Hello,
I try to improve sound stage on dual ES9038q2m based player.
ms gamebar services
I was changed capacitors in audio rail near op amps for muRata C0G.
LPF OP275 replaced to OPA1656.
Changed LDO for TI5907 plus recommended ~2,2uF+100nF.
Resistors in audio rail for Panasonic thin-film 0.1% instead of thic-film probably >2%
Changing capacitors for X7R and additional 100nF G0C in analog supply part.
In DAC section, additional 4x220uF tantal-polimer from AVX and replaced 8x 10uF X7R +100nF G0C.
Still can't achieve sound stage as wide and deep as similar device based on Dual AK4497 with single ended output instead of balanced on ES9038.
Where I can dig more? These 2 devices are similar in design.
I try to improve sound stage on dual ES9038q2m based player.
ms gamebar services
I was changed capacitors in audio rail near op amps for muRata C0G.
LPF OP275 replaced to OPA1656.
Changed LDO for TI5907 plus recommended ~2,2uF+100nF.
Resistors in audio rail for Panasonic thin-film 0.1% instead of thic-film probably >2%
Changing capacitors for X7R and additional 100nF G0C in analog supply part.
In DAC section, additional 4x220uF tantal-polimer from AVX and replaced 8x 10uF X7R +100nF G0C.
Still can't achieve sound stage as wide and deep as similar device based on Dual AK4497 with single ended output instead of balanced on ES9038.
Where I can dig more? These 2 devices are similar in design.
Last edited:
Why do some of the components look burned or scratched on top?
Regarding getting better sound, trying to fix someone else's dac design by replacing a few parts doesn't usually help much. Sometimes using more expensive parts can make an audio device sound worse.
Its more like first you have to understand how a dac should be designed. Then you have to design one yourself, or else find one that easily can be modified into a better design. The dac in the pic does not look like an easy one to make sound better. Too many high pin count flat pack ICs, for one problem. How do you know what all those things are doing to the sound? Do you have schematics? Have access to all the source code for those programmable chips? If not, how you can you determine if the dac SQ is likely to be fixable,?
There was a thermal pad who hold accumulator in place as well.
Don't have any scheme, I can try to do it first after MCU of course, more like 2x LDO powering 2 DACS. 2 LPF and Amps have 9V rail with 19 x 100uF MLCC ones. I have values for resistors and capsnearby
Don't have any source codes for chips, just try to understand what can I do more.
I try to do my best and I have some improvements after mods vs stock ones.
I'm just wondering what I can do more?
That picture isn’t very good.
What I have found to make an improvement in a media player I have is, on the digital portion, to carefully compensate the output of each regulator, starting with the regulator data sheets for esr recommendations. Also the clock decoupling was an area where having the just right ceramic cap made a difference. I found some parts in an old tv dvr that worked well, I wish I knew what they were.
The analog part will be something you’ll need to maybe put in a socket for the opamps and some way to easily swap caps around such as not to damage the board traces from too much meddling.
Power supplies seemed to also benefit from some simple ferrites on the hot wires. Sometimes devices will slip in a 0 ohm resistor that can be substituted for these.
What I have found to make an improvement in a media player I have is, on the digital portion, to carefully compensate the output of each regulator, starting with the regulator data sheets for esr recommendations. Also the clock decoupling was an area where having the just right ceramic cap made a difference. I found some parts in an old tv dvr that worked well, I wish I knew what they were.
The analog part will be something you’ll need to maybe put in a socket for the opamps and some way to easily swap caps around such as not to damage the board traces from too much meddling.
Power supplies seemed to also benefit from some simple ferrites on the hot wires. Sometimes devices will slip in a 0 ohm resistor that can be substituted for these.
Thank You for answer phase.
Better picture: PXL-20210806-162746786-01 — ImgBB
The problem is VSON10 instead of SO8/DIP8 i in theory I have space for 2 or 4 DIP8 sockets in AMP area. But need very precise wireing and its not as safe its portable. I think better opams from 1622 will be Burson V5i or Muses 02/03 who can fit, but it's quite expensive.
Ferrites on main accumulator wires to clean spikes in power line?
Better picture: PXL-20210806-162746786-01 — ImgBB
The problem is VSON10 instead of SO8/DIP8 i in theory I have space for 2 or 4 DIP8 sockets in AMP area. But need very precise wireing and its not as safe its portable. I think better opams from 1622 will be Burson V5i or Muses 02/03 who can fit, but it's quite expensive.
Ferrites on main accumulator wires to clean spikes in power line?
Maybe a silly question, but have someone tried to bypass the opamp and feed DAC chip output (DACR, DACL) directly into output buffer?
I spent some time reading this thread but didn't found anything like that.
I experimented actually with a darlington emitter follower output buffer and the result is amazing IMO, it would be interesting to know others similar experiences.
Before this I played with a couple of op amps (as preamps), opa2134 being most expensive and best sounding, but the results were nowhere near with that of direct output coupling.
Sure, this not an universal solution, but for low impedance and high gain headphones this "geld" may be the best improvement to this board.
I spent some time reading this thread but didn't found anything like that.
I experimented actually with a darlington emitter follower output buffer and the result is amazing IMO, it would be interesting to know others similar experiences.
Before this I played with a couple of op amps (as preamps), opa2134 being most expensive and best sounding, but the results were nowhere near with that of direct output coupling.
Sure, this not an universal solution, but for low impedance and high gain headphones this "geld" may be the best improvement to this board.
Didn't found THD figures of current mode vs voltage mode anywhere in the ES9038Q2M datasheet, I will appreciate a link to such info.
Anyway, considering the major improvement in SQ of the DAC->follower setup, heard without any effort in a A-B test, over the headphone output of "highly acclaimed" low distortion DX3 Pro (v1), I suppose the statement "distortion is *much* higher" refers to figures obtained after opamp, not for the DAC itself.
As for me, I am completely contented with a possible minor measurable distortion of the voltage mode no-opamp setup over the "low distortion" but altered sound produced by the opamp no matter of mode.
Anyway, considering the major improvement in SQ of the DAC->follower setup, heard without any effort in a A-B test, over the headphone output of "highly acclaimed" low distortion DX3 Pro (v1), I suppose the statement "distortion is *much* higher" refers to figures obtained after opamp, not for the DAC itself.
As for me, I am completely contented with a possible minor measurable distortion of the voltage mode no-opamp setup over the "low distortion" but altered sound produced by the opamp no matter of mode.
IME a proper opamp design with this type of dac chip can give better SQ than other approaches. However, a proper opamp design is not trivial to implement well. So we see some people using transformers or other approaches.
IMHO differential summing is often where SQ tends to start to suffering. Its not so much a problem with the I/V stage if layout and component selection are sufficiently optimized. Power supply and decoupling for opamp rails also matters. So maybe think about how to improve or maybe avoid the usual differential summing approach.
Beyond that, AVCC supply design also matters a lot in terms of final SQ.
Those two things, output stage design and AVCC design, make the biggest difference in SQ for these dac chips. There is more too, but SQ improvement is smaller for those other things.
IMHO differential summing is often where SQ tends to start to suffering. Its not so much a problem with the I/V stage if layout and component selection are sufficiently optimized. Power supply and decoupling for opamp rails also matters. So maybe think about how to improve or maybe avoid the usual differential summing approach.
Beyond that, AVCC supply design also matters a lot in terms of final SQ.
Those two things, output stage design and AVCC design, make the biggest difference in SQ for these dac chips. There is more too, but SQ improvement is smaller for those other things.
Last edited:
Since things are so slow in the forum these days, thought I might talk about AVCC voltage regulation a bit.
I have started trying discrete series regulators for AVCC (Vref) type use. IME with some tuning and optimization of a fairly simple circuit, its possible to get better SQ that I have been able to get with IC regulators or even with opamp buffers. A basic regulator topology is introduced reasonably well at: Transistor Series Voltage Regulator | Error Amplifier Working
Referring to figure 17-3 of the linked article, a few 'improvements' might be made to the basic circuit. For example, R1 could be replaced with a simple current source (such as constant current diode); D1 could be replaced with a red LED; and Q1 could be made into a Darlington (integrated or discrete). Also, the series regulator could be preceded by a capacitance multiplier to help clean up any input noise. The example changes are enough to make a nice sounding regulator that is not excessively complex. Avoiding unnecessary complexity makes the process of circuit adjustment simpler.
One consideration about changing Q1 to a Darlington is that regulator loop gain is increased. That can improve regulation, but it can also make assuring stability more of an issue. It is something that calls for availability of a scope with decent bandwidth (100Mhz seems minimal for dac work). It is also important to know how to probe and otherwise make good use of a scope. For some people there may be some study involved to help get up to speed. More discussion and info on regulator testing and adjustment is possible if anyone decides to give this a try.
One of the things I found experimenting with AVCC regulation is that the regulator output caps which help provide frequency stability can also affect a dac's particular sound. To explore the effects of regulator caps on dac sound I mounted some caps on female pin headers, and also mounted male pin headers on the PCB where the AVCC regulator output caps would go. That made it easy to do quick comparisons. Again, more could be said about caps and comparisons if anyone wants to dive in and explore building a discrete AVCC regulator.
Another possible thing to maybe think about is the fundamental difference between a conventional voltage regulator and something like an opamp buffer. Voltage regulators normally use output caps both for frequency compensation and to help supply current for load transients. On the other hand, amplifiers (such as opamp buffers) are often frequency compensated at internal nodes. When amps are used as power supply buffers, an output cap may or may not be used for stability. A possible advantage of compensating at internal nodes is that smaller value caps may be needed. That may mean that use of better quality caps becomes more feasible.
Okay then, those are some thoughts during this sorta lull period in the dac subforum. The critics are free to have their fun.
I have started trying discrete series regulators for AVCC (Vref) type use. IME with some tuning and optimization of a fairly simple circuit, its possible to get better SQ that I have been able to get with IC regulators or even with opamp buffers. A basic regulator topology is introduced reasonably well at: Transistor Series Voltage Regulator | Error Amplifier Working
Referring to figure 17-3 of the linked article, a few 'improvements' might be made to the basic circuit. For example, R1 could be replaced with a simple current source (such as constant current diode); D1 could be replaced with a red LED; and Q1 could be made into a Darlington (integrated or discrete). Also, the series regulator could be preceded by a capacitance multiplier to help clean up any input noise. The example changes are enough to make a nice sounding regulator that is not excessively complex. Avoiding unnecessary complexity makes the process of circuit adjustment simpler.
One consideration about changing Q1 to a Darlington is that regulator loop gain is increased. That can improve regulation, but it can also make assuring stability more of an issue. It is something that calls for availability of a scope with decent bandwidth (100Mhz seems minimal for dac work). It is also important to know how to probe and otherwise make good use of a scope. For some people there may be some study involved to help get up to speed. More discussion and info on regulator testing and adjustment is possible if anyone decides to give this a try.
One of the things I found experimenting with AVCC regulation is that the regulator output caps which help provide frequency stability can also affect a dac's particular sound. To explore the effects of regulator caps on dac sound I mounted some caps on female pin headers, and also mounted male pin headers on the PCB where the AVCC regulator output caps would go. That made it easy to do quick comparisons. Again, more could be said about caps and comparisons if anyone wants to dive in and explore building a discrete AVCC regulator.
Another possible thing to maybe think about is the fundamental difference between a conventional voltage regulator and something like an opamp buffer. Voltage regulators normally use output caps both for frequency compensation and to help supply current for load transients. On the other hand, amplifiers (such as opamp buffers) are often frequency compensated at internal nodes. When amps are used as power supply buffers, an output cap may or may not be used for stability. A possible advantage of compensating at internal nodes is that smaller value caps may be needed. That may mean that use of better quality caps becomes more feasible.
Okay then, those are some thoughts during this sorta lull period in the dac subforum. The critics are free to have their fun.
Wow, this must be a new record. 5x "could", 5x "may", 1x "maybe" and 2x "might" in one post.
So here are my questions:
So here are my questions:
- Are actually building something or are these just thought experiments?
- Is the regulator usable in it's current state?
- Will you post schematics with component values/recommendations?
- Did you measure output impedance vs. frequency, response to current step, noise spectrum?