On the PI Hats the I2S routing looks crappy compared to the board you have at hand.
I think one of the major I2S flaws on the PI is how weird (wildly spread) these folks assigned
the I2S lines to the GPIO header. That alone already prevents from well designed I2S
on a PI-HAT - IMO.
Not sure if a setup with e.g U.FLs from Kali ( Ioan -- a new feature!
) to its HAT
would be the better choice here.
Or e.g. at least offering a separate I2S header on Kali might be another option.
I think one of the major I2S flaws on the PI is how weird (wildly spread) these folks assigned
the I2S lines to the GPIO header. That alone already prevents from well designed I2S
on a PI-HAT - IMO.
Not sure if a setup with e.g U.FLs from Kali ( Ioan -- a new feature!
) to its HAT would be the better choice here.
Or e.g. at least offering a separate I2S header on Kali might be another option.
Last edited:
Attachments
So.
I also did the sync-MCLK-mod surgery.
1. relocated the 51R ( I cut the trace that connected to pin 27 - this way viases 29/27 make some nice pads for the 51R )
2. removed the clock LDO
3. used a 3-wire ribbon (gnd-clock-gnd) for the time being
4. Against earlier plans I'm using the old 51R pad to inject the clock.
5. I bent the ribbon away from the other I2S lines. (routing is not as shown on the image)
Now it's heating up.
PS1: Sorry Vasilis, Not much left of the MamboBerry.
PS2: What you see on top of the caps are NOT my old chewing gums.
@Soundcheck,
Good on the cap damping, I will do that on mine at some point.
Your pictures reminded me I have not posted any of my work on the MB...
Here's some of my sync mode setup.
As on yours, I had too much treble emphasis too with mine and a 51R series R. Try 100R.
Also try it on the other end of the line.
I suspect the u.fl will be better than the ribbon. I just used a single wire, worse choice, but it worked ok once I fiddled with the series R value.
Greg in Mississippi
P.S. You can see the PPS caps I used on the CP-CN and NEG-GRND pins pretty clearly here along with the ones replacing the small electrolytics. AND please excuse the goopy clear stuff, that is hot glue making sure the caps are well secured and won't accidentally tear off pads.
P.P.S Also try a couple of good 10R metal film (Vishay-Dale or PRP) for the output R's!
Attachments
Last edited:
@Soundcheck,
We mostly agree on the ES9022/ES9023. There were several comments saying basically the same as what you summarized in ESS-DAC chip-related threads back a few years (2012-2013?).
But of course we disagree on filter swapping, I have no issues believing that the manufacturer will select filters that make their chip measure best instead of sound best and that using other filters can sound better even when they measure worse. BUT doing that on the ES9022/23 may be hard... IF you upsample to 352/384 AND run it in Sync mode, the filtering used during the upsampling MAY be the dominant one. In an ES9018, you had to upsample to 352/384, run them in Sync Mode, AND turn off the Over-Sampling Filter (OSF) to make this work.
Since my setup does not yet have 352/384 upsampling enabled, I can't say how it sounds on the ES9022/23. BUT even using the different vendor-supplied filters in a PCM5122 or using community-generated filters on a Soekris makes a noticeable and useful difference.
These pix show most of what I did... you can't see both of the PPS caps on the AVCC side of the ES9023 because a cap is in the way, but you can see them on the CP-CN/NEG-GRND side.
You CAN see the caps on the ES9023 legs in a couple of these and one of the ones showing my Sync Mode mod.
I don't have a good picture of the stacked 1uf PPS/4.7uf stock electrolytic on VREG-GRND, I'll try to post one later.
AND as I said before, I picked these parts as ones I've used on other mods that have worked well for the cost. Nope, they aren't Black Gates. Nor are they ones selected by painstaking listening, then trying another such as what Vasilis does and many appear to have done on the Subbu DAC thread. But they worked and worked well... consider this one possible modset and at least a good, economical starting point.
Greg in Mississippi
Attachments
I won't argue about the Odroid possibly being a better base board than the Pi.
BUT there is are few turnkey distro/SW solutions for the Odroid... and plenty for the Pi.
What we need is a fully SW/HW physical configuration-compatible, but more sound-quality-focused version of the Pi that uses with more audiophile-ish power networks (linear regs, more capacitance, possibly additional regulation zones). It MIGHT be bigger, but as long as it has the same standard electronic interfaces AND mounting holes in the same place, it can be a drop-in replacement (well, you may need to do a different case).
AND that, I suspect, would sound better while still leveraging all of the distros and SW out there PLUS the community support.
Yah, I'd rather have a solution like the BBB with a clock that has an integer-relationship to our audio clocks. AND I'd love to have on-board isolation. But if someone made a board like this that was no longer HW/SW compatible with the base Pi, they throw out all of the existing distro/SW/drivers & community support base and have to start from scratch AND provide support too. Not too appealing in my book!
Greg in Mississippi
Last edited:
Yes, that's right as inputed SoundCheck as well, the more usefull way is to go for a Pi or a Pi-like !
Now I'm asking myself what is the better solution to stop some noise : chip isolator or traffo isolation ? Some like Pedja Rogic or John from ECDESIGNS seem clairly prefer the traffo isolation way when I2S input from "computer" boards are involved. Although IanCanada has a good sucess with his isolation electronic chip board to manage the I2S signals !
I have a Pi3 , but no time enough yet to try the distros (Moode from Tim planned in first...)
Hey I see you use the APSA (blue top caned polymer cap) caps as well : I like it a lot too with one of my Subbu board at AVCC.
Joachim G. made also a buffer with a filter for the audio outputt of this chip which improved mostly the bass area.
Some have testimonied better or as good result than the ES9018 high end model.
Now I'm asking myself what is the better solution to stop some noise : chip isolator or traffo isolation ? Some like Pedja Rogic or John from ECDESIGNS seem clairly prefer the traffo isolation way when I2S input from "computer" boards are involved. Although IanCanada has a good sucess with his isolation electronic chip board to manage the I2S signals !
I have a Pi3 , but no time enough yet to try the distros (Moode from Tim planned in first...)
Hey I see you use the APSA (blue top caned polymer cap) caps as well : I like it a lot too with one of my Subbu board at AVCC.
Joachim G. made also a buffer with a filter for the audio outputt of this chip which improved mostly the bass area.
Some have testimonied better or as good result than the ES9018 high end model.
Last edited:
Greg.
You made it the other way around on VREG and NEG!?!? I put the larger one on VREG and the small on NEG. I didn't like the larger cap on NEG ( In line with the Subbu folks) - I had it the other way around earlier.
Playing around with I2S is really a shot in the dark.
BTW. Things improved after day one and several hours running my TaraLabs burn-in CD.
(After 48h on that disc usually the DAC starts showing his real face.)
However. Yesterday I ripped everything apart again and started over.
The less forgiving a system gets, the more care must be taken of every tiniest detail.
You made it the other way around on VREG and NEG!?!? I put the larger one on VREG and the small on NEG. I didn't like the larger cap on NEG ( In line with the Subbu folks) - I had it the other way around earlier.
Playing around with I2S is really a shot in the dark.
BTW. Things improved after day one and several hours running my TaraLabs burn-in CD.
(After 48h on that disc usually the DAC starts showing his real face.)
However. Yesterday I ripped everything apart again and started over.
The less forgiving a system gets, the more care must be taken of every tiniest detail.
Just stepped over my potentially next project.
Soekris R2R HAT dam1231
Probably available by the end of the year.
Might make the Kali obsolete.
Let see how things develop.
Edit:
That DAC also seems to come with (autosensing?)) Toslink/SPDIF inputs.
Interesting will be if GPIOs/RS232 will be used.
Soekris R2R HAT dam1231
Probably available by the end of the year.
Might make the Kali obsolete.
Let see how things develop.
Edit:
That DAC also seems to come with (autosensing?)) Toslink/SPDIF inputs.
Interesting will be if GPIOs/RS232 will be used.
Last edited:
Hi folks.
Just to let you now about the status of my little joint-venture with Tim/Moode.
1.
Current Moode Audio 3.1 now offers the kernels based on Clives audio related patch set.
Beside numerous I2S and other audio related patches, SRs up to 384kHz are available for I2S DACs.
A number of named DACs have been added.
I also applied some performance related optimizations to the kernel.
Kernels are available for armv6 armv7.
2.
We also introduced squeezelite. (For the time being we skipped LMS)
Moode is one of the few, if not the only PI Audio distro, offering MPD and squeezelite on the same image.
3.
Several OS related performance enhancements have been introduced
That's pretty much it from my side for the time being.
My actual goal was to make the earlier discussed kernel enhancements (and squeezelite) available to a wider community. I hope there are a few out there making use of this.
Again. Thx to Tim for joining up.
And once more Thx to Clive for gathering/writing/assembling/testing and providing this quite impressive number of patches to the kernel. A great pity that he stepped out.
Enjoy.
Just to let you now about the status of my little joint-venture with Tim/Moode.
1.
Current Moode Audio 3.1 now offers the kernels based on Clives audio related patch set.
Beside numerous I2S and other audio related patches, SRs up to 384kHz are available for I2S DACs.
A number of named DACs have been added.
I also applied some performance related optimizations to the kernel.
Kernels are available for armv6 armv7.
2.
We also introduced squeezelite. (For the time being we skipped LMS)
Moode is one of the few, if not the only PI Audio distro, offering MPD and squeezelite on the same image.
3.
Several OS related performance enhancements have been introduced
That's pretty much it from my side for the time being.
My actual goal was to make the earlier discussed kernel enhancements (and squeezelite) available to a wider community. I hope there are a few out there making use of this.
Again. Thx to Tim for joining up.
And once more Thx to Clive for gathering/writing/assembling/testing and providing this quite impressive number of patches to the kernel. A great pity that he stepped out.
Enjoy.
Some great hardware , as usual , from Soekris.
I think this benefits everyone , from my perspective this is the best hi-end solution on the market right now.
I think this benefits everyone , from my perspective this is the best hi-end solution on the market right now.
Just stepped over my potentially next project.
Soekris R2R HAT dam1231
Probably available by the end of the year.
Might make the Kali obsolete.
Let see how things develop.
Edit:
That DAC also seems to come with (autosensing?)) Toslink/SPDIF inputs.
Interesting will be if GPIOs/RS232 will be used.
What is high-end? What is the market?
####
If you look at the list of tweaks (HW and SW) on Soekris DACs out there !?!? .....I'm not sure if we talk about a great product anymore
If you look at more critical comments about the Soekris DAC family !?!? .....I'm not sure if we talk about a great product anymore
The active output stages seem to be everything but appreciated. For a reason!?!? Everybody with a little experience can tell what opamps can do to the sound and what difference this or that outputstage will make..
Every reasonable commercial DAC out there comes with better/more sophisticated output stages. No wonder that the first thing people do with the Soekris is going passive out.
Beside that. All of us who played around with resistors in the path, learned that they have quite an impact. The numerous resistors in the Soekris DAC will add quite a flavour to the signal.
The HAT also seems to be a stripped down version. Soekris even got rid of the isolation stage.
And @ 260€ plus tax plus shipment plus power supply that DAC gets pretty close to real well down commercial solutions like a Chord Mojo (currently 415€).
I've never been 100% convinced about the Soekris DACs. That's why I never bought one.
Considering all that. I'm honestly not sure anymore if to go for a 400$/€ HAT DAC solution/project that still comes with this or that compromise.
Actually I'd rather say, I'll stay away from it.
Enjoy.
Last edited:
Hi End , I think , its nothing but a term . Market, I am referring to SBC audio .
Anyway , I think its about time that we (allo) with the help of this community we develop a higher end solution.
AKM4497 .Oscilators with better phase noise than Crystek . Not sure what will be the best output stage..
Anyway , I think its about time that we (allo) with the help of this community we develop a higher end solution.
AKM4497 .Oscilators with better phase noise than Crystek . Not sure what will be the best output stage..
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.