Probably the local regulators do help with isolation of noise produced by different load devices on the board. My guess would be to keep them as is, or maybe improve them with better filter caps at each regulator if you think something like that might help.
Went ahead and ordered some more film caps so I can directly compare Katana and the modded dac once again. I could also see if even more film caps has any further effect on sound quality. It could be that more would be better, or maybe I already have more than enough and am wasting some. Be interesting to see.
Hi Alex,
Yes, you might try looking at post #3306 here:
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-331.html#post5622375
At least hopefully that will help you get up to speed more directly. The most recent things I have been working on since my last summary might require browsing through. If anything isn't clear, please ask and I will try to provide what you need.
The area I am working in right now has just undergone a few test mods to see what will happen. Picture attached below.
I am testing a potential upgrade of some decoupling caps, and moved some wiring around associated with feeding power to the caps. Also, added a loop of bare wire on the right as a ground test point, and grounded a currently unused power trace. This stuff may not mean much now, but eventually I hope it will help improve performance.
Attachments
Last edited:
It would be in your personal preferences somewhere. I don't have access to that myself because it normally requires 'donation' user account perquisites.
To check if your page numbers are normal, the default is 10 posts per page. This post is #3327 on page 333, and the link to it is: https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-333.html#post5624667
The last post on this page will be #3330, etc.
Last edited:
It is in the 'Edit Options' page under the 'Your Control Panel' page, the 'Thread Display Options' section. I have mine set to 50 posts per page, which means a lot less clicking when going through long (and informative) threads such as this. BUT I know that DIY Audio post links I provide don't work for others and ones they provide don't work for me, so I always just provide the post number in the thread.
Greg in Mississippi
Greg in Mississippi
A small update on progress with the modded dac: I removed the high cost AK4137 board from the test setup and installed a new low cost AK4137 with the optional 768kHz PCM crystal option. It works fine at 768kHz, and this AK4137 board also includes a slave mode that high cost board does not have. It looks like AK4137 itself is capable of operating as a PCM slave to the dac clock signals, but not as a DSD slave. That might be okay. It raises the possibility of tweaking ES9038Q2M registers to make it run as master. Haven't looked at doing it any any detail yet. Since we are using 100MHz clocks, possibly we might have to use Clock_Gear register setting to divide that down to run the dac chip at 25MHz. The AK4137 could probably slave at the corresponding sample rate, although it would be non-standard. That might not be a bad thing actually. In that mode it might work a lot more like Katana, although a divided 100MHz clock might not be as low jitter as an undivided 24MHz clock. At least we could try it out and it might open the possibility of having more than one clock for ES9038Q2M, kind of similar to what Katana has.
The other thing is I still don't have a real good handle on how to get DPLL stable as I would like. I did leave it overnight last night and just checked it again now. It has been staying stable at 4 out of 10. It runs at 3 too, although doesn't sound as good as 4. At a setting of 2 it is not stable without some retuning.
Right now I am pushing on a place on the dac PCB (with a plastic stick held in a dial indicator holder) that got DPLL to stabilize where it is. No ferrites in there at this point. It looks like most of the tuning of DPLL stability is by flexing the board a little, but exactly how it works is rather confounding. Ferrites when they are in there seem to mostly help keep RF fields more contained. It's hard to shield effectively in there, but making it easy for EM flux to stay in the region seems to be having a similar end benefit.
I did try rerouting some power wires and putting in a few decoupling caps of known quality. Didn't really fix it, but it may be a little more stable when it does work. Probably will change out the remaining old decoupling caps next time I have the board out.
In the meantime I am getting the 1st modded dac board ready to go back in the test box to see if it behaves the same as the 2nd modded dac with respect to DPLL stability. Just changed the I2S pin header to gold. I already confirmed that both types of AK4137 boards seem to be about the same with respect to DPLL stability.
It strikes me that one possibility might be that with some flexing the dac board its clocking might be slightly pulled in frequency with respect to the 22HMz clock on the AK4137 board. It could be that when they are kind of synchronized, that DPLL is more stable. Its might be something like two grandfather clocks in the same room, eventually they will tend to pull into some kind of synchronicity with each other due to coupled vibrations interacting to reinforce such a mode. With clocks it might be the EM fields and or some cable coupling between the boards. If something like that, I don't know that I will be able to do much with it other than let them stabilize together and try not to otherwise disturb them. I don't have any reason to believe that over-coupling them would necessarily help more if something like natural-synchronization is happening.
While the foregoing is pretty speculative at this point, the behavior of where to push on the dac board to tune DPLL stability doesn't seem to make much sense by itself. Sensitive spots seem to move around and pressure to flex the board tends to cause it to slowly 'give' and over-flex thus de-tuning it. Weird, seems to kind of sum it up. Maybe if I had a really good frequency counter I could test the idea, but I don't have anything like that probably as stable as the dac clocks are. Be nice to have something stable and repeatable down to .1Hz or so (sometimes DPLL instability is rather slow). Doesn't necessarily have to be all that accurate just to see some correlation between clock frequencies syncing together, if in fact they are.
The other thing is I still don't have a real good handle on how to get DPLL stable as I would like. I did leave it overnight last night and just checked it again now. It has been staying stable at 4 out of 10. It runs at 3 too, although doesn't sound as good as 4. At a setting of 2 it is not stable without some retuning.
Right now I am pushing on a place on the dac PCB (with a plastic stick held in a dial indicator holder) that got DPLL to stabilize where it is. No ferrites in there at this point. It looks like most of the tuning of DPLL stability is by flexing the board a little, but exactly how it works is rather confounding. Ferrites when they are in there seem to mostly help keep RF fields more contained. It's hard to shield effectively in there, but making it easy for EM flux to stay in the region seems to be having a similar end benefit.
I did try rerouting some power wires and putting in a few decoupling caps of known quality. Didn't really fix it, but it may be a little more stable when it does work. Probably will change out the remaining old decoupling caps next time I have the board out.
In the meantime I am getting the 1st modded dac board ready to go back in the test box to see if it behaves the same as the 2nd modded dac with respect to DPLL stability. Just changed the I2S pin header to gold. I already confirmed that both types of AK4137 boards seem to be about the same with respect to DPLL stability.
It strikes me that one possibility might be that with some flexing the dac board its clocking might be slightly pulled in frequency with respect to the 22HMz clock on the AK4137 board. It could be that when they are kind of synchronized, that DPLL is more stable. Its might be something like two grandfather clocks in the same room, eventually they will tend to pull into some kind of synchronicity with each other due to coupled vibrations interacting to reinforce such a mode. With clocks it might be the EM fields and or some cable coupling between the boards. If something like that, I don't know that I will be able to do much with it other than let them stabilize together and try not to otherwise disturb them. I don't have any reason to believe that over-coupling them would necessarily help more if something like natural-synchronization is happening.
While the foregoing is pretty speculative at this point, the behavior of where to push on the dac board to tune DPLL stability doesn't seem to make much sense by itself. Sensitive spots seem to move around and pressure to flex the board tends to cause it to slowly 'give' and over-flex thus de-tuning it. Weird, seems to kind of sum it up. Maybe if I had a really good frequency counter I could test the idea, but I don't have anything like that probably as stable as the dac clocks are. Be nice to have something stable and repeatable down to .1Hz or so (sometimes DPLL instability is rather slow). Doesn't necessarily have to be all that accurate just to see some correlation between clock frequencies syncing together, if in fact they are.
Any recommendations on these ES9038Q2M Chinese Boards.
#1
I see there are many variations on these design and that is very popular on here, is it still worth it.
Lusya ES9038Q2M I2S IIS DSD DOP Coaxial Fiber SPDIF Digital Audio DAC Decoder Board support 32bit 384k/DSD64 DSD128 DSD256-in Digital-to-Analog Converter from Consumer Electronics on Aliexpress.com | Alibaba Group
#2
This design looks simpler, with better output connectors.
K.GUSS ES9038Q2M ES9038 I2S Input Decoders Mill Board DAC Decoding Board For Amplifier AMP-in DAC from Consumer Electronics on Aliexpress.com | Alibaba Group
#3
I like that it has Wima capacitors, heatsinks, but not crazy on the digital inputs.
Aliexpress.com : Buy Lusya ES9038 Q2M I2S DSD Decoder Coaxial Fiber input DAC decoding board For hifi amplifier audio A1 011 from Reliable Digital-to-Analog Converter suppliers on LusyaTechnology Store
#4
This is the upgraded version of #3, it has a cheap power switch on the back but it can be modified easily. i'm leaning more towards this one.
Aliexpress.com : Buy Lusya Upgraded version ES9038 Q2M I2S DSD Decoder Coaxial Fiber input DAC decoding board For hifi amplifier audio T0144 from Reliable Digital-to-Analog Converter suppliers on LusyaTechnology Store
i had an experience with a ES9028Q2M board, i couldn't adjust the volume, i try everything... at the end it ended on the trash.
#1
I see there are many variations on these design and that is very popular on here, is it still worth it.
Lusya ES9038Q2M I2S IIS DSD DOP Coaxial Fiber SPDIF Digital Audio DAC Decoder Board support 32bit 384k/DSD64 DSD128 DSD256-in Digital-to-Analog Converter from Consumer Electronics on Aliexpress.com | Alibaba Group
#2
This design looks simpler, with better output connectors.
K.GUSS ES9038Q2M ES9038 I2S Input Decoders Mill Board DAC Decoding Board For Amplifier AMP-in DAC from Consumer Electronics on Aliexpress.com | Alibaba Group
#3
I like that it has Wima capacitors, heatsinks, but not crazy on the digital inputs.
Aliexpress.com : Buy Lusya ES9038 Q2M I2S DSD Decoder Coaxial Fiber input DAC decoding board For hifi amplifier audio A1 011 from Reliable Digital-to-Analog Converter suppliers on LusyaTechnology Store
#4
This is the upgraded version of #3, it has a cheap power switch on the back but it can be modified easily. i'm leaning more towards this one.
Aliexpress.com : Buy Lusya Upgraded version ES9038 Q2M I2S DSD Decoder Coaxial Fiber input DAC decoding board For hifi amplifier audio T0144 from Reliable Digital-to-Analog Converter suppliers on LusyaTechnology Store
i had an experience with a ES9028Q2M board, i couldn't adjust the volume, i try everything... at the end it ended on the trash.
#1 is the only one I might consider. Don't know how good it might be as a candidate for modding. #3 and #4 have switching power supplies a very poor ground return path between the output stage and the dac chip, and cheap SMPS always sound awful, IMHO. Looks like #2 is one without a volume control. In fact, looks like maybe only #1 has volume control and ability to select filters.
Getting back to #1, I think one or more people in the thread may have one like that. Don't know how easy it might be to mod if there were interest in that. People seem to prefer the type with optional display for that. It may save having to do pin lifting to get access to dac registers, again, if that sort of thing were of interest.
If not interested in modding then they are all junk, IMHO. I would throw them in the trash or maybe save them for spare parts, but couldn't stand listening to one for more than a few seconds. Some people don't seem to mind so much though. YMMV.
Last edited:
Thanks Mark, Yeah yo are right #1 seems to be the only worthy option, i'm also looking at the Khadas Tone Board, it measures very good, similar to benchmark Dac's, Topping etc.. For only $99
Review and Measurements of WesionTEK Khadas Tone Board DAC | Audio Science Review (ASR) Forum
have you try this board, the khadas?
No. Looked at the pictures carefully, some time ago. Not that impressed in an absolute sense, although it seems likely a good deal for the price. There is no way it can compare with Benchmark in sound quality, I can promise you.
Measurements claimed to show something like Khadas is nearly like Benchmark in sound quality or something to similar effect would seem to strain credulity. That's trying to say it in a nice way.
Last edited:
Now that I have a dac with sockets in the output stage it might be possible to roll opamps. However, I don't have any of those in stock, and I have my own ideas about what I want to work on next. If I was dissatisfied with the output stage sound then I might have more interest.
Terry, you have so many ideas you would like to try and things to test, why don't you join in with us and mod your own dac like we are doing? If you catch up with where we are now, you could satisfy your curiosity and maybe even have some extra time to work on trying things I would like to suggest.
Mark, I've been tempted many times to grab one of these boards, start
having some fun and contribute. However I've got to be smart with DAC
development time, it can be an endless hole
The new AKM chip is coming soon, so I'm pretty sure I'll be going down that
path. I think it will be a much better proposition on a lot of fronts and just
for the way I like to design and build analog stages.
T
Okay, but what is it about the new AKM chip that makes it so appealing? What 'fronts' are you thinking of? After all, there is still work to do here as I think you must know. Don't you think the diy community could use to have this project left off in a more fully developed state?
Last edited:
Board from the topic can be bought here:
NEW ES9038 ES9038Q2M DAC Decoder board Support IIS DSD DOP 384KHz
I ordered version 1 without a switch. The board came with a switch.
dumohwow,
Thank you for the link to the board like we use. That one looks kind of expensive though with the plastic case. If you are planning to mod the board such as with the output stage board and AVCC board, you might find that the dac board no longer fits in the case. Oh, well. At least you have a free switch.
Thank you for the link to the board like we use. That one looks kind of expensive though with the plastic case. If you are planning to mod the board such as with the output stage board and AVCC board, you might find that the dac board no longer fits in the case. Oh, well. At least you have a free switch.
Interesting day today. But, not so much the good kind of interesting. Took the test board out of the steel case and replaced remaining decoupling caps around the dac chip and clock. Added a little more HF filter capacitance where 5v comes into the dac board. Also, trying out a new block of film caps on the +-15v.
Made the changes, didn't expect much, maybe a slight improvement in DPLL stability from replacing the remaining decoupling caps. However, when I turned it on it sounded worse than before. More bright, scratchy, distorted, metallic, splashy, all those bad things poor quality sigma delta dacs are often associated with. Decided to try reducing DPLL bandwidth, but couldn't get it to stabilize at even 5 out of 10. Figured maybe it didn't like the new decoupling caps on the clock or something. That's okay, I still have the old ones, but what's going on?
Continued to work with it trying to tune DPLL stability with the plastic stick. Some progress there, but not much. It seems considerably more unstable than before. Since pushing down on the dac board with the plastic stick wasn't helping much, I decided to pull up on it a little bit. Ah, ha. A little improvement there. Tried removing a screw that holds the whole test board assembly in the steel server case and lift up one corner of the test board. More improvement. Lightly touched a finger of one hand to one of the clocks on the AK4137 board, and lightly touched another finger of the same hand on dac board clock, all while holding up the corner of the test board. Bingo! Tuned in good sound. However, how soft I touch the clocks, and how I lift up the test board assembly in the corner is all affecting sound quality in a very sensitive way. Small movements make big changes. Hmmm. Also, some of what I hear when DPLL is very out of tune and SQ is bad is a lot like it sounded when I tried swapping AK4137 clocks with supposedly lower jitter ones.
Obviously, there is at least a problem with uncontrolled, poorly defined RF EM fields interacting between the case and the two boards with clocks. Likely a problem due to working with only 2-sided boards and 100MHz and other RF frequencies radiating around, reflecting, and coupling together. Of course, it is confounding to me because there are so many different reflections and sensitive areas on the circuit boards.
Plan for tomorrow morning will be to line the interior of the steel case with copper foil and see what that does. Will also experiment with sensitivity of circuit board distances to case walls. May try to put some absorptive material in there if I can figure out something that works. If my fingers can make a difference, so should a rubber glove filled with salt water (or possibly fairly close).
Although I could probably get this thing working each time I take it out and put it back in, the whole thing is ridiculous. There needs to be some way to get best sound quality working more predictably for all mod builders all the time, without having to fiddle around each time because the RF fields are poorly defined/controlled in space, and exactly how they are interacting to affect sound quality (for better or for worse) is not well understood.
Made the changes, didn't expect much, maybe a slight improvement in DPLL stability from replacing the remaining decoupling caps. However, when I turned it on it sounded worse than before. More bright, scratchy, distorted, metallic, splashy, all those bad things poor quality sigma delta dacs are often associated with. Decided to try reducing DPLL bandwidth, but couldn't get it to stabilize at even 5 out of 10. Figured maybe it didn't like the new decoupling caps on the clock or something. That's okay, I still have the old ones, but what's going on?
Continued to work with it trying to tune DPLL stability with the plastic stick. Some progress there, but not much. It seems considerably more unstable than before. Since pushing down on the dac board with the plastic stick wasn't helping much, I decided to pull up on it a little bit. Ah, ha. A little improvement there. Tried removing a screw that holds the whole test board assembly in the steel server case and lift up one corner of the test board. More improvement. Lightly touched a finger of one hand to one of the clocks on the AK4137 board, and lightly touched another finger of the same hand on dac board clock, all while holding up the corner of the test board. Bingo! Tuned in good sound. However, how soft I touch the clocks, and how I lift up the test board assembly in the corner is all affecting sound quality in a very sensitive way. Small movements make big changes. Hmmm. Also, some of what I hear when DPLL is very out of tune and SQ is bad is a lot like it sounded when I tried swapping AK4137 clocks with supposedly lower jitter ones.
Obviously, there is at least a problem with uncontrolled, poorly defined RF EM fields interacting between the case and the two boards with clocks. Likely a problem due to working with only 2-sided boards and 100MHz and other RF frequencies radiating around, reflecting, and coupling together. Of course, it is confounding to me because there are so many different reflections and sensitive areas on the circuit boards.
Plan for tomorrow morning will be to line the interior of the steel case with copper foil and see what that does. Will also experiment with sensitivity of circuit board distances to case walls. May try to put some absorptive material in there if I can figure out something that works. If my fingers can make a difference, so should a rubber glove filled with salt water (or possibly fairly close).
Although I could probably get this thing working each time I take it out and put it back in, the whole thing is ridiculous. There needs to be some way to get best sound quality working more predictably for all mod builders all the time, without having to fiddle around each time because the RF fields are poorly defined/controlled in space, and exactly how they are interacting to affect sound quality (for better or for worse) is not well understood.