Hello all, I'm new here and this is going to be my project thread. As I briefly stated in my intro here I am looking to build a DAC to use with a power amp that closely mimics the Oppo Sonica DAC, but after reading so many threads here, I now believe there is a possibility to build something even better (somewhat based on this). So after processing the overwhelming amount of options and information overload, I have decided to do this project in two phases.
Phase 1 Goal: Use something off the shelf like the three options available on ebay/aliexpress and replace all possible components with better alternatives.
Phase 2 Goal: Pursue building something from scratch (either using something from audiodesignguide.com or scratch if I can find the help/resources).
For Phase 1, I need help with the following. I am reasonably familiar with electronic components but software is really my strong suite (if I ever get to the point of improving firmware).
1) Which of the three options here present a good starting point or has the best layout of the board to derive the most out of the ES9038PRO.
2) Based on 1, next would be what parts to swap out with better alternatives? (From reading around here, most obvious things would be clocks, caps and op-amps?)
Phase 1 Goal: Use something off the shelf like the three options available on ebay/aliexpress and replace all possible components with better alternatives.
Phase 2 Goal: Pursue building something from scratch (either using something from audiodesignguide.com or scratch if I can find the help/resources).
For Phase 1, I need help with the following. I am reasonably familiar with electronic components but software is really my strong suite (if I ever get to the point of improving firmware).
1) Which of the three options here present a good starting point or has the best layout of the board to derive the most out of the ES9038PRO.
2) Based on 1, next would be what parts to swap out with better alternatives? (From reading around here, most obvious things would be clocks, caps and op-amps?)
Attachments
Probably better to use ES9028PRO. Likely more practical to get better sound quality out of that verses ES9038PRO due to complications with powering AVCC in the latter chip.
Regarding Chinese dac boards using ESS PRO chips, most or all of them tie AVCC_L and AVCC_R together causing problems with left/right channel separation and problems powering AVCC with the best sounding power supply solutions (even for ES9028PRO).
If 8-channel outputs are wanted, don't know of any of the Chinese dacs offering that.
Also, costs of buying a board, replacing most of the parts and most all of the power supplies, buying a case, power transformer(s), etc., gets expensive by the time everything is done. May as well consider how much one can afford to spend on a dac, what it is to be used for, what level of sound quality is required, etc., first.
Another factor might be how much work one is willing to do to have a completed dac meeting the goals.
Often buying and modding a Chinese dac board is not the best solution. At least, one should know what one is really getting into. Lots of people start projects and don't follow through to the end. Sometimes in the end there is a working device that doesn't meet the original goals, sometimes not. Sometimes it can work out well, depends.
Regarding Chinese dac boards using ESS PRO chips, most or all of them tie AVCC_L and AVCC_R together causing problems with left/right channel separation and problems powering AVCC with the best sounding power supply solutions (even for ES9028PRO).
If 8-channel outputs are wanted, don't know of any of the Chinese dacs offering that.
Also, costs of buying a board, replacing most of the parts and most all of the power supplies, buying a case, power transformer(s), etc., gets expensive by the time everything is done. May as well consider how much one can afford to spend on a dac, what it is to be used for, what level of sound quality is required, etc., first.
Another factor might be how much work one is willing to do to have a completed dac meeting the goals.
Often buying and modding a Chinese dac board is not the best solution. At least, one should know what one is really getting into. Lots of people start projects and don't follow through to the end. Sometimes in the end there is a working device that doesn't meet the original goals, sometimes not. Sometimes it can work out well, depends.
Thanks for the reply and advice!
"At least, one should know what one is really getting into." -> THIS, would be great. I do want to learn how these chips work and how to build a solution around them to get the most out of it. Any learning resources that can help enlighten me would be great!
Let me try and expand on the goals here after considering some points above.
Overall, the goal is to learn something new and build something I want and will use. I do want to build a good DAC because not only am I picky about quality (not quite audiophile grade though), whatever I end up making will be hooked up to expensive equipment (Marantz amp and possibly Ascend Acoustics Sierra Towers) so it better sound good. I do have a good size CD collection that will be digitized to some lossless format and I occasionally stream via Spotify. Budget wise, I would like to keep the expenditure leaning more towards quality components around the DAC and less towards the case/other cosmetics etc.
If I manage to learn enough to build my very own board around one of these chips, that would be the ultimate win, but for a short term win, I'm trying to find something that I can get working soon. Heck, if someone has already done this and can share schematics with me, I would be elated to get the board printed and get soldering. I do want to get some hands on time with the hardware and ideally also some time customizing MCU firmware. After reading your suggestion about going with a 9028, I found this kit that fits the bill, but lacks XLR/balanced outs
Could still be worth it.
My daily job though exciting, does have its dull phases and I keep myself challenged by taking up projects that are completely unrelated to work. As and example, the few years ago I took up restoring a vintage motorcycle, a 1972 Honda CB750 K2. Knew next to nothing when I started but learned a LOT and quick. That was loads of fun and I learned so many new things, made great friends worldwide. The bike went on to win Bike of the year that year. Hopefully that also addresses the points on willing to work, following through and meeting the original goals
"At least, one should know what one is really getting into." -> THIS, would be great. I do want to learn how these chips work and how to build a solution around them to get the most out of it. Any learning resources that can help enlighten me would be great!
Let me try and expand on the goals here after considering some points above.
Overall, the goal is to learn something new and build something I want and will use. I do want to build a good DAC because not only am I picky about quality (not quite audiophile grade though), whatever I end up making will be hooked up to expensive equipment (Marantz amp and possibly Ascend Acoustics Sierra Towers) so it better sound good. I do have a good size CD collection that will be digitized to some lossless format and I occasionally stream via Spotify. Budget wise, I would like to keep the expenditure leaning more towards quality components around the DAC and less towards the case/other cosmetics etc.
If I manage to learn enough to build my very own board around one of these chips, that would be the ultimate win, but for a short term win, I'm trying to find something that I can get working soon. Heck, if someone has already done this and can share schematics with me, I would be elated to get the board printed and get soldering. I do want to get some hands on time with the hardware and ideally also some time customizing MCU firmware. After reading your suggestion about going with a 9028, I found this kit that fits the bill, but lacks XLR/balanced outs
Could still be worth it.My daily job though exciting, does have its dull phases and I keep myself challenged by taking up projects that are completely unrelated to work. As and example, the few years ago I took up restoring a vintage motorcycle, a 1972 Honda CB750 K2. Knew next to nothing when I started but learned a LOT and quick. That was loads of fun and I learned so many new things, made great friends worldwide. The bike went on to win Bike of the year that year
. Hopefully that also addresses the points on willing to work, following through and meeting the original goals The kit you found has all the problems I mentioned, plus it does not include the cost of the dac chip. If you buy the kit, the dac chip may not come soldered to the board. You can throw away all the capacitors, the voltage regulators, and the op amps. Probably the resistors too. You will have to live with crippled stereo since AVCC_L and AVCC_R are tied together and the tie is buried inside a multi-layer PCB. That is a project very much not recommended for beginners.
One of the other problems with trying to learn about ESS dacs is that ESS keeps all information about them tightly restricted. One cannot get a data sheet without signing a nondisclosure agreement that is very restrictive.
Also, you did not say how much you want to spend on this project. If you have around $1,000 (or a bit more) and maybe 80 hours of time, you could go to a much better dac, namely AK4499. If you have $500 you might have some trouble putting together a decent RPi dac system based on ES9038Q2M.
One of the other problems with trying to learn about ESS dacs is that ESS keeps all information about them tightly restricted. One cannot get a data sheet without signing a nondisclosure agreement that is very restrictive.
Also, you did not say how much you want to spend on this project. If you have around $1,000 (or a bit more) and maybe 80 hours of time, you could go to a much better dac, namely AK4499. If you have $500 you might have some trouble putting together a decent RPi dac system based on ES9038Q2M.
Last edited:
AK4499 evaluation boards go for $600. An Arduino can be used as an I2C controller, and and USB-to-I2S board can be fitted for about $80. Analog outputs are differential.
Although the eval board can run on +-15v, it starts sounding a lot better if the various subsystems are powered from dedicated regulators. However, that adds some more cost to the project. Could be done incrementally over time though.
Depending on where the balanced outputs will be going, it can help to offset the I/V opamp non-inverting input reference voltage divider from 2.5v to about 1.9v.
The most difficult thing found so far would be if one wanted to put the I/V opamps on their own dedicated +-15v power. Sounds better to do it for sure, but some tiny soldering is needed and some care with power supply sequencing must be observed.
One other puzzle might warrant some R&D: AKM uses Jung regulators to power the dac switched-resistor outputs (similar to what ESS calls AVCC). AKM uses a filter to further shape the Jung regulator outputs, and filter capacitor values affect LF harmonic distortion. Increasing cap values reduces the HD, but flattens out dynamics. There is a graph in the eval board manual showing HD vs cap values. There is a trick circuit I used to overcome the dynamics flattening problem, but it is not my trick and I cannot say what it is. Thus, others will have to find their own solutions or decide to live with some LF HD (as AKM apparently chose to do). Still sounds very good, no worries on that.
Although the eval board can run on +-15v, it starts sounding a lot better if the various subsystems are powered from dedicated regulators. However, that adds some more cost to the project. Could be done incrementally over time though.
Depending on where the balanced outputs will be going, it can help to offset the I/V opamp non-inverting input reference voltage divider from 2.5v to about 1.9v.
The most difficult thing found so far would be if one wanted to put the I/V opamps on their own dedicated +-15v power. Sounds better to do it for sure, but some tiny soldering is needed and some care with power supply sequencing must be observed.
One other puzzle might warrant some R&D: AKM uses Jung regulators to power the dac switched-resistor outputs (similar to what ESS calls AVCC). AKM uses a filter to further shape the Jung regulator outputs, and filter capacitor values affect LF harmonic distortion. Increasing cap values reduces the HD, but flattens out dynamics. There is a graph in the eval board manual showing HD vs cap values. There is a trick circuit I used to overcome the dynamics flattening problem, but it is not my trick and I cannot say what it is. Thus, others will have to find their own solutions or decide to live with some LF HD (as AKM apparently chose to do). Still sounds very good, no worries on that.
Hi edwardmorris,
Didn't mean to dissuade you by my comments. Of course it is possible to build a working AK4499 from info given in the data sheet.
Probably not the more interesting question, seems to me anyway. Others may differ.
IMHO, besides what is in the datasheet there are some other things AKM wants prospective dac designers to know about. Suggested layout ideas might be one. Use of different types of voltage regulators for different loads might be another. There could be more.
AK4499 can be used to make a wide range of dac products. At one end of the spectrum are low-cost (maybe with a catchy name like AK4499 Flagship DAC!) products that produce any kind of sound at all, which might work for some designers' purposes. At the other end of the spectrum are dacs that make the most of the dac chip to produce a truly great sounding dac design. Getting to that second kind of dac is harder and more involved. AKM appears to be giving hints about what they think matters for sound quality in the evaluation board. That is my take on it. Likely we will hear some other opinions too.
Didn't mean to dissuade you by my comments. Of course it is possible to build a working AK4499 from info given in the data sheet.
Probably not the more interesting question, seems to me anyway. Others may differ.
IMHO, besides what is in the datasheet there are some other things AKM wants prospective dac designers to know about. Suggested layout ideas might be one. Use of different types of voltage regulators for different loads might be another. There could be more.
AK4499 can be used to make a wide range of dac products. At one end of the spectrum are low-cost (maybe with a catchy name like AK4499 Flagship DAC!) products that produce any kind of sound at all, which might work for some designers' purposes. At the other end of the spectrum are dacs that make the most of the dac chip to produce a truly great sounding dac design. Getting to that second kind of dac is harder and more involved. AKM appears to be giving hints about what they think matters for sound quality in the evaluation board. That is my take on it. Likely we will hear some other opinions too.
Last edited:
Hey Markw4, I do appreciate you taking the time to reply to a newcomer like me, thank you! And of course I'd love to hear more from others as well.
I've been trying to educate myself more and more as I make time, googling, reading through whatever docs I find etc. and of course numerous posts here, just trying to sound as less stupid as I can when I post here haha.
My hesitation with teh 4499 eval board is the entry point for the price and me being a beginner. Now if just the daughter board was available to play with, that'd have been great. Based on a little more clarity that I now I *think* I have, lets refine the goal and see what can be done. Lets take whatever we can get from the DS and build something that meets the following (eventhough the chip is capable of a lot more).
Instead of targeting something that makes a sound (flagship!!) or something that squeezes out everything from the chip, how about I aim for a reasonably high fidelity stereo output (this is still possibly a little more to the left on the flagship-elite spectrum)? For starters, lay out the daughter board style design first, then fill out the rest following some of the suggestions here like individually powering subsystems etc. How permissiable is it to openly discuss these designs when it comes to the AK chips?
I've been trying to educate myself more and more as I make time, googling, reading through whatever docs I find etc. and of course numerous posts here, just trying to sound as less stupid as I can when I post here haha.
My hesitation with teh 4499 eval board is the entry point for the price and me being a beginner. Now if just the daughter board was available to play with, that'd have been great. Based on a little more clarity that I now I *think* I have, lets refine the goal and see what can be done. Lets take whatever we can get from the DS and build something that meets the following (eventhough the chip is capable of a lot more).
Instead of targeting something that makes a sound (flagship!!) or something that squeezes out everything from the chip, how about I aim for a reasonably high fidelity stereo output (this is still possibly a little more to the left on the flagship-elite spectrum)? For starters, lay out the daughter board style design first, then fill out the rest following some of the suggestions here like individually powering subsystems etc. How permissiable is it to openly discuss these designs when it comes to the AK chips?
Don't know if you have seen the eval board manual, which includes a schematic among other information. It comes on a CD with eval boards, or an AKM distributor can provide it. Don't know for sure why AKM hasn't made it available for download. Turns out they use Jung regulators for the 'Reference Voltage,' one Jung for each of the 4-channels. They also use four NJM7805 regulators, one for each channel, for the other dedicated analog power inputs. For the digital and RF analog stuff, they generally use LDO regulators. I did try using some LDOs instead of the 7805's and it messed up the sound (as LDOs often do for analog audio loads).
So, LDOs are recommended for all power input pins that are not dedicated to a particular dac channel. The dac power pins that are for each channel need to each have a dedicated regulator of a type that sounds good for analog circuits. The Reference Voltage is the more critical of the two for each channel. The I/V opamps need to have a separate set of +-15v regulators that is power sequenced as AKM warns in the section of the data sheet that discusses sequencing of power and signals. If no fail-safe sequencing, then zener diodes should be used to protect dac chip as described in the data sheet, but that may interfere with best sound quality. I could describe how I solved that particular problem.
See where this is going? It's a complicated chip to make sound good.
For the most part it needs 22/24Mhz clocks. A good solution that beats all others I have tried in terms of lowest jitter is JLSounds I2SoverUSB. That's a no brainer, IMHO.
Arduino or other MCU needed for serial control mode, otherwise no DSD capbility.
So, LDOs are recommended for all power input pins that are not dedicated to a particular dac channel. The dac power pins that are for each channel need to each have a dedicated regulator of a type that sounds good for analog circuits. The Reference Voltage is the more critical of the two for each channel. The I/V opamps need to have a separate set of +-15v regulators that is power sequenced as AKM warns in the section of the data sheet that discusses sequencing of power and signals. If no fail-safe sequencing, then zener diodes should be used to protect dac chip as described in the data sheet, but that may interfere with best sound quality. I could describe how I solved that particular problem.
See where this is going? It's a complicated chip to make sound good.
For the most part it needs 22/24Mhz clocks. A good solution that beats all others I have tried in terms of lowest jitter is JLSounds I2SoverUSB. That's a no brainer, IMHO.
Arduino or other MCU needed for serial control mode, otherwise no DSD capbility.
Last edited:
ES9038Q2M is a 2-channel Sabre chip aimed at mobile applications, but a very good dac can be made from one. Not in the same league as AK4499, but no Sabre chip is, IMHO.
Problem is that datasheets are only available under NDA, and we can't discuss certain things in public. Didn't stop some people people, of course, like Khadas Tone Board designers.
Hardware support I think we can talk about here and there is already a lot in the 'ES9038Q2M Board' thread.
Problem is that datasheets are only available under NDA, and we can't discuss certain things in public. Didn't stop some people people, of course, like Khadas Tone Board designers.
Hardware support I think we can talk about here and there is already a lot in the 'ES9038Q2M Board' thread.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.