Asynchronous I2S FIFO project, an ultimate weapon to fight the jitter

Hires, Your 22,579MHz Pulsar measures almost exactly like the 20MHz plot that I had posted. For me it's another confirmation.
Acko, I am just trying to do a sober crosscheck here.. As You see that plot of yours, if confirmed, is just strongly outstanding, and as such, invites for confirmation.. :)
 
Phase noise plot for my particular AX22 oscillator:

1689085293220.png
 
Mark got a couple of the sine output version and done some listening tests already with favourable outcome
… As You see that plot of yours, if confirmed, is just strongly outstanding, and as such, invites for confirmation.. :)
Measurements aside, what’s more important is the confirmation of how much better it sounds in an actual system and whether the level of improvements is worth the costs involved in the upgrade
 
...whether the level of improvements is worth the costs...
Worth the costs for whom? What market segment are we talking about?

Maybe consider that people who make very high end audio gear typically don't sell many of their most expensive products. Some products may sell only, say, maybe, two units per year. These are sometimes referred to as "statement pieces." They are to give some cachet to the company name more than to directly make money. Entry level products tend to be low profit as well, they may exist largely to get customers into the product line in the hopes they can be persuaded to upgrade later. Its the mid-tier products where sales volumes and profit margins combine to make for most of the company income.
 
Last edited:
  • Like
Reactions: 1 users
Ian canada UcPure mkIII with the new UcBalancing board kit and Eaton Ultra capacitors.

When Ian search for beta testers who already have the psu parts needed for an UcCapacitor psu, I signed up for a set of the UcBalancing kit. Ian responded positiv and sendt me one set UcBalancer kit the same day.

Yesterday I put it together in my dddac to feed the Accusilicon 338 clocks (not 318), and everything sounds ok.
It took less than an our to charge the Eatons via the AC input on UcPure. S5 on UcPure is left open for less current when charging from AC input and not DC input (according to the manual).

I will post more about the sound and the difference between my former psu the Linear Pi mkII with UcConditioner board,and this new UcPure psu with Ian’s balancer boards.

ops, I have connected the psu the safe way via the fuse, just to be sure everything works ok, then later on next step is direct pure out from the Eatons.
 

Attachments

  • IMG_1384.jpeg
    IMG_1384.jpeg
    646.3 KB · Views: 200
  • IMG_1385.jpeg
    IMG_1385.jpeg
    643.5 KB · Views: 193
  • Like
Reactions: 4 users
...everything sounds ok...
Wish I could everything looks okay. In particular the I2S wiring from FIFO_Pi to DDDAC appears quite long and improperly grounded. Beyond that, tin pin header interconnects can introduce timing jitter. All critical pin header connections should be gold plated or gold flashed. There should be enough gold to prevent surface corrosion. Otherwise a significant amount of the low jitter performance of good clocks is probably lost before it gets to the dac. Also looks like some other possible problems. Radiated EMI/RFI problems looks like a risk factor.
 
Last edited:
It might help to rotate DDDAC so that its I2S port is more closely aligned with FIFO_Pi output. Then put a lossy metal shield plate/wall between FIFO_Pi and DDDAC, leaving a opening for I2S connections. No doubt RPi and or FIFO_Pi are radiating a lot of EMI/RFI (some of which from RPi might already be getting into the clean side of FIFO_Pi; that's why Station_Pi exists, to help fix that particular coupling mechanism). USB boards radiate too. Better to keep that garbage out of the DAC if possible. Only way to see how much it might help would be to try it.
 
I know that there is many things to do to improve the dac, shorter i2s wiring, better isolation between the boards and a optimized layout for all components. But since there is so many exiting things coming out from Ian, i will wait a bit longer before rebuilding my dac.

ps ; no Rasberry, but Aurender streamer, usb out to WaVeio i2s.
 
Ian canada UcPure mkIII with the new UcBalancing board kit and Eaton Ultra capacitors.

When Ian search for beta testers who already have the psu parts needed for an UcCapacitor psu, I signed up for a set of the UcBalancing kit. Ian responded positiv and sendt me one set UcBalancer kit the same day.

Yesterday I put it together in my dddac to feed the Accusilicon 338 clocks (not 318), and everything sounds ok.
It took less than an our to charge the Eatons via the AC input on UcPure. S5 on UcPure is left open for less current when charging from AC input and not DC input (according to the manual).

I will post more about the sound and the difference between my former psu the Linear Pi mkII with UcConditioner board,and this new UcPure psu with Ian’s balancer boards.

ops, I have connected the psu the safe way via the fuse, just to be sure everything works ok, then later on next step is direct pure out from the Eatons.
Would it be possible to post a couple of pics of little up close on how the transformer gets connected to the ultra capacitors and how the transformer connects to the mains (in your case). It would be of great help and reference for many of us.
 
Rather than make a plan to rebuild then execute it, maybe more useful to make a plan to do a series of experiments then plan a rebuild based the experimental results. For instance, right now you probably don't know how your dac might sound if laid out differently one way as opposed to another way. One thing to try is take things loose and start moving them around to see if the sound changes at all when as anything is moved. That could be wires, boards, whatever. What happens if you separate things more, move them closer, rotate them, stick a reasonably big plate of metal between them? What if its a steel plate? What if its aluminum? Copper?

IOW, we know so little about what's really going on with builds because we don't do enough perturbation testing. We don't know the sensitivity to small or large changes. Yes, its work. But its the only way to learn certain things. There are also ways to measure radiated fields that aren't too expensive. There is Tiny VNA and a set of field probes. There is connecting a scope probe ground to the probe tip and using it as a loop antenna.

To me, the idea of design it on paper, build it so it looks nice, maybe take a few measurements, and then call it done is not enough. Its not the way to learn how to keep making incremental improvements.

Of course, maybe that's just me and the way I like to do my version of the hobby.

EDIT: Also, if somebody would be willing to do the work, then there could be a known good layout developed that people could have some confidence was really better than some other random way. People could at least start from what is already known and then try to make some improvements from there. IOW, build on existing knowledge that has been verified by replicated experiments. Learn together.
 
Last edited:
  • Like
Reactions: 1 users
NeoTheOne

No pictures yet, the dac is in the rack and playing.

Markw4

For me it is just a hobby. I like to play around and try out different things. One thing that I learn is that Ian Jin is making great dac parts, and I have always been impressed of his engagement and skills when it comes to improve already good stuff.
An example is the FiFoPi Q7, which is way better than the previous ones.

I take it slowly, and read about other diy members and their journey in this great hobby. Don’t forget the most important,the music it self. It is easy to overthink and dig a big hole in the ground and loose focus on the music.

bohrok2610

FifoPi improve the sound, I have tried without, and no doubt, FifoPi stays.
 
  • Like
Reactions: 1 user
FifoPi improve the sound, I have tried without, and no doubt, FifoPi stays.
Then you didn't try it the right way "without." Maybe the problem is that Ian doesn't offer a good non-FIFO solution for USB?

Regarding the music, you just said that FIFO_Pi improves the sound so its staying. What if you could improve the sound a lot more, but you just don't know it?