Since the DIL socket has been brought up, I have a question for a future 2022 PC project.
If I remove the Crystek clock off a JCAT USB card and solder the legs with a Mill-Max holder in place of the Crystek, would placing a STS work if the JCAT conforms to the DIP14 standard?
Will it work just like with the FiFoPi?
I can give up on the idea if it's a no go. Just in initial planning phase.
If I remove the Crystek clock off a JCAT USB card and solder the legs with a Mill-Max holder in place of the Crystek, would placing a STS work if the JCAT conforms to the DIP14 standard?
Will it work just like with the FiFoPi?
I can give up on the idea if it's a no go. Just in initial planning phase.
Last edited:
Having worked with a very expensive 24 MHz clock upgrade for a pci audio card, I am reminded that for audio there are only two places that the clock really matters. the first is the ADC which is sampling the analog audio and creating the digitally encoded version of it. The second one that we actually have control over is at the DAC. Since most Xmos and other pci output cards derive their "audio" clocks through calculation from the 24 MHz bus master clock, effort to improve this clock is of secondary importance. Adding a fifo and synchronously re-clocking the data to an actual audio frequency clock that drives the DAC itself can provide far better results and obsoletes any efforts on the bus clock.
@stephen1212b,
Thanks for kicking down the knowledge. It may end up saving me a big investment in the PC rabbit hole build which I was planning next year.
I guess I'm on the right track already as I went the FiFo / 5 MHz TWTMC route first.
I was planning a PC build to topple, but if that's not possible I may just build it as just a secondary complement since I'm due for a new PC build for personal use around that time.
So I'll focus my efforts on maxing out the FiFo build instead of pouring resources onto the PC build side. I don't mind duplicating the effort for the better, but if it's for the worse it gives me an out.
Thanks for kicking down the knowledge. It may end up saving me a big investment in the PC rabbit hole build which I was planning next year.
I guess I'm on the right track already as I went the FiFo / 5 MHz TWTMC route first.
I was planning a PC build to topple, but if that's not possible I may just build it as just a secondary complement since I'm due for a new PC build for personal use around that time.
So I'll focus my efforts on maxing out the FiFo build instead of pouring resources onto the PC build side. I don't mind duplicating the effort for the better, but if it's for the worse it gives me an out.
The aim for the PC if combined with well designed FIFO should be providing the DAC system with as much isolation as possible, shielding+distancing for radiated interference , optical data connection and battery or heavily filtered AC power for electrical interference.
Ideally minimising the source of interference with use of low speed computer (rpi) or other source.
This will also help address other sources of interference, as the PC is not the only one.
Ideally minimising the source of interference with use of low speed computer (rpi) or other source.
This will also help address other sources of interference, as the PC is not the only one.
Last edited:
A 5 MHz clock at the DAC should be awesome. It will limit you to just 44.1 sample rate but that’s 95% of my library. Ideally the fifo clock should be at least 2X the streamer master clock or run the streamer from a buffered ground isolated output of the same clock so that it is pulling the data at the same exact time. Check with your fifo manufacturer for recommended clock frequencies. If you are only feeding SPDIF and not I2S or running a true pcm ladder DAC instead of a delta sigma design then no worries. Enjoy it I am certain you will.
I thought I'd add a couple pictures. So I'm not really 'up there' with you guys and the SoTa drixos and exos but this is still a nice upgrade over my previous Crystek and NDK SDA. Hopefully this helps a few members just onto the ladder of adding better clocks.
Powered by a Salas Ref D...maybe ill change to a Ubib or even might try a much smaller psu such as a PFM Flea.
Feeds Ians Fifo via Ufl and Andreas DIP 8 adaptor board.
It is mounted with earbud silicone covers. Just hangs off the top 2. Far from ideal and is close to the AC inlet but it is working nicely until I can implement a separate chassis for it.
I
Powered by a Salas Ref D...maybe ill change to a Ubib or even might try a much smaller psu such as a PFM Flea.
Feeds Ians Fifo via Ufl and Andreas DIP 8 adaptor board.
It is mounted with earbud silicone covers. Just hangs off the top 2. Far from ideal and is close to the AC inlet but it is working nicely until I can implement a separate chassis for it.
I
Attachments
@lasercut. Yes, agree steps should be taken if going the PC route. I would only PC build if I can scale / bridge LiFePO4 power supplies plus mix in Super Caps. It will be like a big laptop that's off mains.
I also will go USB optical that's a 5-10m distance separation from PC and DAC.
Both should provide minimal interference.
@stephen1212b,
It hasn't hit me yet that I'm close to a 5 MHz dream clock. Maybe it will hit soon, but for now enjoying yours and other's impressions of the 5 MHz clock. I'm sure the 5 MHz will be most enjoyable as long as I don't cut corners or make a misstep in the FiFo build. Fortunately for us, we have Andrea's and Ian's great support if we make a rookie misstep along the way. Not to mention, help from fellow members.
I also will go USB optical that's a 5-10m distance separation from PC and DAC.
Both should provide minimal interference.
@stephen1212b,
It hasn't hit me yet that I'm close to a 5 MHz dream clock. Maybe it will hit soon, but for now enjoying yours and other's impressions of the 5 MHz clock. I'm sure the 5 MHz will be most enjoyable as long as I don't cut corners or make a misstep in the FiFo build. Fortunately for us, we have Andrea's and Ian's great support if we make a rookie misstep along the way. Not to mention, help from fellow members.
Last edited:
Hi !
If someone is interested, that's how I isolated the quartz from the pcb.
From my tests this resulted in the lowest transmitted vibration ratio.
The silicone cast was subpar for my standards... But i want the job done relatively quickly and after the second attempt i call it good enough for me [emoji4]
Surely not the "ultimate" way (whatever that means)... But surely it was the best among the options i have on the table at the moment.
The second "barrier" will be an elastic external case suspension dampened with some proper mass to de-tune the system for the low freq vibration response.
If someone is interested, that's how I isolated the quartz from the pcb.
From my tests this resulted in the lowest transmitted vibration ratio.
The silicone cast was subpar for my standards... But i want the job done relatively quickly and after the second attempt i call it good enough for me [emoji4]
Surely not the "ultimate" way (whatever that means)... But surely it was the best among the options i have on the table at the moment.
The second "barrier" will be an elastic external case suspension dampened with some proper mass to de-tune the system for the low freq vibration response.
Last edited:
... the mold was done with a 3dPrinter (ABS filament).
Done in two pieces in order to get released the two pegs that went trough the board in order to mechanically fix the silicon cover.
The pcb was drilled in two specific points and then the pegs was dragged trought them.
silicone used is "prolastix" from "prochima" (an italian company), shore 30 (tested 10-20-30-40, 30 was the one that attenuated the broadest vibration freq. spectrum).
If someone is interested i can share the STL files (keeping in mind that is good for f's > 5-6Mhz due to the phisical dimension and board constraints)
Done in two pieces in order to get released the two pegs that went trough the board in order to mechanically fix the silicon cover.
The pcb was drilled in two specific points and then the pegs was dragged trought them.
silicone used is "prolastix" from "prochima" (an italian company), shore 30 (tested 10-20-30-40, 30 was the one that attenuated the broadest vibration freq. spectrum).
If someone is interested i can share the STL files (keeping in mind that is good for f's > 5-6Mhz due to the phisical dimension and board constraints)
Which shore scale? Which frequencies did you test? What was the vibration test setup?
Thanks,
Bryan
Shore "A"
sampled at 500 / 1000 / 2000 / 4000 / 8000 / 10000 / 20000 Hz.
the eccentric mass was selected each time to be in a manageble range.
The comparison was done for each "solution" vs each frequency.
Used a purposely built test bench that i was using at work during some similar tests for a product i was developing the past year.
Vibration generator was made with an eccentric tunable mass with a servo controlled motor.
The sensor is a flex pcb with accelerometer attached at the end. the weight of the quartz case was very similar to the one of the accelerometer/pcb assembly, so in this case i was not adding any additional mass to the sensor probe.
Last edited:
I'm jealous. Would be interesting to try some silicone gels.
Also, how did you mount the board for the test?
-Bryan
- Status
- Not open for further replies.