If 96Khz was OK but the 48Khz, it seems that the clock board didn't switch between Fs, what kind of clock board you are using now?
Ian
Im using the single XO version with 48/96 setup.
And the problem is even the lock light is on, there is no output from the aes of the interface board.
Last edited:
Hi mcluxun,
To switch between different Fs automatically, I suggest the dual xo or si570 clock board, however if you really want to play 48Khz by single clock board, you can :
1, use a 12.2880 MHz XO, or
2, use a 24.5769 MHz XO with short TP48 to GND, please refer "I2S FIFI Firmware V3.80 Upgrading" for details
If you have no sound with FIFO LOCK LED lighted, please check your I2S source or your system connections. All FIFO KIT were carefully tested bit perfect.
If you can't fix the problem, I also suggest posting pictures of your system, a lot of people would be very glad to help.
Have a good weekend.
Ian
Some magic just happen I unplugged and plugged back in everything. It worked!!
Here it is a "cheap" raspberry pi based network player:
Are you using raspberry i2s output?
Im really bad at software and Ive been trying to compile a linux kernel for this but didnt succeed. Do you have a compiled image where the i2s is active you want to share?
Thanks!
Some magic just happen I unplugged and plugged back in everything. It worked!!
Here it is a "cheap" raspberry pi based network player:
Very nice
. I never thought PI can work as a player.It works as an USB based player or memory card based player? Can you give some more details? I'm interested in it. It seems the problem was the connections. Please be very carefully about the 'backdoor' when you pull out the cable. PCB can easily get damaged if it wasn't soldered very well.
Ian
Yes it's rpi i2s output from the board.
Here is my repo for kernel. https://github.com/lkong/RaspbianKernel
For the kernel I've made some optimizations and replace stock scheduler with bfs . In short it's more of a machine focused on playback.
To use the i2s output you have to download everything from my repo and replace/copy accordingly (/boot to /boot and /lib to /lib)
Then edit /boot/config.txt there should be a line something like "kernel=kernel.img" change it to "kernel=kernel_asoc_BFS.img"
reboot and choose the sound card it should look like "snd_xxx_1541a"
Hardware wise you have to solder some pins on the P5 header.
some reference here: RPi Low-level peripherals - eLinux.org
And wire the output as I mentioned in former posts.
It's more likely to be a computer outputting i2s through onboard sound card.Very nice
It seems the problem was the connections. Please be very carefully about the 'backdoor' when you pull out the cable. PCB can easily get damaged if it wasn't soldered very well.
Ian
The rpi is running a linux system with modified kernel.
It's more than a sd card player. I installed a hqplayer network adapter on it (for serious listening). So it could be controller remotely and headless from other computers.
Also I have squeezelite/squeezeslave running on it, so it could be a full function squeeze network player.
It can run airplay and mplayer as well.
I put some epoxy around the backdoor kit so it's less scary.
Im more of a computer science guy who just picked the solder iron I'd like to help out if you got any software problem with it.
Yes it's rpi i2s output from the board.
Here is my repo for kernel. https://github.com/lkong/RaspbianKernel
For the kernel I've made some optimizations and replace stock scheduler with bfs . In short it's more of a machine focused on playback.
To use the i2s output you have to download everything from my repo and replace/copy accordingly (/boot to /boot and /lib to /lib)
Then edit /boot/config.txt there should be a line something like "kernel=kernel.img" change it to "kernel=kernel_asoc_BFS.img"
reboot and choose the sound card it should look like "snd_xxx_1541a"
Hardware wise you have to solder some pins on the P5 header.
some reference here: RPi Low-level peripherals - eLinux.org
And wire the output as I mentioned in former posts.
Thanks for this!
I'm going to try this on my Raspberry Pi running RaspyFi
Thanks mcluxun,
You did good job on soldering the backdoor connector
.I'm interested in buying a RPI, will let you know if I have any question then.
Ian
90.3168MHz/98.3040MHz OCXO to test, -135dBc/Hz@100Hz
A friend of mine sent email to me a couple of months ago, he told me that he got problem when playing 90.3168/98.3040 MHz clock on my Dual XO clock board and FIFO KIT. I replied, yes of course, Dual XO clock board does not support 90.xxx/98.xxx MHz XO so far. Then the guy sent another email to me and asked if there is any possible revising the hardware/software somehow to make it working for those frequencies, because they are really nice clocks. He said he would send them to me for evaluation during R&D time if there is any chance to make it.
I was quite exciting when I received them, they are a pair of 90.316800/98.304000 MHz OCXO. Phase noise only -105dBc@10Hz, -135dBc@100Hz. It seems they are exactly what I have been looking for for years. According to the specifications, the phase noise performance would be much better than the Si570, as well as CCHD951.
I thought OCXO could be very hard to use because of the high power consumption of the oven. But it seems I was wrong, this OCXO is different, only 150mA at start up and turns to 50mA after two minutes. It would be suitable for the DUAL XO clock board and my ESS9018 DAC. LifePO4 battery cell would be good choice to power it. But I have to revise the Dual XO clock board hardware, and probably firmware, to make it running at quad speed mode to support 90.3168/98.3040 MHz XOs.
The guy didn’t tell me where he got them and how much they are. Also, there was no any manufacture information on the spec he gave to me. I think he just wants to keep that information as secret for now for some reason.
I couldn't wait to test them. Will it be another planet or just no big difference?
Ian
A friend of mine sent email to me a couple of months ago, he told me that he got problem when playing 90.3168/98.3040 MHz clock on my Dual XO clock board and FIFO KIT. I replied, yes of course, Dual XO clock board does not support 90.xxx/98.xxx MHz XO so far. Then the guy sent another email to me and asked if there is any possible revising the hardware/software somehow to make it working for those frequencies, because they are really nice clocks. He said he would send them to me for evaluation during R&D time if there is any chance to make it.
I was quite exciting when I received them, they are a pair of 90.316800/98.304000 MHz OCXO. Phase noise only -105dBc@10Hz, -135dBc@100Hz. It seems they are exactly what I have been looking for for years. According to the specifications, the phase noise performance would be much better than the Si570, as well as CCHD951.
I thought OCXO could be very hard to use because of the high power consumption of the oven. But it seems I was wrong, this OCXO is different, only 150mA at start up and turns to 50mA after two minutes. It would be suitable for the DUAL XO clock board and my ESS9018 DAC. LifePO4 battery cell would be good choice to power it. But I have to revise the Dual XO clock board hardware, and probably firmware, to make it running at quad speed mode to support 90.3168/98.3040 MHz XOs.
The guy didn’t tell me where he got them and how much they are. Also, there was no any manufacture information on the spec he gave to me. I think he just wants to keep that information as secret for now for some reason.
I couldn't wait to test them. Will it be another planet or just no big difference?
Ian
Attachments
Last edited:
Ian, are these OCXO's from this supplier:
http://www.diyaudio.com/forums/vend...e-low-power-consumption-ocxo.html#post3586014
That is very low phase noise for a ~100MHz oscillator, especially close to
carrier. Generally speaking 5 to 10MHz is optimum for low phase noise and
the further you get away (higher) the phase noise, especially close in
suffers.
Bear in mind when comparing a 10MHz XO to 100MHz, you have to allow
for 20dB poorer phase noise of 100 due to shorter period for same jitter.
They might be SC-cut crystals by the look of the cans, if this is the
case they will be -very- expensive.
One other way to get very good close in phase noise with 100MHz OCXO's is
the use of frequency multiplier and also phase locking using low freq
oscillator as the ref. The tradeoff is usually poorer noise floor.
I am sure you will get a lot of interest in these
Get them in and cooking! Tell us what you think.
I am currently using SC-cut 11.2896MHz OCXO's for DAC's that have
-120dBC / 10Hz offset from carrier. However I am finding there is a lot of
subjective difference with various topologies of PS, squaring circuit, even the
OP loading makes significant difference to sound.
So there appears more to it than just a nice phase noise plot.
Z
Last edited:
Ian, are these OCXO's from this supplier:
http://www.diyaudio.com/forums/vend...e-low-power-consumption-ocxo.html#post3586014
That is very low phase noise for a ~100MHz oscillator, especially close to
carrier. Generally speaking 5 to 10MHz is optimum for low phase noise and
the further you get away (higher) the phase noise, especially close in
suffers.
Bear in mind when comparing a 10MHz XO to 100MHz, you have to allow
for 20dB poorer phase noise of 100 due to shorter period for same jitter.
They might be SC-cut crystals by the look of the cans, if this is the
case they will be -very- expensive.
One other way to get very good close in phase noise with 100MHz OCXO's is
the use of frequency multiplier and also phase locking using low freq
oscillator as the ref. The tradeoff is usually poorer noise floor.
I am sure you will get a lot of interest in these
Get them in and cooking! Tell us what you think.
I am currently using SC-cut 11.2896MHz OCXO's for DAC's that have
-120dBC / 10Hz offset from carrier. However I am finding there is a lot of
subjective difference with various topologies of PS, squaring circuit, even the
OP loading makes significant difference to sound.
So there appears more to it than just a nice phase noise plot.
Z
Thanks zenelectro for the link and the info. I think that's exactly what it is. But the price is totally out of my expectation, it's too much expensive than Si570 or CCHD957. I don't know that is reasonable for an OCXO or it's not worth it.
I did some search on digikey and found many OCXO's with similar performance are not as expensive as this one. Such as:
http://www.abracon.com/Precisiontiming/AOCJY.pdf
http://www.conwin.com/datasheets/cx/cx207.pdf
The only thing it that there was nothing with audio frequencies. However if going with ESS9018 internal clock, 100MHz SC-cut OCXO would be a good choice.
Any way, I'll going real listening test with these 90.xxx/98.xxx OCXOs to see what's gonna be happened.
Thank again
Ian
Last edited:
They look a lot like some parts I found at a Russian OXCO vendor ?Magic Xtal Ltd.? develops and fabricates all kind advanced Oven Control Crystal Oscillators (OCXOs). Portfolio of the company contains ultra-miniature vacuum sealed OCXOs with 1.5 cm3 volume and 150 mW power consumption, high-frequency low phase The specs are similar. However I was quoted around $110 ea in quantities of 10.
The datasheet looks similar as well http://magicxtal.com/upload/iblock/040/MXO37-8D-14D.pdf
They are probably a piece of the Russian Military Industrial Complex looking for some real income. . . However the NSA will be watching.
The datasheet looks similar as well http://magicxtal.com/upload/iblock/040/MXO37-8D-14D.pdf
They are probably a piece of the Russian Military Industrial Complex looking for some real income. . . However the NSA will be watching.
HAHAHAHAHAHA. Mate, when I read Ian's post the first thing in my mind was how will qusp's jitter-audio-nervosa handle seeing these!
Me, I wish I had the tools in the lab (read - garage) where I could measure some of these changes myself then play around with some variations. I want to know what performance I'm getting in-situ. Slowly getting there.
Zenelectro, have you done any measurements looking at changes of PSU, squaring etc? you've been at this clock biz for longer than I have ...
Ian, we're all very jealous and look forward to reading your impressions!
Last edited:
I found this one:
CTS Valpey Corporation
http://www.ctsvalpey.com/Collateral...equencyControl/Oscillators/OCXO/Model 144.pdf
I don't know it's the same guy or not
Dear diyAudio Jitter Fighters,
my name is Mauro Gavinelli and I am a member of the ULNOCXO team.
I followed your very interesting discussion initiated by iancanada with the post #2972 on this thread and I see that you need some more information to better understand what is in iancanada hands and what we are proposing, so I try to resume the whole story.
We are few friends sharing a passion for the top level audio and we like to experiment seeking for a (quasi) absolute sound.
We tested the iancanada products and we appreciated them so much that we wanted to listen them with the lowest jitter clocks available at an affordable price; when we came to look to the market we contacted many OCXO dealers and discovered that no one had an off-the-shelf component with all the characteristics needed to be fitted directly on the board.
We started to look for someone capable to produce such custom component and found only one european dealer (we are located in EC area - Italy) that gave us a positive answer.
The price of the ULNOCXO samples was quite high, but we decided to go on and the sound we listened was a real quantum leap.
The test was not easy due to the fact that the iancanada products (basically the current Dual XO board) was designed when the 90.316800 and 98.304000 MHz clocks were not yet available and the power supply should have been improved in order to guarantee a such low phase noise, but still in that conditions the quality appeared so high that we decided to contact iancanada in order to try to adapt the Dual XO board to these frequencies.
We understood that this couldn't be a quick process due to the above mentioned problems he had to overcome, so, while we were waiting a feedback from iancanada (that came with the a.m. post), we evaluated with our provider the possibility to share this great listening experience transforming a couple of samples in a real product.
They have been expressly designed on our specification, tailor made for Audio applications, so we posted the ULNOCXO thread to know if there is an interest for an OCXO with this high performances.
From here on the story is in this thread.
I read with great interest the discussion and see that all the evaluations are correct, I would like just to put in evidence a couple of points that in my opinion are not enough highlighted.
- Comparison with of the shelf items:
in the thread are mentioned many OCXOs with apperently similar phase noise but often the data sheet values refers to the 10 MHz version while the two ULNOCXO samples runs very near to 100 MHz. zenelectro is right when he says in his post #2973 that 10 MHz have a 20 dB advantage vs. the 100 MHz versions (I say even 25dB) so a comparison can be made only between the TEST data of our two ULNOCXO samples and the 100 MHz datasheet values (Magic and CTS)
The ULNOCXO samples appears to have lower phase noise.
- Power consumption:
Steady state: the two ULNOCXO samples requires 165 mW while the other comparable OCXOs requires 200 or 250 mW
Warmup: the two ULNOCXO samples requires 627 mW while the other comparable OCXOs requires 1000 or 1100 mW
- Control Voltage:
It is our opinion that for audio quality a low phase noise is much more important than a precise frequency so per our mandatory request the Frequency Control Voltage function has not been implemented lowering further the phase noise.
- The ULNOCXO samples are a Magic rebrand:
well, obviously we wasn't and won't be told by our provider, maybe they are! But comparing the measured phase noise of the ULNOCXO samples with the best Magic datasheets data (MXO37H/14) the maximum ULNOCXO samples figures (the only one guaranted values by the producer) are better then the Magic minimum figures. If they are a rebrand for sure they are special series.
- Price:
as told above we are located in Italy and the only way to sell something that is not a second hand item is to operate by means of a company and this has a certain cost; we are still working to find solutions to reduce the price of these new OCXOs, but consider that looking deeply to the performances they are really exceptional and also our procurement cost can't be reasonably low.
1audio: with reference to your post #2976, which Magic product are you referring to for that price, in term of frequency and phase noise figures?
For your convenience I report here the measured phase noise of the ULNOCXO samples(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)
We are glad that iancanada could spend a word on these ULNOCXO samples and we hope he can test them at best, possibly upgrading his board, in order to know his opinion not only on the base of datasheet numbers, but mainly on his listening experience; yes it's a subjective experience, but in the end it is the one that really cares us, audio lovers.
Mauro
my name is Mauro Gavinelli and I am a member of the ULNOCXO team.
I followed your very interesting discussion initiated by iancanada with the post #2972 on this thread and I see that you need some more information to better understand what is in iancanada hands and what we are proposing, so I try to resume the whole story.
We are few friends sharing a passion for the top level audio and we like to experiment seeking for a (quasi) absolute sound.
We tested the iancanada products and we appreciated them so much that we wanted to listen them with the lowest jitter clocks available at an affordable price; when we came to look to the market we contacted many OCXO dealers and discovered that no one had an off-the-shelf component with all the characteristics needed to be fitted directly on the board.
We started to look for someone capable to produce such custom component and found only one european dealer (we are located in EC area - Italy) that gave us a positive answer.
The price of the ULNOCXO samples was quite high, but we decided to go on and the sound we listened was a real quantum leap.
The test was not easy due to the fact that the iancanada products (basically the current Dual XO board) was designed when the 90.316800 and 98.304000 MHz clocks were not yet available and the power supply should have been improved in order to guarantee a such low phase noise, but still in that conditions the quality appeared so high that we decided to contact iancanada in order to try to adapt the Dual XO board to these frequencies.
We understood that this couldn't be a quick process due to the above mentioned problems he had to overcome, so, while we were waiting a feedback from iancanada (that came with the a.m. post), we evaluated with our provider the possibility to share this great listening experience transforming a couple of samples in a real product.
They have been expressly designed on our specification, tailor made for Audio applications, so we posted the ULNOCXO thread to know if there is an interest for an OCXO with this high performances.
From here on the story is in this thread.
I read with great interest the discussion and see that all the evaluations are correct, I would like just to put in evidence a couple of points that in my opinion are not enough highlighted.
- Comparison with of the shelf items:
in the thread are mentioned many OCXOs with apperently similar phase noise but often the data sheet values refers to the 10 MHz version while the two ULNOCXO samples runs very near to 100 MHz. zenelectro is right when he says in his post #2973 that 10 MHz have a 20 dB advantage vs. the 100 MHz versions (I say even 25dB) so a comparison can be made only between the TEST data of our two ULNOCXO samples and the 100 MHz datasheet values (Magic and CTS)
The ULNOCXO samples appears to have lower phase noise.
- Power consumption:
Steady state: the two ULNOCXO samples requires 165 mW while the other comparable OCXOs requires 200 or 250 mW
Warmup: the two ULNOCXO samples requires 627 mW while the other comparable OCXOs requires 1000 or 1100 mW
- Control Voltage:
It is our opinion that for audio quality a low phase noise is much more important than a precise frequency so per our mandatory request the Frequency Control Voltage function has not been implemented lowering further the phase noise.
- The ULNOCXO samples are a Magic rebrand:
well, obviously we wasn't and won't be told by our provider, maybe they are! But comparing the measured phase noise of the ULNOCXO samples with the best Magic datasheets data (MXO37H/14) the maximum ULNOCXO samples figures (the only one guaranted values by the producer) are better then the Magic minimum figures. If they are a rebrand for sure they are special series.
- Price:
as told above we are located in Italy and the only way to sell something that is not a second hand item is to operate by means of a company and this has a certain cost; we are still working to find solutions to reduce the price of these new OCXOs, but consider that looking deeply to the performances they are really exceptional and also our procurement cost can't be reasonably low.
1audio: with reference to your post #2976, which Magic product are you referring to for that price, in term of frequency and phase noise figures?
For your convenience I report here the measured phase noise of the ULNOCXO samples(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)
We are glad that iancanada could spend a word on these ULNOCXO samples and we hope he can test them at best, possibly upgrading his board, in order to know his opinion not only on the base of datasheet numbers, but mainly on his listening experience; yes it's a subjective experience, but in the end it is the one that really cares us, audio lovers.
Mauro