The Well Tempered Master Clock - Building a low phase noise/jitter crystal oscillator

Status
Not open for further replies.
J test

Here's two crystals in a cheap Sony CD-490, side by side.
The J test measurement could be better, but the difference is still clearly visible.
I just made the player crystal connection plug-able and exchanged them.
No name low profile crystal to the left, a KDS one on the right.
The picture shows the bottom spread of the 11kHz jitter test signal.
 

Attachments

  • Sony CDP-490 tiny + KDS xtal jitter.jpg
    Sony CDP-490 tiny + KDS xtal jitter.jpg
    150.3 KB · Views: 950
Updated schematics, PCBs, BOMs and Assembly guide

Attached the assembly guides of all the boards available. They contain the schematic of each board, the BOM to source the parts and the instructions to build each circuit.

TWTMC-AIO was added, finished and bare boards available.

- TWTMC-C: Colpitts-Clapp oscillator revised
- TWTMC-D: Driscoll oscillator revised
- TWTMC-P: Pierce oscillator
- TWTMC-AIO: Dual Driscoll oscillator with power supply
- TWTMC-D&D: daughter board (power supply, dividers, and so on) revised to supply the Driscoll oscillator and the oven board
- TWTMC-OVN: oven board for SC-Cut crystals Driscoll oscillator
- TWTMC-DIL: DIL adapter for Ian's Fifo board
 

Attachments

hi there,

i know this is a high end design, but maybe there is a ''budget'' oscillator available that would be cheaper than the commercial ones (like the popular crystek's or ndk's) that would perform better than these and ''give a taste''....?

i'm thinking 10$ or so ballpark...

i'm about sure that there would be a good demand for the boards if so, given the good subjective results shown here...
 
Last edited:
hi there,

i know this is a high end design, but maybe there is a ''budget'' oscillator available that would be cheaper than the commercial ones (like the popular crystek's or ndk's) that would perform better than these and ''give a taste''....?

i'm thinking 10$ or so ballpark...

i'm about sure that there would be a good demand for the boards if so, given the good subjective results shown here...
You won't be able to buy the x-tal of this design for that money...
 
I hope that the seemingly systematic design of a top notch oscillator takes acoustic feedback inte consideration. The sound pressure and vibration can be substantial.

How do one preserve these precious dBc/Hz in reality?

//


Bravo. This is a very valid point.

Vibration can easily induce high levels of "close-in" noise on the oscillator output.

That is why I still remain slightly sceptic to the idea of close-in PN as being the single most mportant parameter for audio clocks.
In a listening room environment, all that hard work of lowering the close-in PN is wasted once the enclosure and the xtal start vibrating to the music.

Yet upgrading the clock seems to improve the percieved sound...

Can it be something else than just close-in PN??
 
I
how much then?
the (my) idea would be to compete with popular commercial clocks, if possible...

You'll find the answer to your question if you read Andreas comments at the very start of this thread.

It might be possible to build a really good clock with very cheap crystals, but then you'd need measurement equipment (source analyzers) that easily cost ~ 100 grand, to sort out the xtals and optimize the design through measurement.

The only way of getting around this was to:
1- start with a good xtal.
2- start investigating topologies to find the best one.

Thanks to Andreas efforts, and input from Gerhard, 1audio, ecdesigns and all other people, i think we're getting close.

...So no, you can't find a 10 buck audio clock that outperforms a well designed oscillator.

Best wishes.

PS:
There are some tricks one can throw in to improve performance of simple, cheap oscillators, but the easiest way out is to start with a decent xtal.
 
Last edited:
Bravo. This is a very valid point.



Vibration can easily induce high levels of "close-in" noise on the oscillator output.



That is why I still remain slightly sceptic to the idea of close-in PN as being the single most mportant parameter for audio clocks.

In a listening room environment, all that hard work of lowering the close-in PN is wasted once the enclosure and the xtal start vibrating to the music.



Yet upgrading the clock seems to improve the percieved sound...



Can it be something else than just close-in PN??



Not to say that there aren't also other factors but the effects of audio vibrations are eliminated with headphones (and hence easily testable) but we should consider the surprisingly large low frequency ground vibrations that should be mechanically isolated. Spring for an old Agilent or build one of our own testing circuits?
 
Bravo. That is why I still remain slightly sceptic to the idea of close-in PN as being the single most mportant parameter for audio clocks.
In a listening room environment, all that hard work of lowering the close-in PN is wasted once the enclosure and the xtal start vibrating to the music.

Yet upgrading the clock seems to improve the percieved sound...

Can it be something else than just close-in PN??

It could be that both close in PN AND low vibration is important.

//
 
Vibration

Well, I guess that was my question. If environmental vibration is an issue, and if these sorts of grommets would be effective to reduce that vibration, and if they need a different mounting hole, would it make sense on the next group buy to modify the boards so that they can use these grommets?
 
There are two questions here that need to be addressed. First- is vibration at the crystal an issue in audio systems. We don't have measurements to say one way or another. It should be pretty easy- take a system, play a J-test and "excite" it with high spl at a medium frequency that will be well separated from any other spurs in the J-test signal.

Second needs an understanding of vibration isolation (not something I see done well in turntables). First both the mass and the energy need to be understood. A rubber isolator can work but one rated for 10 pounds (5 KG) will do essentially nothing with 5 grams on it. There is no substitute for mass in reducing vibration. second, the isolator will need to be very loose to isolate from structure born vibration but that means its more sensitive to acoustic vibration. My short suggestion would be to put the whole oscillator in a heavy metal can on springs, but not until there is some evidence its an issue. Gerhard above made note of where this is a real issue- helicopters and radar, or vibration at 100 dB greater than we are discussing and sensitivity easily 10-100 times greater as well.

Its possible that the small packaged oscillators with tiny crystals will be less sensitive that ones with large crystals in metal cans.

However if you have some extra space shuttle tiles maybe it can make a difference? Femtosecond Clocks The plot seems to be one that can be duplicated pretty easily and seems to show close in phase noise (no $150K test system needed) although you would need a pretty clean ADC to be sure of what you measure.
 
Status
Not open for further replies.