The performance of the latest circuit is exemplary. Almost too good. I would want to cross check the instrument (timepod?) with a known source just to be sure its accurate. There is a note on Wenzel's site about doing just that.
1/F noise in the amplifier is the main limit to close in phase noise if everything else is right.
Very interesting design from the two academics. Following on from that I believe it would be possible to use an ultralow noise audio opamp as the active element in a crystal oscillator. Most would have enough gain at 5 or 10 MHz to work and using a transformer at the input to get the best noise matching is a great idea.
1/F noise in the amplifier is the main limit to close in phase noise if everything else is right.
Very interesting design from the two academics. Following on from that I believe it would be possible to use an ultralow noise audio opamp as the active element in a crystal oscillator. Most would have enough gain at 5 or 10 MHz to work and using a transformer at the input to get the best noise matching is a great idea.
Hi Terry,
just do it, the second oscillator we are testing is just the York design with some variations. The new Driscoll is ready, while we are optimizing the York type. The original circuit has higher broadband noise against the Driscoll and also the output level is too low, so we are working solve these issues.
The prototype board of the modified York is on the way, time for final tuning.
Hi Demian,
Yes, the gear is the Timepod and we are sure it performs accurate measurements since we have compared the phase noise plots of a pair of known OCXO that was measured using the Aeroflex PN900 and the Agilent E5052A. The plots are identical.
From the many tests we have done in the last months we have observed that the close-in phase noise (below 10 Hz) is not affected much by the active device, the New Driscoll uses an ordinary 3904.
In the region below 10Hz from the carrier the phase noise is hevily affected by the flicker noise of the crystal.
That fits the photos I have stolen from Rubiola's book and placed at Jocko's
< DIYHiFi.org • View topic - To clock or not to clock, that is the question... >
Methinks the York design could be even better. They only get an operating
Q of 490K from 1.39 Meg unloaded . Part of it could come from the fact that
the crystal is driven from a +- 50-Ohm source, and the impedance at the
input side of the step up transformer is also not 0. With a crystal R0 = 53 ohm
most resonance losses thus happen outside of the resonator.
Tomorrow I'll get a board for a folded cascode Driscoll that fits the
Morion 5 MHz SC crystal.
Runs on +-5V.
Cheers, Gerhard
BTW I've not yet implanted the replacement attenuators, should happen
really soon now.
< DIYHiFi.org • View topic - To clock or not to clock, that is the question... >
Methinks the York design could be even better. They only get an operating
Q of 490K from 1.39 Meg unloaded . Part of it could come from the fact that
the crystal is driven from a +- 50-Ohm source, and the impedance at the
input side of the step up transformer is also not 0. With a crystal R0 = 53 ohm
most resonance losses thus happen outside of the resonator.
Tomorrow I'll get a board for a folded cascode Driscoll that fits the
Morion 5 MHz SC crystal.
Runs on +-5V.Cheers, Gerhard
BTW I've not yet implanted the replacement attenuators, should happen
really soon now.
Last edited:
Hi guys, perhaps someone here can help me out. I’m thinking of using a fan out buffer like the lmk1c11x and normally I wouldn’t give a damn about the added 7,5fs of rms jitter but with the awesome jitter values Andrea gets with the new clock design it means almost doubling the jitter.
So in short: Do you think I can neglect the jitter added by such a buffer or should I try to avoid them and compensate by a more complicated pcb design?
Any suggestions are welcome ^^
Greetings Oli
So in short: Do you think I can neglect the jitter added by such a buffer or should I try to avoid them and compensate by a more complicated pcb design?
Any suggestions are welcome ^^
Greetings Oli
This is awesome, I made 6 pcb's accomodating the York design already and curious if your modifications would fit on the pcb. I also found the same amplitude problem. Not sure how to proceed, monolithic or ic and how this fits comparing measurements of the paper vs yours. Hard to amplify low levels RF without paying with LF noise..
Can't wait
All those clock buffer chips are for telecom applications and they completely ignore the
noise outside 12KHz-20MHz to arrive at their wonderful specs. That standard is
completely OK for telecom apps, but neglecting the entire 1/f region below 12 KHz
is not what we really want.
The chips may be OK, but that must be verified.
I see, thanks. Was so happy to find a buffer with such low jitter values but I guess it was too good to be true looking at the phase noise diagrams of the chips I found until now and it seems they add a lot(!) more jitter if one includes the lower frequency domain, so I guess I’ll try to ditch the fan out buffer and so I don’t ruin all the work that went into those clocks ^^
looking at the phase noise diagrams of the chips I found until now and it seems they add a lot(!) more jitter if one includes the lower frequency domain, so I guess I’ll try to ditch the fan out buffer and so I don’t ruin all the work that went into those clocks ^^I see, thanks. Was so happy to find a buffer with such low jitter values but I guess it was too good to be true
An LTC6957 might be better suited, if you want to be flexible and want to use an IC.
More to discuss later, just a quick note - to set things straight, it's worth
noting the terms of use of this design.
From page 1 of the 'York' white paper:
"Reuse
This article is distributed under the terms of the Creative Commons Attribution (CC BY) licence. This licence
allows you to distribute, remix, tweak, and build upon the work, even commercially, as long as you credit the
authors for the original work. More information and the full terms of the licence here:"
About The Licenses - Creative Commons
Not trying to rain on the parade but the original designer(s) should be respected.
TCD
I could imagine that the White Rose is on quite thin ice here. The IEEE might not
like it that they put their header page in front of an IEEE transaction article that
is normally behind the IEEE pay - WallOfShame.
Not that my sympathy is on the IEEE side. Maybe the univ has bought their
own work back.
like it that they put their header page in front of an IEEE transaction article that
is normally behind the IEEE pay - WallOfShame.
Not that my sympathy is on the IEEE side. Maybe the univ has bought their
own work back.
Hi All,
So ... just a couple of days of vacation and suddenly there are so many remarkable news in this thread. Andrea - it looks to be outstanding and I feel like quoting mterbekke here with his comment "can't wait", and again, many thanks for doing this work ...
I am thinking that alongside battery supply these oscillators may be the most positive transformative contribution to a/my audio system I will ever have heard ....
@Hifoli:
Depending on how many outputs you need another diyaudio member (JohnW) in one of the other threads (can't remember which one, sorry) mentions that he has measured some/the best phase noise/jitter values with OnSemi's tinylogic. And in the tinylogic UHS series there's a dual buffer:
NC7WZ16
Cheers,
Jesper
So ... just a couple of days of vacation and suddenly there are so many remarkable news in this thread. Andrea - it looks to be outstanding and I feel like quoting mterbekke here with his comment "can't wait", and again, many thanks for doing this work
... I am thinking that alongside battery supply these oscillators may be the most positive transformative contribution to a/my audio system I will ever have heard ....
@Hifoli:
Depending on how many outputs you need another diyaudio member (JohnW) in one of the other threads (can't remember which one, sorry) mentions that he has measured some/the best phase noise/jitter values with OnSemi's tinylogic. And in the tinylogic UHS series there's a dual buffer:
NC7WZ16
Cheers,
Jesper
Jesper,
Good old 74ACxxx logic at 5 V has the lowest phasenoise I've measured - Tinylogic has higher Phasenoise - a second issue with tinylogic is its edge control circuitry which gives 3 defined output drive levels during the switching edge...
I still use Tinylogic due to its lower package impedancies and no inherent Gate crosstalk - however I parallel Gates where possible to reduce there noise contribution and increase drive current.
References?
Just curious ... about the various designs in this thread ... what were the some references you drew from? E.g., university or college textbook, electronics magazine, Elektor, web sites , etc.
In your OP (post #1), you vaguely referenced Jocks orig. "low-noise" regulator when he himself professed to having lifted from some ham radio book.
Just curious ... about the various designs in this thread ... what were the some references you drew from? E.g., university or college textbook, electronics magazine, Elektor, web sites , etc.
You can find tons of documents on the web about Clapp (1948), Pierce (1938) and Driscoll (1973) oscillators ("the various designs in this thread").
Simply do a Google search, you will found several interesting documents explaining how the above oscillators work.
Moreover, do a Google search about "oscillator phase noise". Again you will find tons of documents explaining the impact of phase noise in oscillator quality, then you can compare yourself the various plots published in this thread.
Sorry but I have no time to do it for you.
Hey thanks, that's a nice tip, I'll try it without buffers and try to make it up by proper routing/termination, I guess it will come to down to the amount of reflection that I'll get. Only problem I have is that with the output resistance of the driver I'm dealing with, the calculation for source termination of parallel transmission lines (Rsource = Z0 -Rdriver*N , N = Number of parallel transmission lines) gets negative so there is no Rsource I can set to completely eliminate Reflections so I just will put in 22Ohms and see how well that works out.
If it won't work that way I now have a plan B and I'll go with 74ac... in parallel
greetings Oli
References.
Some designs are proprietary. So who knows where yours came from. It's too bad you don't give References.
I'm not sure if KHashmi316 is a troll but sure as hell he's amusing, I leave this here ;D
--
Sorry, but it ain't worth my time to search for messages the OLD FASHIONED WAY.
If you want to impress, try using hyperlinks.
PSSSSSTT: That's why the Internet was invented
- Status
- Not open for further replies.