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

Status
Not open for further replies.
Disabled Account
Joined 2019
Petrus :) Glou glou miam miam :yes: :cool:


Not exactly a Bordeaux nore a 100% Merlot technically but who cares :D ...



I believe it's more the soil that the fact it's a near 100% Merlot that give its complex maturity & flavor (here the ferous is good for the music)... btw I like very much as well the almost more affordable R2R (Rolls to Royce :D) and very near St Emilion as the Pomerol parcell :cool: ! Though it's a big Cut in the pocket but worth it.. at least once in a life but it's more expensive than the Total Excellium per liter !



Ah thanks I remember now this story of never shutdown the Xtals. I believed first it was for their integrity and not their good aging... Well I'm not confortable to have always something switched on nowadays ! The good info is my 5.xx AT & AC will last till my liver becomes so fat that I will look like a purple punk goose from the Perigord ! :D
 
Both oscillators are without oven, we have measured them with and without the oven and there are no appreciable differences, the phase noise plots are superimposable.
Of course there is a difference in Allan Deviation, but since we look for the best short term stability we have removed the oven to make the board easier to assemble.

In the Differential oscilaltor we have used the SSM2212.
We also have tried the SMBT3904 but the phase noise performance is worse.

AFAIK, SSM2212 (MAT12) has the lowest 1/f noise of just about any low noise BJT. The noise is almost flat to 1Hz. MAT14 is similar, might even be better, hard to say from data sheet.
I'm assuming SMBT3904 is similar to 2N3904, if so the SSM will be lower noise especially at low frequencies.
Might be worth trying SSM in Driscoll if it is tolerant of the much higher capacitances. If Driscoll doesn't like the higher capacitances of SSM, 1/4 of MAT14 has much lower C and super low noise above around 4mA (0.5nV/rt Hz at 10Hz).

The 5 / 6 MHz results are *really* good. It's a a shame the 11.2896 didn't come a bit closer to the 10MHz 'York' differential specs as many of us will require 45 MHz so using your 11MHz with 2 doublers is not quite as good.
Is it possible to use 3 x doublers with an amplifier in the chain to derive 45MHz from 5?

Again, top effort.

TCD
 
We have already tested the SSM2212 in the Driscoll circuit but there is no phase noise difference against the 3904.
In the Driscoll oscillator the crystal is placed in the emitter circuit instead of the feedback loop, so is not much affected by the 1/f noise of the active device, the phase noise plot with the SSM and the 3904 are superimposable.

You could try to use 3 doublers but the problem is that you loose a lot of output power.
Maybe in the future we could ask for a different crystal cut at 11/12 and 22/24 MHz in HC/47 size.
 
What worries me a little are the size of the circuit boards and the many housings that are needed. For a Dac with 22/24 MHz you need at best 6 housings and the corresponding power supply units.
I would prefer an all in one solution here.

They are external devices, they should not stacked with any other device, you have to keep them far from the source and the DAC using SMA terminated coax 50 ohm cables.

We are just testing an all in one solution, but it's only for AT-Cut crystals and Pierce oscillators.
The All in One solution TWTMC-PXO-AIO uses AT-Cut crystals at 5.6448 and 6.144 MHz, then there is a pair of doublers on board so you can get 11/12 MHz or 22/24 MHz.

The new Driscoll and Differential oscillators with SC-Cut crystals are SOTA devices that need separate supply and cases.
BTW with the TWRPS-pp regulator you can supply 1 oscillators plus a pair of doublers, and with batteries (suggested) you can supply all the oscillators and doublers.
 
No, I have not contacted Croven but I expect the same reply for small quantities.

Anyway is not a problem of manufacturer since Laptech makes very very good crystals, it's a problem of price.
As you can see the 5/6 Sc-cut crystals are more expensive than the 11/22/24 MHz, almost double priced.
As I said in the future I will get some 11/12/22/24 MHz with different cut in HC-47/U package and I expect the same price of the 5/6 MHz one.
 
A few months ago I contacted Croven Crystals and sent an inquiry for crystal quartz 16.9344 MHz / HC43U / SC cut -3rd OT.
They responded that the minimum order quantity - 25 pieces ( $ 47.90 per piece). Other qty. pricing can be quoted upon request. Lead time is currently 12-14 weeks.

Also in the specifications was looking for a highly polished crystal and it's important detail in the reply:
"We would also recommend an etch finish not polished finish which is non-standard and will drive cost very high.???

After that I sent another inquiry for 5 pieces and got the reply.

They told me that CC is probably not the place for hobbyists due to their pricing structures. As a supplier of mainly hi-reliability, low phase noise and g-sens crystals, their market is now mainly aerospace,
military and communication infra-structure applications.They have the capability to build more commercial grade products, but the costs reflect their overhead rates.

Because this isn't a standard frequency, they will need to buy a minimum lot of raw material quartz and so the pricing is high for such a small qty. 5 pcs would be $120 each, and at this time they don't expect
shorter lead times that is 12-14 weeks.

As for the answers,everything is correct and professional ,but from all shown, it seems that Laptech is here still the best option as a base for HQ crystal oscillator.
 
Just a clarification

We have no intentions to contact any other crystal manufacturers, we have designed the oscillators on the Laptech crystals specs and we are happy with them.

Of course everyone is free to contact other manufacturers and buy the crystals from them, but we cannot assure that the new oscillators will work correctly with different crystals.
There are a lot of parameters to be kept in account (ESR, Q, overtone and so on), so with different crystals the new oscillators could not start or could work with a wrong drive level.
 
In this You are absolutely right.
Each time the oscillator would need a redesign, fit to the new crystal. And the result controlled with tests.

This all would not have started, if the York Univ. Oscillator would not have such a good performance at 10MHz.. So one could get away with the doublers.
But already what You have done is excellent, so never mind..

Ciao, George
 
We have no intentions to contact any other crystal manufacturers, we have designed the oscillators on the Laptech crystals specs and we are happy with them.

Of course everyone is free to contact other manufacturers and buy the crystals from them, but we cannot assure that the new oscillators will work correctly with different crystals.
There are a lot of parameters to be kept in account (ESR, Q, overtone and so on), so with different crystals the new oscillators could not start or could work with a wrong drive level.

You mentioned Nofech, and I just followed up.I had no intention of researching the market either.When I received an offer from Laptech for 10 pieces (moq) wanted to see what another manufacturer
has to offer. Honestly, I wanted to see if it was possible to order samples ( 2pcs) instead of 10pcs. from Laptech.It’s an old shabby stunt but you can’t know if you haven’t tried it.

As pointed out, as stated, crystal from Croven Crystals is a commercial grade, and this can be seen in the finishing.
Quote: "We would also recommend an etch finish not polished finish which is non-standard and will drive cost very high".
Laptech crystal is higly polished.


Also, Croven Crystals - Motional Resistance Rs = 80ohms max. Laptech crystal - Motional Resistance Rs = 60ohms max.
All of my Laptech crystals measured a resistance below 45ohms and a few below 40ohms.
Croven Crystal - Q factor = 500
Laptech crystal - Q factor = 500
All of my Laptech crystals measured a Q factor over 500 and two of them have touched 600.

Of those two offerings, I think Laptech is a better choice.

I don’t think there’s need to look any better than Laptech crystals.They represent high quality and exceed the requirements of the audio industry.
After all, the results you've already got just confirm that.
 
Last edited:
@tom59hifi

I understood what you meant, but I just wanted to make it clear that to achieve such results it needs years of hard work and so we have no time to start a new developement with different crystals.

Moreover we cannot assure that our oscillators will work with different crystals, so we cannot support anyone decides to use different crystals.
 
Status
Not open for further replies.