The original board was the green version like this: https://www.ebay.com/itm/403818062410?hash=item5e056ea24a:g:KTkAAOSwYFpbQu65&amdata=enc:AQAHAAAA4OXzgw0oawkP+mpWo17Ciki3hILJp0CT/8Govx5gwtgc7FAI9oU0JKdubBNqIM376KN8YM7Y0A7YP1lIz0qp0sDErzxs0dAXrVUndejUWOq/ZkL679+aKE/IA2FguYIPRWrmUa/0qrDti3LTJmvHyYVI5GHn8u+4/l2WGTnUDXNmdNHJANEuqaAj9gWq4D4zGe1cvazQw5exPxPw+QZ/4eZ6gHpLr7RysqYkQr5N00HOW1/CTdln14ppxAkPKxOHJqJJ9LqB2c4fWwyjCCEpz6CghafC/y/DUbw8m1kYAN9X|tkp:BFBM5oeMwcBh
Some people also used the similar but usually a bit less expensive (and more cheaply made) blue one: https://www.ebay.com/itm/125646507390?hash=item1d411d8d7e:g:1T8AAOSwmZljjCuQ&amdata=enc:AQAHAAAAoPaN2+AM7UFiOKWkFQVxrXvPXKY1DsxAfQ8oWbCre3Y6LTmBS2kaHrJ+wLg2T48w+H/Law/OV3WH+eG2N8pXd8SuEdg5jpTxCz4YJX+l7f+K4PaiU5XIFYzudSvnc6/4K9/lN+25C7ObOSHfoXuL94l8VzV/ajxteEXq1/3gQTqkuHHKi9D0F6lTNoyj1WJbRhRI+bpbK1WugJhuSnvMBwU=|tkp:Bk9SR-SHjMHAYQ
Either way, I would not especially recommend these boards today unless you want to repeat the journey that went on in this thread. It was good for learning and experimenting, but its hard to get especially good SQ out of a two layer board, etc. At the time I started on it nobody had much experience with the dac chip and Chinese boards. People hoped there would be a fairly easy way to make them sound better, but that didn't turn out the case. There was a hard way though, which was to redo pretty much everything. Biggest improvements according to most people came from replacing the output stage with a more proper one, and by improvements to the Vref power supply. Another thing that often helped was to do as ESS recommended and set DPLL_BANDWIDTH to the lowest stable value using the I2C control registers.
Some people also used the similar but usually a bit less expensive (and more cheaply made) blue one: https://www.ebay.com/itm/125646507390?hash=item1d411d8d7e:g:1T8AAOSwmZljjCuQ&amdata=enc:AQAHAAAAoPaN2+AM7UFiOKWkFQVxrXvPXKY1DsxAfQ8oWbCre3Y6LTmBS2kaHrJ+wLg2T48w+H/Law/OV3WH+eG2N8pXd8SuEdg5jpTxCz4YJX+l7f+K4PaiU5XIFYzudSvnc6/4K9/lN+25C7ObOSHfoXuL94l8VzV/ajxteEXq1/3gQTqkuHHKi9D0F6lTNoyj1WJbRhRI+bpbK1WugJhuSnvMBwU=|tkp:Bk9SR-SHjMHAYQ
Either way, I would not especially recommend these boards today unless you want to repeat the journey that went on in this thread. It was good for learning and experimenting, but its hard to get especially good SQ out of a two layer board, etc. At the time I started on it nobody had much experience with the dac chip and Chinese boards. People hoped there would be a fairly easy way to make them sound better, but that didn't turn out the case. There was a hard way though, which was to redo pretty much everything. Biggest improvements according to most people came from replacing the output stage with a more proper one, and by improvements to the Vref power supply. Another thing that often helped was to do as ESS recommended and set DPLL_BANDWIDTH to the lowest stable value using the I2C control registers.
This reply was helpful!
Yes, I am putting off investing more in "older" ESS or AKM dacs (and low-$ Chinese kit boards based on them) because of the NEW line of DACs both these and other DAC manuf's have recently released.
That said, I am interested in pursuing the analog end of these kits. I continue to be impressed with my other experiments with discrete output stages (such as in the various i/v threads). Yes, i/v does not apply to voltage-out dacs. But a discrete output might be a option for single-bit DACs like delta-sigma. Or perhaps op-amp rolling or tweaking was discussed in this thread? Yes?
The opamp of choice for these modern sigma-delta dacs is generally OPA1611/1612. If that doesn't sound best, you are probably doing something wrong. Now, it may be possible to design a discrete opamp that sounds better, but its not a trivial problem. There is RF mixed in with the audio coming out of a DAC. Making opamps RF tolerant (so as to avoid intermodulation of RF with audio in semiconductor junctions) could be an issue. Please see attached.
Right ... CAREFUL decoupling -- not going overboard with electrolytic uF -- and even snubbering of opamps and their regs and PSUs. I posted pictures of what I did with a Meier headphone amp. Gets a bit messy ... wired like a xmas tree. But sonics improve quite a bit.
See:
https://www.diyaudio.com/community/threads/improving-a-meier-audio-headphone-amp.394022/
See:
https://www.diyaudio.com/community/threads/improving-a-meier-audio-headphone-amp.394022/
Well, some of that stuff can help or not, depending. It can help a lot to get a wide bandwidth scope and an active low capacitance probe (even a used from on ebay could suffice, although you may have to make a ground spring for one of those). Its that you should measure to see if there is evidence of a problem before hanging components around without really knowing. Sometimes adding components on a hunch can make things worse. Also, I would agree with what someone in the other thread said about cleaning up excessive flux residue. It can expose problems that might otherwise have been visually evident. At least try. MG chemicals makes some high purity isopropanol which along with a flux brush can do a lot to help. After that its possible to use full strength 'Simple Green' and a fresh flux brush, followed by a distilled/deionized water rinse. It can be be worthwhile.
Hi, after a while i decided to make a dual mono DAC board es9038q2m. I use Rcore transformer and LT1674 to main 5v, then LT1674 on mainboard to main 3.3v for oscillator and other 3.3v components, ltc6655 and opa1612 for AVCC and VREF, tps65131 +-15v and low voltage through LM317/337. Is this layout ok, even though it's not complete
You would have to provide Gerbers or equivalent if you want comments on layout. Routing and use of planes/fills and multiple layers is big part of overall layout.
A schematic and parts values would help too if you would like feedback on those things.
Also, if you use Amanero, they usually sound better if run of clean +5v power, instead of USB power.
In addition, I might have some concern about putting what looks like a Bluetooth receiver under the Amanero.
Another possible concern is that I only see one clock. ES9038Q2M can be operated that way, but it may not be the best way, depending. Don't know the clock frequency either. 100MHz clocks offer the most sample rates but higher clock frequencies may increase measured distortion, as well as increase jitter/phase-noise.
IOW, all sorts of details can matter.
A schematic and parts values would help too if you would like feedback on those things.
Also, if you use Amanero, they usually sound better if run of clean +5v power, instead of USB power.
In addition, I might have some concern about putting what looks like a Bluetooth receiver under the Amanero.
Another possible concern is that I only see one clock. ES9038Q2M can be operated that way, but it may not be the best way, depending. Don't know the clock frequency either. 100MHz clocks offer the most sample rates but higher clock frequencies may increase measured distortion, as well as increase jitter/phase-noise.
IOW, all sorts of details can matter.
Last edited:
What board layout program are you using?
For myself, I always use the ratsnest view to start optimizing component location and orientation.
If using KiCad or maybe some other programs, there are or may be useful ways available for organizing signal types using color-coded groups, individualized color-coding, etc. It can be worth the time to utilize such features to help optimize layout. You have to decide which signals are most critical and give them routing priority. Then the next lower level of priority, and so on. Yet while doing that type of thing always be aware of overall optimization too.
For myself, I always use the ratsnest view to start optimizing component location and orientation.
If using KiCad or maybe some other programs, there are or may be useful ways available for organizing signal types using color-coded groups, individualized color-coding, etc. It can be worth the time to utilize such features to help optimize layout. You have to decide which signals are most critical and give them routing priority. Then the next lower level of priority, and so on. Yet while doing that type of thing always be aware of overall optimization too.
Last edited:
You can do better than you are, better clocking, etc. However it adds complexity and cost. The upside is you have the potential of significantly better performance.
A lot of the sound and overall performance of a dac comes from the quality of the circuitry surrounding the dac chip, often that surrounding circuitry is more important to final result than choice of dac chip. To put it another way, the dac chip kind of sets a limit of the best you can do if all the circuitry is the best it can be. That being the case it may be a better tradeoff to use some other dac chip and focus more on the support circuitry.
In some well known professional dacs the dac chip is actually AK4493. It is somewhere in the middle of AKMs offerings, yet can perform very well for a lot of purposes. They tend to sound good, for one thing.
A lot of the sound and overall performance of a dac comes from the quality of the circuitry surrounding the dac chip, often that surrounding circuitry is more important to final result than choice of dac chip. To put it another way, the dac chip kind of sets a limit of the best you can do if all the circuitry is the best it can be. That being the case it may be a better tradeoff to use some other dac chip and focus more on the support circuitry.
In some well known professional dacs the dac chip is actually AK4493. It is somewhere in the middle of AKMs offerings, yet can perform very well for a lot of purposes. They tend to sound good, for one thing.
Hi,
After dealing with es9038q2m green board and 9038pro chinese board, I've decided to develop my own custom 9028pro board, the performance/price ratio is pretty good (with chinese sourcing)
Current specs:
Only differential output
Optical input
spdif input
Usb input with customized xu208 board ( i2s and i2c management)
8 thp210 opamps
2 lt3045 for avcc
Powered by one 5v power supply ( board has 2 DC DC converters (5v to 9v) for opamp supplies stabilized by 2 lt3042.)
3 boards stacked ( xu208 , custom power board (including input management)+ es9028pro custom board
I'm very happy with it , it's the first board which works better with thd compensation turned off ... I'm now limited by my custom ADC measurement board ( cs5381)
I will share some pictures soon...
cheers
Occip
After dealing with es9038q2m green board and 9038pro chinese board, I've decided to develop my own custom 9028pro board, the performance/price ratio is pretty good (with chinese sourcing)
Current specs:
Only differential output
Optical input
spdif input
Usb input with customized xu208 board ( i2s and i2c management)
8 thp210 opamps
2 lt3045 for avcc
Powered by one 5v power supply ( board has 2 DC DC converters (5v to 9v) for opamp supplies stabilized by 2 lt3042.)
3 boards stacked ( xu208 , custom power board (including input management)+ es9028pro custom board
I'm very happy with it , it's the first board which works better with thd compensation turned off ... I'm now limited by my custom ADC measurement board ( cs5381)
I will share some pictures soon...
cheers
Occip
Hi,
The output is pretty clean.
Chinese es9028q2m board lacks of good output stage and bad avcc supply. if you want to tweak it , you have to externalize the avcc supply and the output stage with long wires , not so good for SNR and thd. A fully custom board with a tqfp 64 package is easy to solder and to debug.
The es9028pro is also less prone to ess hump.
I've found also that the thp210 opamp is easy to manage compared to opa1632 (which consume a lot of current )
For me es9028pro + lt3045 + thp210 is the magic combo
. Please forgive the quality of soldering after many debugging
The output is pretty clean.
Chinese es9028q2m board lacks of good output stage and bad avcc supply. if you want to tweak it , you have to externalize the avcc supply and the output stage with long wires , not so good for SNR and thd. A fully custom board with a tqfp 64 package is easy to solder and to debug.
The es9028pro is also less prone to ess hump.
I've found also that the thp210 opamp is easy to manage compared to opa1632 (which consume a lot of current )
For me es9028pro + lt3045 + thp210 is the magic combo
. Please forgive the quality of soldering after many debugging Hi, there. I just bought a device, thinking about mod. Lack of resolution and transparency. I think the shunt of electrolytes (wima 0.1uF) or just replace them on Nichicon. I have a linear PSU 12V. What else I can do with it? Please give me an advice!
https://www.dastereo.ru/uploads/def...d14b8cd9ece3c2809ea6f85256f31e411ca30337.jpeg
https://www.dastereo.ru/uploads/def...d14b8cd9ece3c2809ea6f85256f31e411ca30337.jpeg
Hi,
The device you have needs two main things to start sounding quite a bit better. (1) a proper 3-opamp output stage, and (2) a proper AVCC power supply/regulator. Its a lot of work, usually more work than its worth unless maybe your aim is start learning about dac design. Another thing that would probably help would be to run the output stage opamps on a +- dual rail linear power supply. The single supply option is not as good. If you had asked before getting it we probably would have warned you about it then. Sorry if that's not what you wanted to hear.
Also, schematic of what you probably have attached, if that helps at all.
The device you have needs two main things to start sounding quite a bit better. (1) a proper 3-opamp output stage, and (2) a proper AVCC power supply/regulator. Its a lot of work, usually more work than its worth unless maybe your aim is start learning about dac design. Another thing that would probably help would be to run the output stage opamps on a +- dual rail linear power supply. The single supply option is not as good. If you had asked before getting it we probably would have warned you about it then. Sorry if that's not what you wanted to hear.
Also, schematic of what you probably have attached, if that helps at all.
Attachments
Hi @Markw4. The default opamp is muses 8920 from ali (non -original). I have OPA2134 and a few one-channel opamps. I don't understand why this Dac doesn't work with my Amanero? I use the standart connection pins (BCLK, FSCLK, DATA and GRD). Maybe I should use other scheme?
I also talk with my friend about realization AVCC power supply/regulator.
I also talk with my friend about realization AVCC power supply/regulator.
We had an audio matrix with 8x Mic- and 8x Line-Inputs and 16x Line-Outputs and every channel had a OPA1632 and Low-Noise LDOs from LT for every supply voltage. Only low value resistors were used and alot of other techniques were applied to get a better signal quality. But the system got pretty hot and when you hooked it up to a PA system you could hear the noise floor.
These shiny opamps often make more trouble than someone might think.
An interesting alternative is the LME49724.
BTW, TI seems to have a successor to the OPA1632 in the pipeline -> OPA1633
The problem is that with only one opamp you can only operate the dac in voltage output mode. The output stage needs to operate the dac in current output mode to produce better sound. Old ESS application note attached with applicable info.
The opamp of choice for ESS and for AKM has for quite some time been OPA1612. If you can't get it sounding good with that opamp, then there are other problems that need to be fixed.
Regarding Amanero, the dac should work but you have to select to use I2S input using the jumpers and or dip switches and or mode button on your particular dac board. It depends on how the MCU is programmed to allow for input selection. Usually the jumpers and or dip switches also allow selection of the dac filter.
The opamp of choice for ESS and for AKM has for quite some time been OPA1612. If you can't get it sounding good with that opamp, then there are other problems that need to be fixed.
Regarding Amanero, the dac should work but you have to select to use I2S input using the jumpers and or dip switches and or mode button on your particular dac board. It depends on how the MCU is programmed to allow for input selection. Usually the jumpers and or dip switches also allow selection of the dac filter.
Last edited: