I'm looking at the moment but I need a board with an Ethernet PHY that supports hardware packet timestamping so that I can use it for source duties for synchronized network streaming. I've found some candidates but of course they are more expensive.
And I have a toddler so free time is very limited
Thanks Zek
yes it is in the other thread. there i want to have a cross-check DAC for my modification on the CM6631A+9023 boards.
sorry for the confusion...
thanks mark i will do that but i don´t believe they will react for the win10 driver.......will see.
if a wonder is ahppend and the DAc is working...i will start an extra thread for modification...
chris
yes it is in the other thread. there i want to have a cross-check DAC for my modification on the CM6631A+9023 boards.
sorry for the confusion...
thanks mark i will do that but i don´t believe they will react for the win10 driver.......will see.
if a wonder is ahppend and the DAc is working...i will start an extra thread for modification...
chris
I'm looking at the moment but I need a board with an Ethernet PHY that supports hardware packet timestamping so that I can use it for source duties for synchronized network streaming. I've found some candidates but of course they are more expensive.
And I have a toddler so free time is very limited
Interesting. Just wondering, if receiver nodes are receiving time stamped audio, how closely to their real time clocks have to stay in sync? Just thinking about NTP network latency. Maybe a local dedicated NTP server in the middle of the node cluster somewhere?
I like this option, it suits me much better as a guy coming from software. Also it could, in theory, allow me to do functional tests.
For the record, I'm also interested in this project and in participating. There seems to be a whole toolchain which I'll need to be purchasing, which is holding me up a little bit at the moment.
May be it's posted here before but this thread contains a lot of interesting information regarding the ESS dac chips and their "hump" in THD.
Oops! We ran into some problems. | Audio Science Review (ASR) Forum
Oops! We ran into some problems. | Audio Science Review (ASR) Forum
eljoantonyn,
Interesting information perhaps, but not all that useful, IMHO. The issue seems to be characteristic to the dac chips themselves, and there is only so much that can be done to minimize distortion by dac end-use product designers, especially given the price point some dacs have to sell at.
Interesting information perhaps, but not all that useful, IMHO. The issue seems to be characteristic to the dac chips themselves, and there is only so much that can be done to minimize distortion by dac end-use product designers, especially given the price point some dacs have to sell at.
Hi Mark,
I was busy the last time. I think on this page the jumper of the board are explained better here:
AK4137 Raspberry Pi IIS DAC Decoder Board Digital Player Netzwerk player Unterstutzung I2S 32bit 384 karat DSD256 Fur Raspberry Pi 2B 3B 3B + in AK4137 Raspberry Pi IIS DAC Decoder Board Digital Player Netzwerk player Unterstutzung I2S 32bit 384 karat DSD256 Fur Raspberry Pi 2B 3B 3B + aus Verstarker auf AliExpress.com | Alibaba Group
You can set input and optput signal type and the kind of the filter. What do you think is missing here?
And how do I have to connect such a board to the ES9038Q2M board. the AK4137 has a MCLK output the ES9038Q2M Board not. Should I leave the MCLK signal from the AK4137 open?
Hi Ampi,
It looks a lot better with the jumper settings explained. Although it does not make accessible all possible AK4137 settings as described in the AK4137 data sheet, it seems offer enough settings for most purposes.
One reason I recommended using AK4137 with the dacs we typically mod in this thread is because it leaves open the possibility upsample and convert to DSD things like SPDIF inputs as the more expensive AK4137 boards can do. Aside from that reason, upsampling between the source device and the dac board can help to reduce jitter.
It seems to me that the situation when using RPi might be a little different. RPi I2S clocking is known to be jittery, but reclocker boards, such as Allo Kali, are available to address that issue. Besides reclockers, there also FIFO boards made to work with RPi, and they can work very well to reduce jitter in music player applications.
More specifically, I wonder when using RPi with one of our modded dacs, if a better solution in that case might be to use something like Allo Kali reclocker (if it supports DSD, not sure), or perhaps Iancanada FIFO Pi, and then upsample and convert audio files to DSD in software such as HQplayer and Rune can do, and other software can do. The quality of the upsampling and conversion to DSD might be better if processed offline using however much time it takes to render output files. Because AK4137 has to do what it does in real-time, that puts some constraints on how it has to work (constraints that don't apply to offline processing). If the dac will only be used for listening to your music files, one of the above choices might be a better way to go.
On the other hand that might not be so good for a general purpose dac, such as what I have been working on. Also, I find I get the best results with AK4137 if it is clocked in a somewhat complicated way, something which may be impractical with the RPi AK4137 board.
I guess the bottom line is if all things considered you want an AK4137 for whatever reason like the price is good and it does what what you want, then I don't see any reason not to use it with RPi.
EDIT: I would add that I found replacing the output pin headers on my AK4137 boards with high quality gold flashed, or gold plated pin headers, and using gold ribbon cable pin connectors helped to keep jitter to a minimum. Other possibilities might include soldering wires to the bottom of the pin connectors and don't use the nickel pins at all. I would still recommend using ribbon cable with every other wire a ground for best signal quality to the dac board. The dac board should also use gold pin headers or soldered connections. More info at:
ES9038Q2M Board - Page 204 - diyAudio
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-305.html#post5582674
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-305.html#post5585394
In addition to all the foregoing, if you can keep jitter to a minimum, you may be able to futher improve sound quality by reducing DPLL bandwidth in the dac registers, assuming you have hooked up an Arduino or other device to take control of the dac board I2C bus.
It looks a lot better with the jumper settings explained. Although it does not make accessible all possible AK4137 settings as described in the AK4137 data sheet, it seems offer enough settings for most purposes.
One reason I recommended using AK4137 with the dacs we typically mod in this thread is because it leaves open the possibility upsample and convert to DSD things like SPDIF inputs as the more expensive AK4137 boards can do. Aside from that reason, upsampling between the source device and the dac board can help to reduce jitter.
It seems to me that the situation when using RPi might be a little different. RPi I2S clocking is known to be jittery, but reclocker boards, such as Allo Kali, are available to address that issue. Besides reclockers, there also FIFO boards made to work with RPi, and they can work very well to reduce jitter in music player applications.
More specifically, I wonder when using RPi with one of our modded dacs, if a better solution in that case might be to use something like Allo Kali reclocker (if it supports DSD, not sure), or perhaps Iancanada FIFO Pi, and then upsample and convert audio files to DSD in software such as HQplayer and Rune can do, and other software can do. The quality of the upsampling and conversion to DSD might be better if processed offline using however much time it takes to render output files. Because AK4137 has to do what it does in real-time, that puts some constraints on how it has to work (constraints that don't apply to offline processing). If the dac will only be used for listening to your music files, one of the above choices might be a better way to go.
On the other hand that might not be so good for a general purpose dac, such as what I have been working on. Also, I find I get the best results with AK4137 if it is clocked in a somewhat complicated way, something which may be impractical with the RPi AK4137 board.
I guess the bottom line is if all things considered you want an AK4137 for whatever reason like the price is good and it does what what you want, then I don't see any reason not to use it with RPi.
EDIT: I would add that I found replacing the output pin headers on my AK4137 boards with high quality gold flashed, or gold plated pin headers, and using gold ribbon cable pin connectors helped to keep jitter to a minimum. Other possibilities might include soldering wires to the bottom of the pin connectors and don't use the nickel pins at all. I would still recommend using ribbon cable with every other wire a ground for best signal quality to the dac board. The dac board should also use gold pin headers or soldered connections. More info at:
ES9038Q2M Board - Page 204 - diyAudio
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-305.html#post5582674
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-305.html#post5585394
In addition to all the foregoing, if you can keep jitter to a minimum, you may be able to futher improve sound quality by reducing DPLL bandwidth in the dac registers, assuming you have hooked up an Arduino or other device to take control of the dac board I2C bus.
Last edited:
Hi Mark,
I finally got around to mess with my so far unsuccessful project with this new v 1.07 board. My problem was that when I hooked power to the board the neg rail on the 15 vdc psu would drop to zero, the right rail drops to 13 vdc. The trafo would also start getting warm. I took the board out today and did some continuity checks on the board and found that the neg input to the board had a short to ground. I also noticed that this board had the J6 jumper already soldered in where as on my first board that did work until I killed it by accident did not. I removed the jumper, problem solved. Reassembled everything and have it singing again. The only mods done so far is replacing opamp, adding 3300 uF to the AVCC rails, added a WavIO usb to I2S board input into the I2S inputs, and added a bunch of film caps on the PSU rails. Using LT1083 based PSU. Sounds pretty good, I'll audition it against my other dacs when I get the chance.
Paul
I finally got around to mess with my so far unsuccessful project with this new v 1.07 board. My problem was that when I hooked power to the board the neg rail on the 15 vdc psu would drop to zero, the right rail drops to 13 vdc. The trafo would also start getting warm. I took the board out today and did some continuity checks on the board and found that the neg input to the board had a short to ground. I also noticed that this board had the J6 jumper already soldered in where as on my first board that did work until I killed it by accident did not. I removed the jumper, problem solved. Reassembled everything and have it singing again. The only mods done so far is replacing opamp, adding 3300 uF to the AVCC rails, added a WavIO usb to I2S board input into the I2S inputs, and added a bunch of film caps on the PSU rails. Using LT1083 based PSU. Sounds pretty good, I'll audition it against my other dacs when I get the chance.
Paul
I'm a bit late to this 9038 party, so thanks to all that have supplied so much valuable information.
I've just bought two of the 1.07 boards and I'm going to use these as a DAC 'replacement' for my MiniDSP 2x4HD using the MiniDSP's I2S output. I've mocked up a PCM5122 proof of concept using two old Raspberry PI IQAudio PI-DACs and it worked so now the work needs to be repeated for these 9038 boards.
I've got dual linear toroidal/LM317/337 PSU powering the 9038 boards at the specified 15V, my question is, is there any issue with dropping the voltage to 12V so I can use my overrated PSU to power the MiniDSP (a 12v device) and the two 9038 boards? I've dialled the LM317/337 back to 12V and it's powering the DAC fine, no dropouts, no apparent loss of quality.
So far, I've just replaced the op-amps with a pair LM4562s and will replace the AMS1117 with a suitable LDO in the next few days. After that I'll be doing a little bit of reading back over the last 415 pages.
I've just bought two of the 1.07 boards and I'm going to use these as a DAC 'replacement' for my MiniDSP 2x4HD using the MiniDSP's I2S output. I've mocked up a PCM5122 proof of concept using two old Raspberry PI IQAudio PI-DACs and it worked so now the work needs to be repeated for these 9038 boards.
I've got dual linear toroidal/LM317/337 PSU powering the 9038 boards at the specified 15V, my question is, is there any issue with dropping the voltage to 12V so I can use my overrated PSU to power the MiniDSP (a 12v device) and the two 9038 boards? I've dialled the LM317/337 back to 12V and it's powering the DAC fine, no dropouts, no apparent loss of quality.
So far, I've just replaced the op-amps with a pair LM4562s and will replace the AMS1117 with a suitable LDO in the next few days. After that I'll be doing a little bit of reading back over the last 415 pages.
@damiangt3
+-12v is okay for the opamp output stage. Regarding the AMS1117, may as well just leave it there to provide power to the MCU. The important subsystems should be disconnected from the existing 3.3v bus and be given their own dedicated 3.3v regulators to power each one. The most critical power supply is for AVCC_L and AVCC_R. Some information about recommmended circuitry can be found on the ESS download page, such as: http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
Also, I will attach a long list that could use some paring down of posts of possible interest. The ones going way back to the beginning may not be the most useful.
+-12v is okay for the opamp output stage. Regarding the AMS1117, may as well just leave it there to provide power to the MCU. The important subsystems should be disconnected from the existing 3.3v bus and be given their own dedicated 3.3v regulators to power each one. The most critical power supply is for AVCC_L and AVCC_R. Some information about recommmended circuitry can be found on the ESS download page, such as: http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
Also, I will attach a long list that could use some paring down of posts of possible interest. The ones going way back to the beginning may not be the most useful.
Last edited:
Hi Mark, and everyone interested,
I burned my project in for about 10 hours before giving it a listen on a decent sound system. I compared it directly to a AK4490 build from JL Sounds usb to ds. The AK4490 has separate LT1083 psu's, one for the DAC and one for the xmos based isolated async I2S converter. also a bog standard =/- 15 v lm 317/337 dual supply psu for the lme 49710 based output stage board.
My ES9038 build includes a wavIO xmos based isolated async usb to I2S board with it's own separate LT1083 psu. So this gives me three input options which I like. The ES9038q2m sounded pretty good in the garage using optical spdif in, but when I hooked it and gave it a listed using USB input and Foobar Asio (32 bit) playback it didn't sound very involving at all. Mids seemed thin and highs unnatural with poor decay. I was disappointed because I thought it sounded really good in the garage so I switched to coax input (24 bit), much, much better. I kind of remember trying the wavio with a Wolfson dac several years ago and not liking much it which was why it was in my parts bin in the first place. I'll need to take a closer look at the wavio to see if I have something set up wrong, many folks seem to like it.
It was much closer with coax input to the AK4490, but couldn't match the realism and liquidity of the AK4490. At this point the ES9038 build is still cheaper than the AK4490 if you discount the usb to I2S board, but it's getting close cost wise. You get all of the DIY fun and frustration with the ES9038 board, but a superior engineered plug and play product with the AK4490.
The only other mod I am considering doing is a separate psu for the avcc. That I think I can handle without blowing up the board. After that I'll wait until someone comes out with a nicely engineered ES9038 board for a reasonable price that I can drop into my build (hint, hint).
Paul
I burned my project in for about 10 hours before giving it a listen on a decent sound system. I compared it directly to a AK4490 build from JL Sounds usb to ds. The AK4490 has separate LT1083 psu's, one for the DAC and one for the xmos based isolated async I2S converter. also a bog standard =/- 15 v lm 317/337 dual supply psu for the lme 49710 based output stage board.
My ES9038 build includes a wavIO xmos based isolated async usb to I2S board with it's own separate LT1083 psu. So this gives me three input options which I like. The ES9038q2m sounded pretty good in the garage using optical spdif in, but when I hooked it and gave it a listed using USB input and Foobar Asio (32 bit) playback it didn't sound very involving at all. Mids seemed thin and highs unnatural with poor decay. I was disappointed because I thought it sounded really good in the garage so I switched to coax input (24 bit), much, much better. I kind of remember trying the wavio with a Wolfson dac several years ago and not liking much it which was why it was in my parts bin in the first place. I'll need to take a closer look at the wavio to see if I have something set up wrong, many folks seem to like it.
It was much closer with coax input to the AK4490, but couldn't match the realism and liquidity of the AK4490. At this point the ES9038 build is still cheaper than the AK4490 if you discount the usb to I2S board, but it's getting close cost wise. You get all of the DIY fun and frustration with the ES9038 board, but a superior engineered plug and play product with the AK4490.
The only other mod I am considering doing is a separate psu for the avcc. That I think I can handle without blowing up the board. After that I'll wait until someone comes out with a nicely engineered ES9038 board for a reasonable price that I can drop into my build (hint, hint).
Paul
Paul,
Now maybe you are starting to see why In consider the unmodded ES9038Q2M boards to sound awful in their stock form. Almost everything needs some fixing, but at least dac chip itself is quite usable, and board has a nice ground plane that lends itself pretty well to modding. Depends if you want to get into it that deep or not.
For nearly finished products, Allo Katana ES9038Q2M RPi hat dac can be very good, but by the time you fix it up with good linear power supplies the cost looks a lot closer to $500. Still it handily beats most or perhaps all other options at that price range and competes very well sound quality-wise with $1,000+ dacs. Iancanada sells parts to make dual mono ES9038Q2M dacs, but if you buy all the odds and ends for a recommended build from Ian, there is some assembly (including oven soldering) to do, and the costs can add up quickly probably to as much or more than Katana (although I didn't add everything up to be sure exactly). Sound quality can be pretty similar to Katana, with one enthusiast who has Katana, Iancanada, and TP dac setups subjectively rating Katana as perhaps slightly favored vs the others at one point in time (subject to change of course as dac offerings continue to change and improve).
Included in what I would refer to as the semi-kit dac market is TP which keeps improving the quality if its offerings. Not so cheap after all the parts and pieces are ordered, and considering that PRO version Sabre chips require more power supplies (vs Q2M mobile chips), with lower noise being one possible benefit. Some soldering is required for some parts. Also, some reading of multiple threads required to collect all the necessary information. With any of the above options one can end up with a very nice dac. TP and Iancanada dacs more customizable vs Katana. Alas, there is no free lunch, nor any cheap and really good dac. There is at least one or perhaps more low cost dac boards that measure quite well, but I would not classify any in that group as equivalent in sound quality to the ones I mentioned specifically.
Looked at in the above light, modding an ES9038Q2M dac is one way to end up with pretty good dac, but there is a whole lot of work involved. Fully modded I would say it can beat Katana on SQ, IMHO of course, but nobody fully mods one so moot point in reality. Maybe easier than building a ship in a bottle, not sure. At least there is the option to stop modding at any point if the dac is deemed to sound good enough, or if one has had enough of dac modding.
One area that I haven't really finished exploring is to see if there is a more cheap and easy way to get AK4137 sounding its best if used with the dac board.
Also, we have never fully explored in this thread how to make a synchronous mode music player dac, either an RPi hat, or a USB version (although occip did provide a driver for use with RPi and a wiring hookup diagram, IIRC).
Now maybe you are starting to see why In consider the unmodded ES9038Q2M boards to sound awful in their stock form. Almost everything needs some fixing, but at least dac chip itself is quite usable, and board has a nice ground plane that lends itself pretty well to modding. Depends if you want to get into it that deep or not.
For nearly finished products, Allo Katana ES9038Q2M RPi hat dac can be very good, but by the time you fix it up with good linear power supplies the cost looks a lot closer to $500. Still it handily beats most or perhaps all other options at that price range and competes very well sound quality-wise with $1,000+ dacs. Iancanada sells parts to make dual mono ES9038Q2M dacs, but if you buy all the odds and ends for a recommended build from Ian, there is some assembly (including oven soldering) to do, and the costs can add up quickly probably to as much or more than Katana (although I didn't add everything up to be sure exactly). Sound quality can be pretty similar to Katana, with one enthusiast who has Katana, Iancanada, and TP dac setups subjectively rating Katana as perhaps slightly favored vs the others at one point in time (subject to change of course as dac offerings continue to change and improve).
Included in what I would refer to as the semi-kit dac market is TP which keeps improving the quality if its offerings. Not so cheap after all the parts and pieces are ordered, and considering that PRO version Sabre chips require more power supplies (vs Q2M mobile chips), with lower noise being one possible benefit. Some soldering is required for some parts. Also, some reading of multiple threads required to collect all the necessary information. With any of the above options one can end up with a very nice dac. TP and Iancanada dacs more customizable vs Katana. Alas, there is no free lunch, nor any cheap and really good dac. There is at least one or perhaps more low cost dac boards that measure quite well, but I would not classify any in that group as equivalent in sound quality to the ones I mentioned specifically.
Looked at in the above light, modding an ES9038Q2M dac is one way to end up with pretty good dac, but there is a whole lot of work involved. Fully modded I would say it can beat Katana on SQ, IMHO of course, but nobody fully mods one so moot point in reality. Maybe easier than building a ship in a bottle, not sure. At least there is the option to stop modding at any point if the dac is deemed to sound good enough, or if one has had enough of dac modding.
One area that I haven't really finished exploring is to see if there is a more cheap and easy way to get AK4137 sounding its best if used with the dac board.
Also, we have never fully explored in this thread how to make a synchronous mode music player dac, either an RPi hat, or a USB version (although occip did provide a driver for use with RPi and a wiring hookup diagram, IIRC).
Last edited:
I Thank God I still have my turntable. In many ways DACs still haven't fully captured their entertainment value. I do like TP products , they do nice work, I own a dual mono Opus DAC of theirs, still sounds great. My issue like many others is my limited budget to indulge my audio hobby, so looking for a deal is tempting.
for <100USD the only sensible ES9038q2m option is Khadas Toneboard. very well designed 4 layer board. good clocks. virtually unmodable. add linear +5V PS. all schematics provided. put it in a nice box.