGING.
3.I think that best thing will be to use "0"angle cut crystal unit without any temperature conpenzation , ofcourse with god oscillator unit , low phase noise design (with special power supply).
REGARDS
Sasa
By long term, I was referring to frequency accuracy over a period of time as opposed to phase noise over a short period of time. TCXO can not be guarenteed to have any better phase noise than a standard oscillator set up. The ones in GPS are low phase noise, but that is because the circuits and crystals are designed as such. SAW stabilized oscillator circuits have the lowest phase noise I have seen personally.
Alvaius
Alvaius
Also, don't forget that Rubidium doesn't spit out 10MHz directly! These units already contain a PLL, locked to the RF signal generated by the Rubidium atomic resonance. So, in fact, the Rb unit output is coming from a PLL anyway, with all it's associated phase noise. The only real benefit, as mentioned before, is the precise long-term frequency stability, which is of little use in audio. I rather doubt you're going to hear jitter at 0.001 Hz. The same goes for TCXO and OCXOs... there's no point to stabilizing the frequency as such - efforts to do so will generally result in a compromise in other performance aspects of the oscillator, such as phase noise. So, IMHO, a good, plain XO circuit seems just the thing for audio applications. Then again, wouldn't it be nice if we could alter the PLL inside the Rb unit to directly derive the frequency we want? Now <i>that</i> might be a worthwhile project...
WRT SAW stabilized oscillators, I wonder somewhat about how susceptible such a circuit might be to microphonic effects? Then again, a crystal is really a mechanical device too...
WRT SAW stabilized oscillators, I wonder somewhat about how susceptible such a circuit might be to microphonic effects? Then again, a crystal is really a mechanical device too...
We've been using and selling a 10MHz Rubidium oscillator (called Chronos) for the past 2 years (now discontinued, all 75 pieces sold).
It was designed as a master reference timing source for digital audio.
It can be interfaced to a dCS 992 master clock.
The dCS 992 has a reference input which can accept this 10MHz sinewave as it has a 'software' DDS.
The DDS controls 'computer temperature calibrated' VCXOs. The PLL has user selectable time constants.
Output frequencies are 44.1 or 48kHz and multiples.
The result is very low jitter and a huge sound improvement.
The only slight drawback is the output stability is governed by the VCXOs, but we typically get +/- 0.002ppm.
We can also recalibrate the VCXOs (whilst in the unit) automatically by using the 10MHz source as a reference, you will need an oven and a day or two!
For Chronos go to: http://www.timelord.co.jp/consumer-audio/audiophile.html
under Product Search click Chronos (sorry, only in Japanese).
For dCS 992 go to: http://www.dcsltd.co.uk/clock.htm
It was designed as a master reference timing source for digital audio.
It can be interfaced to a dCS 992 master clock.
The dCS 992 has a reference input which can accept this 10MHz sinewave as it has a 'software' DDS.
The DDS controls 'computer temperature calibrated' VCXOs. The PLL has user selectable time constants.
Output frequencies are 44.1 or 48kHz and multiples.
The result is very low jitter and a huge sound improvement.
The only slight drawback is the output stability is governed by the VCXOs, but we typically get +/- 0.002ppm.
We can also recalibrate the VCXOs (whilst in the unit) automatically by using the 10MHz source as a reference, you will need an oven and a day or two!
For Chronos go to: http://www.timelord.co.jp/consumer-audio/audiophile.html
under Product Search click Chronos (sorry, only in Japanese).
For dCS 992 go to: http://www.dcsltd.co.uk/clock.htm
Attachments
![chronos[1].gif](/community/data/attachments/7/7045-539f59770a8a6cc71c2d9d4632fd3109.jpg)
Wow, that canned Rb XO draws 1A @ 5V when warmed up...
We have a Rb standard at work. Like others have said, there's a lamp and PLL in there to generate the 10 MHz from the atomic transition.
YIG would be nice, but they don't come in the right frequency range.
What I'd really like to see is someone characterize the 1-1000 Hz region phase noise of a bunch of XOs, VXCOs, OXCOs, and modern PLL and DDS chips. And something like an HP 8662, for kicks.
We have a Rb standard at work. Like others have said, there's a lamp and PLL in there to generate the 10 MHz from the atomic transition.
YIG would be nice, but they don't come in the right frequency range.
What I'd really like to see is someone characterize the 1-1000 Hz region phase noise of a bunch of XOs, VXCOs, OXCOs, and modern PLL and DDS chips. And something like an HP 8662, for kicks.
Lowest phase noise references that I know of are the sapphire whispering gallery mode resonators, run at cryogenic temperatures, with a superconducting (high purity Niobium) enclosure. These typically operate at several GHz.
The Q values of these structures can run up to 10^10, and consequently the circulating powers are enormous.
See for example http://www.npl.co.uk/server.php?show=ConWebDoc.974
or
http://www.psi.com.au/Pages/LibraryPublished/FCS2004_-_Review_of_Sapphire_Oscillators.pdf
The Q values of these structures can run up to 10^10, and consequently the circulating powers are enormous.
See for example http://www.npl.co.uk/server.php?show=ConWebDoc.974
or
http://www.psi.com.au/Pages/LibraryPublished/FCS2004_-_Review_of_Sapphire_Oscillators.pdf
Chronos Timelord Rubidium Master Clock
Hello from a new member, Steve (angst46)
I recently purchased a Chronos Timelord Master clock and I am trying to find an English manual and more information about this unit. I would also like to but another but I understand that very few were made.
If anyone could help me I would appreciate it greatly.
Many thanks,
Steve (Australia)
Hello from a new member, Steve (angst46)
I recently purchased a Chronos Timelord Master clock and I am trying to find an English manual and more information about this unit. I would also like to but another but I understand that very few were made.
If anyone could help me I would appreciate it greatly.
Many thanks,
Steve (Australia)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.