Let me introduce myself, I am the representative of a team that in the coming months intends to propose on the market Ultra Low Noise OCXOs tailored for audio purposes.
Two samples was realized for the 90.316800 and 98.304000 MHz frequencies, the test provided the following phase noise figures(dBc/Hz):
90.316800 MHz : -65(@1Hz) -100(@10Hz) -130(@100Hz) -159(@1kHz) -168(@10kHz) -170(@100kHz)
98.304000 MHz : -67(@1Hz) -101(@10Hz) -131(@100Hz) -159(@1kHz) -168(@10kHz) -170(@100kHz)
The other interesting characteristics are:
- 3.3 V Power Supply
- HCMOS Output
- DIL14 Standard Pinout and Size
- Only 50 mA Operating Current
- Less then 190 mA Warmup Current
In attachment you can find the whole preliminary datasheets.
The final price is heavily influenced by the production volumes, this survey is intended to understand if it exists enough interest for this OCXOs to start the production.
If the survey reveals enough interest the target price will be in the range 370.00 to 400.00 Euro per item + Shipment and VAT (21% now) if delivered in a EU country.
If you are interested in these OCXO this is the survey spreadsheet page: https://docs.google.com/spreadsheet/ccc?key=0AtPgpBGq_lTvdE5Na25jUk9yWTBTYTFfT1RWVzVfZUE&usp=sharing
If you are looking for different fequencies you can state them in the survey form.
Kind regard
The ULN.OCXO team
Two samples was realized for the 90.316800 and 98.304000 MHz frequencies, the test provided the following phase noise figures(dBc/Hz):
90.316800 MHz : -65(@1Hz) -100(@10Hz) -130(@100Hz) -159(@1kHz) -168(@10kHz) -170(@100kHz)
98.304000 MHz : -67(@1Hz) -101(@10Hz) -131(@100Hz) -159(@1kHz) -168(@10kHz) -170(@100kHz)
The other interesting characteristics are:
- 3.3 V Power Supply
- HCMOS Output
- DIL14 Standard Pinout and Size
- Only 50 mA Operating Current
- Less then 190 mA Warmup Current
In attachment you can find the whole preliminary datasheets.
The final price is heavily influenced by the production volumes, this survey is intended to understand if it exists enough interest for this OCXOs to start the production.
If the survey reveals enough interest the target price will be in the range 370.00 to 400.00 Euro per item + Shipment and VAT (21% now) if delivered in a EU country.
If you are interested in these OCXO this is the survey spreadsheet page: https://docs.google.com/spreadsheet/ccc?key=0AtPgpBGq_lTvdE5Na25jUk9yWTBTYTFfT1RWVzVfZUE&usp=sharing
If you are looking for different fequencies you can state them in the survey form.
Kind regard
The ULN.OCXO team
Last edited:
First, I should express my sincere gratitude to ULNOCXO those who offered me a valuable opportunity to test a pair of ULNOCXOs.
Before trying ULNOCXOs, I had just tested NDK NZ2520SA prototype samples of 90.316800 MHz and 98.304000 MHz by replacing NDK NZ2520SD regular products of 45.1584 MHz and 49.152 MHz on Chiaki's Dual Mono ES9018 DAC (The DAC module is coupled with SDTrans384 SD memory transport in a synchronous master clocking scheme).
In general, when we use a crystal oscillator device of high-frequency as a master clock source for ES9018 DAC, we should accept a certain trade-off. For instance, the higher frequency we use for seeking a better resolution of ES9018 output, the worse phase noise property of master clock we should accept.
When I compared NDK 90s MHz ones with NDK 40s MHz ones, I realized the trade-off. The 90s MHz oscillators brought a higher resolution, at the same time it accompanied a feeling of something too sharp or harsh.
I changed NDK 90s MHz ones with ULNOCXO 90s MHz ones. The adverse results I recognized in selecting higher frequencies disappeared! The ULNOCXOs improved all the aspect of resulting sounds, such as resolution/separation, timbres, a nature of bass, staging in 3D, fast response of percussion and moreover they brought a joy of listening music to me.
I thought those ULNOCXOs were quite "preiswert" ( a German word) even if one piece costed approximately 500 Euro. I would not hesitate to purchase one if it were sold now.
However, we should remember that the best performance of the OCXO must be obtained only when we prepare a high quality power supply and a DAC module commensurate with the OCXOs.
One more important thing. The best OCXO performance does not directly mean the best performance of the total systems. Some expensive systems in a professional domain may demonstrate far better sounds and music without employing such OCXOs.
Before trying ULNOCXOs, I had just tested NDK NZ2520SA prototype samples of 90.316800 MHz and 98.304000 MHz by replacing NDK NZ2520SD regular products of 45.1584 MHz and 49.152 MHz on Chiaki's Dual Mono ES9018 DAC (The DAC module is coupled with SDTrans384 SD memory transport in a synchronous master clocking scheme).
In general, when we use a crystal oscillator device of high-frequency as a master clock source for ES9018 DAC, we should accept a certain trade-off. For instance, the higher frequency we use for seeking a better resolution of ES9018 output, the worse phase noise property of master clock we should accept.
When I compared NDK 90s MHz ones with NDK 40s MHz ones, I realized the trade-off. The 90s MHz oscillators brought a higher resolution, at the same time it accompanied a feeling of something too sharp or harsh.
I changed NDK 90s MHz ones with ULNOCXO 90s MHz ones. The adverse results I recognized in selecting higher frequencies disappeared! The ULNOCXOs improved all the aspect of resulting sounds, such as resolution/separation, timbres, a nature of bass, staging in 3D, fast response of percussion and moreover they brought a joy of listening music to me.
I thought those ULNOCXOs were quite "preiswert" ( a German word) even if one piece costed approximately 500 Euro. I would not hesitate to purchase one if it were sold now.
However, we should remember that the best performance of the OCXO must be obtained only when we prepare a high quality power supply and a DAC module commensurate with the OCXOs.
One more important thing. The best OCXO performance does not directly mean the best performance of the total systems. Some expensive systems in a professional domain may demonstrate far better sounds and music without employing such OCXOs.
Last edited:
As for a DC +3.3V power supply I used a modified TI TPS7A4700 evaluation board(Ceramic capacitors on the board were replaced with multiple types of film capacitors based on my experiences). A big CDE 100 uF film capacitor is connected in the middle of the power line for decoupling though it does not appear in the photo. At the power input pin of the oscillator, I located/ does not locate/ an additional 47 uF Sanyo OS CON for a further decoupling.
I observed a maximum draw just after an initial power on was approximately 200 mA and it decreased gradually as time went and a stable draw after one minutes was approximately 50 mA.
I observed a maximum draw just after an initial power on was approximately 200 mA and it decreased gradually as time went and a stable draw after one minutes was approximately 50 mA.
Hi modmix,
Even if presently we are mainly involved with the higher frequencies the production of the OCXO for lower frequencies is just a matter of numbers, with a reasonable interest we can start the production.
I suggest you to sign up your request into our survey page so we can trace the requests and to inform you when the OCXO will be available, just report the desired frequency/ies in the "Notes" column.
For you convenience I report again here the survey page link:
https://docs.google.com/spreadsheet/ccc?key=0AtPgpBGq_lTvdE5Na25jUk9yWTBTYTFfT1RWVzVfZUE#gid=0
Excuse me for this, but... wouldn't getting an used Rubidium, and then building a very good clock converter/transformer be better and somewhat (for what it is and does) cheaper?
The Rubidium wouldn't be included, just the converter/transformer circuit.
I'm after 22.5792MHz, 24.576MHz and 100MHz frequencies.
The Rubidium wouldn't be included, just the converter/transformer circuit.
I'm after 22.5792MHz, 24.576MHz and 100MHz frequencies.
OCXO, or especially Rubidium address non-problems in audio reproduction - long term drift/absolute accuracy aren't real human audio perceptual problems
OCXO can control some close in phase noise - but there's no psychoacoustic evidence or signal theory support for the need for much lower phase noise at offsets a tiny fraction of the smallest perceptual critical band size of ~100 Hz
OCXO can control some close in phase noise - but there's no psychoacoustic evidence or signal theory support for the need for much lower phase noise at offsets a tiny fraction of the smallest perceptual critical band size of ~100 Hz
Last edited:
Talking about phase noise in case of audio reproduction by a dac deals with the question of how much is added to the signal due to deviations of the clock from the assumed equally spaced time ticks.
I can report from my trials (confirmed by some friends) that lower phase noise improves sound quality. A better Rb clock was where all that started. Now I beyond HP 10811 in terms of low phase noise. Each and every step was easy to tell.
Sorry for th OT.
Ulli
This is my experience too. Its not just the bass that is improved - the whole soundstage gest much more stable, as set in rock and tones, low and high, get more analogue like but still with definition - thats how a low close in phase noice "sound". You don't get this with a -140dbc at 10khz but -75dbc at 100hz clock.
//
//
Rb is usually specified where good long term accuracy is required (And when locking to GPS is not possible) but it sometimes does not have spectacularly good PN numbers.
It is of course total overkill in audio, but can have a place in precision timing and for some very narrow band radio work.
I note that these modules lack the ability to be electrically pulled by a few hundred ppb, which I always thought was a standard feature of an OCXO, this means that locking a low phase noise source to an external noisy but very precise source is not possible which limits somewhat the utility of these units (For audio it means you cannot follow word clock, for other uses it means no ability to lock to cesium via GPS).
One of the standard tricks in precision timing is to lock a source having good PN at large offsets to one having low phase noise close in, with the loop bandwidth defining where the phase noise contributions cross over, this requires the second source to have a tuning input.
Wenzel associates have some very good stuff on low noise sources and precision timing generally, and their trick for finessing the noise from an LDO is well worth having in the toolkit, as are the harmonic multiplier circuits (PLLs usually have better performance if you multiply the reference rather then dividing the output, a passive diode multiplier can have lower PN then a cmos or pecl divider chain).
I would note that layout and termination **really** matter and far too often you see some ocxo mounted on 6 inch wires with no local decoupling of the UHF sort and the output run along a few inches of unscreened wire past the power supply.... Of course it sounds different!
I don't see the Allan variance curve for these units on the datasheet?
73 Dan.
It is of course total overkill in audio, but can have a place in precision timing and for some very narrow band radio work.
I note that these modules lack the ability to be electrically pulled by a few hundred ppb, which I always thought was a standard feature of an OCXO, this means that locking a low phase noise source to an external noisy but very precise source is not possible which limits somewhat the utility of these units (For audio it means you cannot follow word clock, for other uses it means no ability to lock to cesium via GPS).
One of the standard tricks in precision timing is to lock a source having good PN at large offsets to one having low phase noise close in, with the loop bandwidth defining where the phase noise contributions cross over, this requires the second source to have a tuning input.
Wenzel associates have some very good stuff on low noise sources and precision timing generally, and their trick for finessing the noise from an LDO is well worth having in the toolkit, as are the harmonic multiplier circuits (PLLs usually have better performance if you multiply the reference rather then dividing the output, a passive diode multiplier can have lower PN then a cmos or pecl divider chain).
I would note that layout and termination **really** matter and far too often you see some ocxo mounted on 6 inch wires with no local decoupling of the UHF sort and the output run along a few inches of unscreened wire past the power supply.... Of course it sounds different!
I don't see the Allan variance curve for these units on the datasheet?
73 Dan.
Phase Noise Roll-off Listening Test
Dear diyAudio Forumers,
the discussion on the benefits of a better phase noise is very interesting and your experiences shows clearly that the quality of the D/A conversion clock is a key point in the listening experience.
If and/or how a better close-in phase noise vs. a better noise floor influences the quality of the sound is a very interesting issue.
To handle it from a mathematical point of view is not easy and a test implies the availability of two oscillators of the same nature that differ "only" for the phase noise profile. In general this is not feasible, but we managed this issue in the OCXO design process and tuning the OCXO parameters we got two gemini OCXOs that exhibit as a unique difference the phase noise profile as shown in the annexed picture.
With this unique couple of devices we had the possibility to perform, probably first in time, an experiment capable to answer to the above doubt!
We tested the OXCOs pair on different systems and the conclusion was always a clearly audible sound improvement using the OCXO with lower close-in phase noise.
Thanks also to this test the OCXO design phase is now complete and the production is going to start soon, we will make an effort to cover, with at list few units, all the audio frequencies in the first production lot, from 11.2896 to 100 MHz.
Currently there is no plan to include the Allan variance curve in the final datasheet that will be released with the phase noise profile values.
Dear diyAudio Forumers,
the discussion on the benefits of a better phase noise is very interesting and your experiences shows clearly that the quality of the D/A conversion clock is a key point in the listening experience.
If and/or how a better close-in phase noise vs. a better noise floor influences the quality of the sound is a very interesting issue.
To handle it from a mathematical point of view is not easy and a test implies the availability of two oscillators of the same nature that differ "only" for the phase noise profile. In general this is not feasible, but we managed this issue in the OCXO design process and tuning the OCXO parameters we got two gemini OCXOs that exhibit as a unique difference the phase noise profile as shown in the annexed picture.
With this unique couple of devices we had the possibility to perform, probably first in time, an experiment capable to answer to the above doubt!
We tested the OXCOs pair on different systems and the conclusion was always a clearly audible sound improvement using the OCXO with lower close-in phase noise.
Thanks also to this test the OCXO design phase is now complete and the production is going to start soon, we will make an effort to cover, with at list few units, all the audio frequencies in the first production lot, from 11.2896 to 100 MHz.
Currently there is no plan to include the Allan variance curve in the final datasheet that will be released with the phase noise profile values.
Attachments
100 MHz Close in Phase Noise Test
Prior to the upcoming availability of the Ultra Low Noise OCXOs we can offer the opportunity to test two samples of the new 100 MHz Ultra Low Noise OCXO designed for asynchronous conversion, the samples have different phase noise profile so the comparison we described in our previous post is possible.
The test can be easily performed in asynchronous mode on the Buffalo III board modified for DIL14 Clock pinout.
The only request from us is that the tester is located inside EU.
If someone is interested can contact us, we'll be glad to arrange the test.
Prior to the upcoming availability of the Ultra Low Noise OCXOs we can offer the opportunity to test two samples of the new 100 MHz Ultra Low Noise OCXO designed for asynchronous conversion, the samples have different phase noise profile so the comparison we described in our previous post is possible.
The test can be easily performed in asynchronous mode on the Buffalo III board modified for DIL14 Clock pinout.
The only request from us is that the tester is located inside EU.
If someone is interested can contact us, we'll be glad to arrange the test.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.