Of course. I am just using the I/V stage to turn my voltmeter (scope) into an ampmeter.
EDIT: I thought I explained how that calculation worked quite clearly. Do I need to explain better, or are there any questions?
Last edited:
I just fired up my 1.07 board to find out what the encoder switch do.
It's everscroll volume and quick pushed in chooses the inputs(menu button on screen pcb does the same). Slow press on the encoder or the menu button chooses the filter type.
My board didn't come with the remote so I've yet to test it. It has an IR receiver chip soldered on so I will be trying out my apple clone remote when that arrives.
I've have to say I like the little screen, it makes the dac stand out from my other Arcam rdac which only have leds.
It's everscroll volume and quick pushed in chooses the inputs(menu button on screen pcb does the same). Slow press on the encoder or the menu button chooses the filter type.
My board didn't come with the remote so I've yet to test it. It has an IR receiver chip soldered on so I will be trying out my apple clone remote when that arrives.
I've have to say I like the little screen, it makes the dac stand out from my other Arcam rdac which only have leds.
Last edited:
Found the complete dac Ebay listing! Good starting point for someone that doesn't want to spend to much time on building, but like doing mods: ES9038 ES9038Q2M DAC Decoder board Support IIS DSD DOP 384KHz with Amanero USB | eBay
Auction text says it works with Apple remote even if it doesn't come with one, maybe all 1.07 board supports the remote??
Auction text says it works with Apple remote even if it doesn't come with one, maybe all 1.07 board supports the remote??
That complete DAC ebay link raises some concerns.
They are encouraging people to use single-ended 12v or 9v power supplies.
I hope the (not original Italian) Amanero board supports SPDIF out or the ability to upsample will probably be lost.
They keep advertising -120dB distortion in those ads which is truly dishonest. -75dB is more like it for that voltage output stage.
Despite what it says on the front of the case, it is not a Power Amplifier.
As usual, eye-candy out competes sound quality. Too bad.
They are encouraging people to use single-ended 12v or 9v power supplies.
I hope the (not original Italian) Amanero board supports SPDIF out or the ability to upsample will probably be lost.
They keep advertising -120dB distortion in those ads which is truly dishonest. -75dB is more like it for that voltage output stage.
Despite what it says on the front of the case, it is not a Power Amplifier.
As usual, eye-candy out competes sound quality. Too bad.
Last edited:
So this is for the whole thing as pictured? In a box with display? The text suggests it's for the board?
Edit: this one is an auction... ES9038 ES9038Q2M DAC Decoder board Support IIS DSD DOP 384KHz with Amanero USB | eBay
Its probably for everything as pictured. The boards sell for less. You can always send a message to the seller asking for clarification.
Its probably for everything as pictured. The boards sell for less. You can always send a message to the seller asking for clarification.
Last edited:
Audiophonics just released the raspi DAC board with 9038q2m. They have i/v conversion and low jitter clock as they say and it could work stand alone.
They say the 1.2 V for Vref is very important and has its own reg...(no word about AVCC). I wonder why the newer versions of the here discussed board skip this reg. though?
All new products on sale - Audiophonics
They say the 1.2 V for Vref is very important and has its own reg...(no word about AVCC). I wonder why the newer versions of the here discussed board skip this reg. though?
All new products on sale - Audiophonics
to my understanding 1.2v are for DVDD digital core. a little unclear what Vref they mean.
some explanation comes from here:
Buffalo II
some explanation comes from here:
Buffalo II
The Buffalo II post by Russ White is correct. Also true that there is an internal DVDD regulator at least for mobile. Don't know why some Chinese boards have an external regulator there and other boards don't have it. Bypass capacitance is needed either way.
Regarding AVCC, by now I think more and more people understand that ESS has very good reasons for recommending very high quality power for AVCC. What ESS has not said but the best DAC makers know is that (1) built-in reconstruction filters are a weakness, and (2) there is some advantage to high-quality upsampling, at least for the lowest sample rates of 44.1 and 48.
In regard to reconstruction filters, the reason there are so many filter options is because they are all audible. Ideally they should not be since the idea is to hear the music not the DAC. Not just for ESS DACs either. Same for AK would be my understanding.
Fixing AVCC is relatively low-cost and probably shouldn't be something for DAC makers to brag about. Fixing other things gets progressively more expensive with reconstruction filtering probably being the most expensive.
Not only does external reconstruction filtering require an external DSP chip, it is hard to find one fast enough to allow the DAC to work at the highest sample rates. DAC clock speeds may have to be lowered to make external DSP practical which means there is an engineering trade-off decision to be made for one set of features vs. another, with each at different cost points. Cost may also be affected in that some options would need a more expensive PRO type DAC chip.
The best DACs built around ESS or AKM chips that I know of do more than chip manufacturers recommend. Benchmark and Crane Song may currently be the top two brands and examples of technology aimed at professional users in recording, mixing, mastering, sound for film production, etc. Minimum price for a finished box in that class is maybe right around $2k for 2-channels.
Regarding AVCC, by now I think more and more people understand that ESS has very good reasons for recommending very high quality power for AVCC. What ESS has not said but the best DAC makers know is that (1) built-in reconstruction filters are a weakness, and (2) there is some advantage to high-quality upsampling, at least for the lowest sample rates of 44.1 and 48.
In regard to reconstruction filters, the reason there are so many filter options is because they are all audible. Ideally they should not be since the idea is to hear the music not the DAC. Not just for ESS DACs either. Same for AK would be my understanding.
Fixing AVCC is relatively low-cost and probably shouldn't be something for DAC makers to brag about. Fixing other things gets progressively more expensive with reconstruction filtering probably being the most expensive.
Not only does external reconstruction filtering require an external DSP chip, it is hard to find one fast enough to allow the DAC to work at the highest sample rates. DAC clock speeds may have to be lowered to make external DSP practical which means there is an engineering trade-off decision to be made for one set of features vs. another, with each at different cost points. Cost may also be affected in that some options would need a more expensive PRO type DAC chip.
The best DACs built around ESS or AKM chips that I know of do more than chip manufacturers recommend. Benchmark and Crane Song may currently be the top two brands and examples of technology aimed at professional users in recording, mixing, mastering, sound for film production, etc. Minimum price for a finished box in that class is maybe right around $2k for 2-channels.
this may depend on MCKL frequency. at higher rates internal regulator may not supply enough power.
Understandably, people are probably trying to make sense out of why some boards have a regulator and some don't. I am not at liberty to tell you what ESS says about it, unfortunately. But I can tell you I don't know why some Chinese boards have regulators and some don't. Hopefully, they are not using reject chips or anything like that. Could be they are using some old information from previous ESS chips. At this point we just don't know and probably nothing to be gained by speculation.
Last edited:
Markw4
I've been looking at your shots of the DAC3 (finally figured out the magnification) and their crystal runs at 50Mhz! Any idea why this might be?
The other thing I am noticing is that a lot of their bulk decoupling is chosen at 330uF 6V!. They also have some large 1000uF 6V caps close to their DAC. I suppose this might be for AVCC.
The other thing I find interesting is that their clock is located relatively far away from the DAC chips compared to other implementations that favor the crystal to be close. Instead they have a couple of ICs that are located closer to the DAC which is counter to what I've been noticing.
Then they locate their DAC to the far edge of the board. Now this forces longer traces towards the DAC chip because they lose proximity to the DAC from one side. This contrasts with what Twisted Pear had done where they attempt to cluster as much as physically possible next to the DAC chip.
I've been looking at your shots of the DAC3 (finally figured out the magnification) and their crystal runs at 50Mhz! Any idea why this might be?
The other thing I am noticing is that a lot of their bulk decoupling is chosen at 330uF 6V!. They also have some large 1000uF 6V caps close to their DAC. I suppose this might be for AVCC.
The other thing I find interesting is that their clock is located relatively far away from the DAC chips compared to other implementations that favor the crystal to be close. Instead they have a couple of ICs that are located closer to the DAC which is counter to what I've been noticing.
Then they locate their DAC to the far edge of the board. Now this forces longer traces towards the DAC chip because they lose proximity to the DAC from one side. This contrasts with what Twisted Pear had done where they attempt to cluster as much as physically possible next to the DAC chip.
Last edited:
There are at least 5 clocks in the box. The one closest to the DAC chip IIRC is 30Mhz.
6v decoupling caps are probably for 3.3v or maybe lower voltage rails. Most of the DAC runs at 3.3v. I don't know exactly what they do for AVCC, but 1000uf caps usually some application limitations. Caps for HF switching supplies might have lowest ESR and inductance, at least for electrolytics. I know Benchmark has a lot of expensive test equipment and they measure a lot. If one has the equipment and time to perform a series of investigations then one can find out first hand what works and what doesn't.
They are also using a multi-layer circuit board vs. the Chinese DAC two layer. With enough layers there can be ground and power planes and PCB structures that approximate something like coaxial cable and which are very mechanically stable. Given some of those things they have some really good options available that we don't.
The DAC is on the side of the board, I would guess, to keep it as far away as possible from unwanted signals, noise, interference, etc. It has an analog half and a digital half. ESS recommends a digital ground plane under part of it and an analog ground plane under other areas. The Chinese DACs have one ground plane that is broken in places to allow some routing of signals that can't cross each other if they are both on the top layer.
What it kind of looks like T.P. is doing is trying to make many different options available to people who want their own custom version of a DAC. That is very different from what Benchmark is trying to do. Also, it seems kind of unlikely that T.P. has the same engineering resources and test equipment budget Benchmark has. No doubt, the latter operates in a much higher volume market than the former which means Benchmark can probably absorb more design and development costs.
6v decoupling caps are probably for 3.3v or maybe lower voltage rails. Most of the DAC runs at 3.3v. I don't know exactly what they do for AVCC, but 1000uf caps usually some application limitations. Caps for HF switching supplies might have lowest ESR and inductance, at least for electrolytics. I know Benchmark has a lot of expensive test equipment and they measure a lot. If one has the equipment and time to perform a series of investigations then one can find out first hand what works and what doesn't.
They are also using a multi-layer circuit board vs. the Chinese DAC two layer. With enough layers there can be ground and power planes and PCB structures that approximate something like coaxial cable and which are very mechanically stable. Given some of those things they have some really good options available that we don't.
The DAC is on the side of the board, I would guess, to keep it as far away as possible from unwanted signals, noise, interference, etc. It has an analog half and a digital half. ESS recommends a digital ground plane under part of it and an analog ground plane under other areas. The Chinese DACs have one ground plane that is broken in places to allow some routing of signals that can't cross each other if they are both on the top layer.
What it kind of looks like T.P. is doing is trying to make many different options available to people who want their own custom version of a DAC. That is very different from what Benchmark is trying to do. Also, it seems kind of unlikely that T.P. has the same engineering resources and test equipment budget Benchmark has. No doubt, the latter operates in a much higher volume market than the former which means Benchmark can probably absorb more design and development costs.
Last edited:
Regarding 100MHz verses lower frequency clocks for ESS DAC chips, it is an engineering trade-off question. If one would like to be able to play back 384kHz PCM and whatever high speed DSD then a fast DAC clock is needed. If one is willing to forego being able to play back some of those formats then there other things one can do to improve SQ for more the more common sample rates. Like maybe fix the reconstruction filter problem with an outboard DSP, fix the oversampling filter overs problem at the same time. Takes a very fast DSP though. Probably not easy to find one at least at a reasonable cost that could work at the highest sample rates.
While we are thinking about it we may as well ask why people want super high sample rates anyway. Presumably because they think it will provide better SQ. Turns outthere is a limit to frequencies and timing differences people can hear. They may be able to hear shorter duration timing errors in some case than correspond with the highest frequencies they can hear. If so, it would just prove there are some things about aural perception that are nonlinear between the time and frequency domains. Wouldn't surprise me, but we don't have all the final answers yet, IMHO. Maybe someday.
Anyway, if we can get the SQ we want by fixing known problems rather than by hoping that higher sample rate formats will do the trick, we very well may end up with better results. That's the direction it looks like Benchmark decided to go.
While we are thinking about it we may as well ask why people want super high sample rates anyway. Presumably because they think it will provide better SQ. Turns outthere is a limit to frequencies and timing differences people can hear. They may be able to hear shorter duration timing errors in some case than correspond with the highest frequencies they can hear. If so, it would just prove there are some things about aural perception that are nonlinear between the time and frequency domains. Wouldn't surprise me, but we don't have all the final answers yet, IMHO. Maybe someday.
Anyway, if we can get the SQ we want by fixing known problems rather than by hoping that higher sample rate formats will do the trick, we very well may end up with better results. That's the direction it looks like Benchmark decided to go.
Thanks for the comments. It is apparent that the DAC3 and most other DACs which also use the ESS Sabre chips actually are very different. Possibly the only commonality is that they use some of the same chips but the engineering approach differs.
Having recognized that. Here is my progress on my 9038q2m. For some reason my system and the only thing that has been changing is this DAC project changed for the better the past week and a half or so. Could this be due to burn in? I don't know but it changed for the better. Psychological no. The slight hardness was convincing me to quickly put together the ESS recommended IV section as well as the AVCC circuit.
Then the OSCON Sepc caps I had ordered arrived. So this saturday morning. I removed the stock caps as well as the 330uF Panasonic HFQs I had added and put in 470uF SEPC
OS-CON™, SEPC Series - Panasonic Electronic Components - Radial Leaded, Can | Online Catalog | DigiKey Electronics
for the AVCC line. This is powered by the LT3042 3.3V reg which replaced the AMS117.
Then I listened and this was the result. Bottom octaves were better defined. More surprisingly was that it cleaned up a lot of the hardness I was hearing. The music was much more agile. Then I listened some more and the cymbals and brushwork was noticeably improved..a lot of the splashiness was gone and it was more defined. Vocals were now spot on. Individual singers could now be easily discerned when in groups.
What I then put on was a 317 preregulator to feed the LT3042 and the sound benefited as the smaller details were a little more obvious but nothing large a small difference but for the small cost of a 317 board from Ebay Less than $4, it will stay.
The project is still far from finished as this will possibly delay my IV section and what will be interesting is what will be the result of adding that OP amp AVCC section as opposed to feeding from the LT3042.
The oscillator is still stock and no ASRC board has been added as yet.
By far the most improvement to the sound had emanated from upgrading the AVCC supply on this board. The OP amp choice is relatively small in comparison at this point.
I will let the new caps burn in a few days and see what happens.
Having recognized that. Here is my progress on my 9038q2m. For some reason my system and the only thing that has been changing is this DAC project changed for the better the past week and a half or so. Could this be due to burn in? I don't know but it changed for the better. Psychological no. The slight hardness was convincing me to quickly put together the ESS recommended IV section as well as the AVCC circuit.
Then the OSCON Sepc caps I had ordered arrived. So this saturday morning. I removed the stock caps as well as the 330uF Panasonic HFQs I had added and put in 470uF SEPC
OS-CON™, SEPC Series - Panasonic Electronic Components - Radial Leaded, Can | Online Catalog | DigiKey Electronics
for the AVCC line. This is powered by the LT3042 3.3V reg which replaced the AMS117.
Then I listened and this was the result. Bottom octaves were better defined. More surprisingly was that it cleaned up a lot of the hardness I was hearing. The music was much more agile. Then I listened some more and the cymbals and brushwork was noticeably improved..a lot of the splashiness was gone and it was more defined. Vocals were now spot on. Individual singers could now be easily discerned when in groups.
What I then put on was a 317 preregulator to feed the LT3042 and the sound benefited as the smaller details were a little more obvious but nothing large a small difference but for the small cost of a 317 board from Ebay Less than $4, it will stay.
The project is still far from finished as this will possibly delay my IV section and what will be interesting is what will be the result of adding that OP amp AVCC section as opposed to feeding from the LT3042.
The oscillator is still stock and no ASRC board has been added as yet.
By far the most improvement to the sound had emanated from upgrading the AVCC supply on this board. The OP amp choice is relatively small in comparison at this point.
I will let the new caps burn in a few days and see what happens.
Mikett, Good news. Don't know how far you will take the project, but I can tell you if you keep modding as suggested it will keep audibly improving. With each new mod you might think you are getting really close, but with the next mod you will probably find out not not all, you just got what seems like that much closer again. Eventually, it gets trickier to figure out what better really is. You can mod more and find out it seems to be getting worse, when in reality you may have just made it enough better to expose another problem that was previously masked. You aren't near that point yet. The smallest problems with power supplies don't show up until most everything else has been done. That is what I found anyway, and I never did try to take any shortcuts with AVCC.
Anyway, please keep us informed. If you get to the point where a recommended mod seems to make things worse in some way, say, maybe too bright in the bass, or maybe too much cymbals, then please let us know and we will try to work through it with you.
Anyway, please keep us informed. If you get to the point where a recommended mod seems to make things worse in some way, say, maybe too bright in the bass, or maybe too much cymbals, then please let us know and we will try to work through it with you.
Hi Mikett,
Could you please post some pictures of your board?
I'm curious about the 317 preregulator and the 470uF SEPC.
I already add LT3042 and I will change the oscillator on my board soon, so I think I can add your mods too.
thank you for your help
A little update on my own progress:
Still working on Arduino software to allow adjustment of harmonic distortion compensation. Want to have that working before I start on measurements. Tried various ways to input numerical data such as some measurement instruments do, but nothing quick and simple works very well. So going to keyboard input. Turns out Arduino doesn't directly support formatted input from a terminal program. ASCII characters in a buffer are all you get, so if a number is needed such as a signed integer then one has to write one's own code to do the conversion and any error handling. What a bother, but its getting there.
Also, noticed my I/V differential output opamp running a little warm. Found a couple of millivolts peak to peak of ~6MHz RF on the output pins. Added some COG caps as I probably should have done in the first place (from inverting to output, and from non-inverting to ground), which is closer to the filtering ESS recommends. Opamp is a little cooler, RF on output is nearly zero, and I believe distortion is reduced. Don't know how RF was getting into there. Couldn't measure any on power rails or non-inverting input. Maybe could have been due to radiated RFI, or maybe too small for me to see with my scope. As usual, I was not setup to measure audio distortion before and after. Maybe I will get better setup for the next project, I hope so. This experience is providing a lot of motivation.
Still working on Arduino software to allow adjustment of harmonic distortion compensation. Want to have that working before I start on measurements. Tried various ways to input numerical data such as some measurement instruments do, but nothing quick and simple works very well. So going to keyboard input. Turns out Arduino doesn't directly support formatted input from a terminal program. ASCII characters in a buffer are all you get, so if a number is needed such as a signed integer then one has to write one's own code to do the conversion and any error handling. What a bother, but its getting there.
Also, noticed my I/V differential output opamp running a little warm. Found a couple of millivolts peak to peak of ~6MHz RF on the output pins. Added some COG caps as I probably should have done in the first place (from inverting to output, and from non-inverting to ground), which is closer to the filtering ESS recommends. Opamp is a little cooler, RF on output is nearly zero, and I believe distortion is reduced. Don't know how RF was getting into there. Couldn't measure any on power rails or non-inverting input. Maybe could have been due to radiated RFI, or maybe too small for me to see with my scope. As usual, I was not setup to measure audio distortion before and after. Maybe I will get better setup for the next project, I hope so. This experience is providing a lot of motivation.
Last edited:
Markw4, those oscillations are the very reason why when my parts bin is done I am done. There is no way some guy like me with no instrumentation can put something decent without a guess. So it must come down to kits which are proven and even then a scope at least is required and a decent scope these days.
OK so I started laying out the textbook circuit by ESS and I discovered I read the wrong number and did not realize I did not have the 100pfs.
But so far this is what is happening.
The AVCC conditioning and IV circuit will be done on two dual op amps per channel. I thought about using a 5534 and then possibly a 797 for AVCC and I was afraid of stability and compensation. So I opted to use something more stable or the likelihood of stability is greater.
I am planning on trying to use the SEPC as the input filter for the conditioning circuit. This will cause a slow ramp up of the AVCC output and I am not sure if that is good but since AVCC has a range of voltage I suppose that is OK. The picture is one channel. So there will be four additional op amps. A second Sulzer board will be added for dual mono for each AVCC and IV output.
The perfboard is larger than need be. Where markw4 goes below. I plan on putting my external board so it is a bridge sitting above the 9038q2m board. That way the leads are minimized in the final layout.
What I have not figured out as yet is that my offset divider is on the external board. Do I need to bring that back to the 3038q2m board?
I will probably hack away at the 9038q2m next weekend. So I will bring the DAC outputs to the external board, take some 3.3v from the LT3042, feed AVCC back to the DAC and output sound entirely from my external contraption. At least hoping to.
OK so I started laying out the textbook circuit by ESS and I discovered I read the wrong number and did not realize I did not have the 100pfs.
But so far this is what is happening.
The AVCC conditioning and IV circuit will be done on two dual op amps per channel. I thought about using a 5534 and then possibly a 797 for AVCC and I was afraid of stability and compensation. So I opted to use something more stable or the likelihood of stability is greater.
I am planning on trying to use the SEPC as the input filter for the conditioning circuit. This will cause a slow ramp up of the AVCC output and I am not sure if that is good but since AVCC has a range of voltage I suppose that is OK. The picture is one channel. So there will be four additional op amps. A second Sulzer board will be added for dual mono for each AVCC and IV output.
The perfboard is larger than need be. Where markw4 goes below. I plan on putting my external board so it is a bridge sitting above the 9038q2m board. That way the leads are minimized in the final layout.
What I have not figured out as yet is that my offset divider is on the external board. Do I need to bring that back to the 3038q2m board?
I will probably hack away at the 9038q2m next weekend. So I will bring the DAC outputs to the external board, take some 3.3v from the LT3042, feed AVCC back to the DAC and output sound entirely from my external contraption. At least hoping to.
Attachments
Last edited: