@Ian, Regarding the headphone amp you linked to on ebay, I do encourage people to build a good LME49600 headphone amp to go with this DAC since many people don't have an amplifier with the resolving power to make audible everything this DAC can do. If building a HPA, I would use the same rules as for the DAC circuitry though which means a ground plane, proper decoupling, and high quality passive components.
Regarding a buffer for this DAC, I'm not sure exactly what is meant by that term. Maybe if we look at some schematics more will become clear. Below are some schematics taken from the ESS document I referred to previously on their download page. It takes one of those complete schematic circuits for each of the two stereo dac output channels. The box that shows the I/V stage is the same stuff I put on one end of the bottom of my DAC board, except the I/V feedback resistors should be 820 ohms for this dac. The box showing the differential output/filter is what I reused the socket and pads on the top of the board for, except I did not implement the filter. The two resistors that are between the boxes serve some function for both of the boxes. Also, Abraxalito suggested changing some of the filter component values back in a post here: http://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-173.html#post5479364 And, I would agree with him that scaling the filter values would be good. On the other hand, I just had some .01% 10k SMD thin film resistors in stock which I used to make a differential summing circuit without the filtering. But, the filtering is probably a good idea. (Actually, I think somebody sketched up a schematic of what I did, so maybe I can find that for you, but the point would still be the same: some fabrication is necessary -- Hey, Just found it!)
Thing is, you can't buy boards on ebay to do all this. You would need to fabricate all this circuitry in the schematics twice, once for the left output channel and once for the right. That's what I did, and it was somewhat challenging for an old man with poor eyesight and sometimes a bit shaky hands. Somehow it worked though, even though at times I had my doubts.
So, all the above is why I have been asking you if you are sure you are up to this. If not, then I would again remind that Mikett started with another DAC board that he says is much easier to modify and much more of what is needed can be purchased already assembled on ebay. So, if that sounds like it might be more doable then maybe worth going to take a look at his thread and asking him to tell you a little more about how hard it was to fabricate. His thread is here: Moving from 9038Q2M to a 9028Pro board
Regarding a buffer for this DAC, I'm not sure exactly what is meant by that term. Maybe if we look at some schematics more will become clear. Below are some schematics taken from the ESS document I referred to previously on their download page. It takes one of those complete schematic circuits for each of the two stereo dac output channels. The box that shows the I/V stage is the same stuff I put on one end of the bottom of my DAC board, except the I/V feedback resistors should be 820 ohms for this dac. The box showing the differential output/filter is what I reused the socket and pads on the top of the board for, except I did not implement the filter. The two resistors that are between the boxes serve some function for both of the boxes. Also, Abraxalito suggested changing some of the filter component values back in a post here: http://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-173.html#post5479364 And, I would agree with him that scaling the filter values would be good. On the other hand, I just had some .01% 10k SMD thin film resistors in stock which I used to make a differential summing circuit without the filtering. But, the filtering is probably a good idea. (Actually, I think somebody sketched up a schematic of what I did, so maybe I can find that for you, but the point would still be the same: some fabrication is necessary -- Hey, Just found it!)
Thing is, you can't buy boards on ebay to do all this. You would need to fabricate all this circuitry in the schematics twice, once for the left output channel and once for the right. That's what I did, and it was somewhat challenging for an old man with poor eyesight and sometimes a bit shaky hands. Somehow it worked though, even though at times I had my doubts.
So, all the above is why I have been asking you if you are sure you are up to this. If not, then I would again remind that Mikett started with another DAC board that he says is much easier to modify and much more of what is needed can be purchased already assembled on ebay. So, if that sounds like it might be more doable then maybe worth going to take a look at his thread and asking him to tell you a little more about how hard it was to fabricate. His thread is here: Moving from 9038Q2M to a 9028Pro board
Attachments
Last edited:
I don't listen to headphones, can't stand the things.
A buffer is simply a line stage amplifier with unity gain.
I don't see the need to stick strictly to what ESS suggests, most often, the manufacturer's designs are done by electronics engineers, not audio engineers and they are designed from the POV of what is electrically going to work, not what sounds good.
There will be multiple ways of implementing an output stage that sounds good, not just the ESS recommended way, take CD players for instance, while many stick to the reference designs provided by Phillips and Sony, the best sounding ones throw those designs away and design their own, for the reason I explained above - they are designed by electronics engineers, not audio engineers.
An example of this is Marantz's HDAM, which is basically a diamond buffer, it uses a bunch of transistors and discrete components instead of op amps. Many people find this sounds much better, although some don't.
In short, I see no reason to stick rigidly to ESS's reference design, there will be more than one way of achieving good SQ and with much less hassle than fabricating from scratch, after all, people have been designing unity gain output stages for a long time and there are many proven designs, so it is largely a matter of finding one that works effectively with this ESS board.
A buffer is simply a line stage amplifier with unity gain.
I don't see the need to stick strictly to what ESS suggests, most often, the manufacturer's designs are done by electronics engineers, not audio engineers and they are designed from the POV of what is electrically going to work, not what sounds good.
There will be multiple ways of implementing an output stage that sounds good, not just the ESS recommended way, take CD players for instance, while many stick to the reference designs provided by Phillips and Sony, the best sounding ones throw those designs away and design their own, for the reason I explained above - they are designed by electronics engineers, not audio engineers.
An example of this is Marantz's HDAM, which is basically a diamond buffer, it uses a bunch of transistors and discrete components instead of op amps. Many people find this sounds much better, although some don't.
In short, I see no reason to stick rigidly to ESS's reference design, there will be more than one way of achieving good SQ and with much less hassle than fabricating from scratch, after all, people have been designing unity gain output stages for a long time and there are many proven designs, so it is largely a matter of finding one that works effectively with this ESS board.
Ian, you might have some points there, except these DACs are different than what you are familiar with. Also, it isn't that there is a need to stick rigidly to ESS reference designs, as you can plainly see from the schematics I didn't stick to it exactly. However, I did stick to the things that are important for good sound quality. You can believe it or not as you wish.
I guess I'm not sure if is clear yet why the I/V stages, the differential stage, the filters, all the other stuff? It's all for good sound quality not for rigid compliance to arbitrary useless nonsense.
If you don't understand the need for this or that thing that seem tedious and or unnecessary you can always ask about it and what effect there may be on sound quality.
Or if you don't think I am giving you good advice then I will be happy to leave you alone to do as you like. No problem there.
But, I cannot in good faith advise you to use a LME49600 diamond buffer as a line level buffer given it would raise the very lowest distortion you could get from -120dB to -105dB. That's giving away 15dB of distortion for no useful purpose except maybe if you wanted to drive headphones with it. At least you could get enough current that way. Most people would not want the extra distortion for a general purpose line level buffer.
And by the way, no line level buffer is needed. The differential output stage can drive a preamp or power amp input just fine. So, why would anyone want another buffer anyway? You could always ask if a line level buffer would be needed for anything.
If you are thinking of skipping the I/V and differential stage stuff altogether because you CD player doesn't need it or something like that, and instead going into directly into a line level buffer then you would be increasing the best distortion floor you could get from around -120dB to closer to -70dB, which by the way will have pretty poor sound quality.
Lastly, if you think this is frustrating for you, it is for me too. Maybe we just aren't a good match for working together over the internet, I don't know.
I guess I'm not sure if is clear yet why the I/V stages, the differential stage, the filters, all the other stuff? It's all for good sound quality not for rigid compliance to arbitrary useless nonsense.
If you don't understand the need for this or that thing that seem tedious and or unnecessary you can always ask about it and what effect there may be on sound quality.
Or if you don't think I am giving you good advice then I will be happy to leave you alone to do as you like. No problem there.
But, I cannot in good faith advise you to use a LME49600 diamond buffer as a line level buffer given it would raise the very lowest distortion you could get from -120dB to -105dB. That's giving away 15dB of distortion for no useful purpose except maybe if you wanted to drive headphones with it. At least you could get enough current that way. Most people would not want the extra distortion for a general purpose line level buffer.
And by the way, no line level buffer is needed. The differential output stage can drive a preamp or power amp input just fine. So, why would anyone want another buffer anyway? You could always ask if a line level buffer would be needed for anything.
If you are thinking of skipping the I/V and differential stage stuff altogether because you CD player doesn't need it or something like that, and instead going into directly into a line level buffer then you would be increasing the best distortion floor you could get from around -120dB to closer to -70dB, which by the way will have pretty poor sound quality.
Lastly, if you think this is frustrating for you, it is for me too. Maybe we just aren't a good match for working together over the internet, I don't know.
Last edited:
My understanding was that the best approach in SQ terms is to take the raw DAC output and feed it into an output stage, thus bypassing whatever output stage is built into the board, which means that SQ is then dependent on the implementation and design of the output stage.
There will be more than one design of output stage that will work well, different designs will have different pros and cons, I was hoping that people had tried various output stages with this board and could share their experiences. One thing I am wondering about is discrete vs op amp, many people swear by discrete output stages as being superior to opamps.
I'm just interested in exploring different options before making a decision as to what I will try to build and implement.
The way I see it, most of the effort should be put into the power supply and output stage, that way, the DAC board can be changed and all the effort and expense involved in creating the power supply and output stage isn't wasted.
There will be more than one design of output stage that will work well, different designs will have different pros and cons, I was hoping that people had tried various output stages with this board and could share their experiences. One thing I am wondering about is discrete vs op amp, many people swear by discrete output stages as being superior to opamps.
I'm just interested in exploring different options before making a decision as to what I will try to build and implement.
The way I see it, most of the effort should be put into the power supply and output stage, that way, the DAC board can be changed and all the effort and expense involved in creating the power supply and output stage isn't wasted.
@Ian - I don't know if anyone's mentioned this before, but seems to me that @opc's NTD1 project might be of interest to you as you're looking at discrete stages.
opc's NTD1 project
One drawback of his approach (at least as it seems to me) is his use of high voltage supplies and consequent high power dissipation. There are other ways to crack that particular nut which don't call for double-digit Watt dissipation.
opc's NTD1 project
One drawback of his approach (at least as it seems to me) is his use of high voltage supplies and consequent high power dissipation. There are other ways to crack that particular nut which don't call for double-digit Watt dissipation.
Mark put a lot of effort in trying to help you, but since you have made up your mind that you know better, this is what is going to happen to you....
ESS designers also put a lot of effort into testing and recommending circuits that will get the best out of their chips. I agree with Mark that those recommendations should at least be the starting point for design of the ancillary circuits that support their DAC designs. If you think you can do better, good luck.
Maybe you should check out what Twisted Pear is doing with I/V stages to understand better. They are known to have worked quite closely with ESS.
You need to educate yourself about the difference between current output dacs, and voltage output dacs and why a good I/V stage is so important for ESS dacs..
Seems to me here there's a missing word otherwise some people might draw the wrong conclusions. That missing word is 'measurements' to wit '...that will get the best measurements out of their chips'.
Thanks for that, I'll check it out.
I'm only considering a discrete output stage because I'm still at the stage in my project where I'm open to all options and gathering info in order to guide my design choices.
I've done a lot of reading on the subject in the last few days, to the point where my head hurts, and there simply is no consensus about whether discrete or op amp is better, in SQ terms. Certainly, op amps are cheaper, simpler, easier to implement and much less space consuming, and there are a good number of people who say that modern op amps can sound every bit as good as discrete, but conversely there are still many who maintain that a discrete solution is better.
Which is going to be subjectively better with this ES9038 board is a rather more difficult question to answer, probably the answer is that both discrete and op amp solutions can work well with this board and give great SQ, with the most important thing being the implementation and the quality of the power supply.
I'm only considering a discrete output stage because I'm still at the stage in my project where I'm open to all options and gathering info in order to guide my design choices.
I've done a lot of reading on the subject in the last few days, to the point where my head hurts, and there simply is no consensus about whether discrete or op amp is better, in SQ terms. Certainly, op amps are cheaper, simpler, easier to implement and much less space consuming, and there are a good number of people who say that modern op amps can sound every bit as good as discrete, but conversely there are still many who maintain that a discrete solution is better.
Which is going to be subjectively better with this ES9038 board is a rather more difficult question to answer, probably the answer is that both discrete and op amp solutions can work well with this board and give great SQ, with the most important thing being the implementation and the quality of the power supply.
Look, I don't think I know better than anyone, I am merely asking questions and if that upsets some people, then so be it.
I know that a good I/V stage is important, I never said otherwise, all I said was that there will be more than one way of implementing a good sounding output stage and that I would like to consider a number of options before making a final decision, which is just good common sense.
I might end up using a valve output stage, or I might end up doing something with op amps close to what ESS recommends, or I might go for a discrete solution, at this moment in time I'm still trying to weigh up the pros and cons of each, to not become wedded to one particular solution until I am confident I have been thorough in exploring the possible options. If some people get upset because of this then it is a failing of theirs, not mine.
I won't be pushed into choosing any particular path until I am ready to make the decision about which path to take and I simply haven't reached that point yet.
"To avoid voltage drops on the wires between regulator and load circuit, and to avoid ground-induced noise and ripple" so called "remote sensing option" are implemented into the circuits, such a Salas regs (BiB, ReflektorD) and Superreg as well. "In this way, the regulator regulates not just its local output but rather the voltage directly at the load."
SSLV1.1 builds & fairy tales
SSLV1.1 builds & fairy tales
Last edited:
ESS doesn't specify a number for AVCC line noise. What they do is say the following:
AVCC:
"The choice of opamp varies for each design and budget. The power supply circuit shown in Figure 4 (opamp AVCC supply) is necessary to provide a clean 3.3V to the analog section of the Sabre DAC. The opamp used for this circuit is the most critical for optimizing the THD and SNR performance. Use of a poorly designed opamp will degrade both the THD and SNR."
"The +3.3V required for the analog section of the DAC needs to be as quiet and well regulated as possible. "The design uses the +3.3V DVCC as a reference and filters this signal with a RC filter. The opamp is configured in a buffer stage to supply the +3.3V for AVCC. As shown in Figure 8, the decoupling capacitors should be kept close to the supply pins and the filter should be close to the opamp input. The output trace with should be large and provide a very low impedance path to the DAC AVCC pins."
Then they suggest by example to measure DNR and THD+N at the end of the output stage, to listen, and to try to optimize performance that way. Since as engineers we don't have a good way to translate listening into specifications, they don't have much advice on how to do that part. In the video on their website Martin Mallison of ESS says he is not good at the listening part but they have people who are good at that. They do have some tables showing how different op amps used to make an AVCC supply can affect output DNR and THD, so I think they would suggest comparing AVCC solutions by measuring and listening at the line level output.
My own position in this thread has been to the effect that since most people here don't seem to be equipped for measuring state of the art DAC circuits, that we should try to make use of proven designs that others have already measured. We can listen of course, and I think we all do. But, listening alone without measuring is fraught with problems arising from human perceptual biases. I could say more about that, but will leave off here.
AVCC:
"The choice of opamp varies for each design and budget. The power supply circuit shown in Figure 4 (opamp AVCC supply) is necessary to provide a clean 3.3V to the analog section of the Sabre DAC. The opamp used for this circuit is the most critical for optimizing the THD and SNR performance. Use of a poorly designed opamp will degrade both the THD and SNR."
"The +3.3V required for the analog section of the DAC needs to be as quiet and well regulated as possible. "The design uses the +3.3V DVCC as a reference and filters this signal with a RC filter. The opamp is configured in a buffer stage to supply the +3.3V for AVCC. As shown in Figure 8, the decoupling capacitors should be kept close to the supply pins and the filter should be close to the opamp input. The output trace with should be large and provide a very low impedance path to the DAC AVCC pins."
Then they suggest by example to measure DNR and THD+N at the end of the output stage, to listen, and to try to optimize performance that way. Since as engineers we don't have a good way to translate listening into specifications, they don't have much advice on how to do that part. In the video on their website Martin Mallison of ESS says he is not good at the listening part but they have people who are good at that. They do have some tables showing how different op amps used to make an AVCC supply can affect output DNR and THD, so I think they would suggest comparing AVCC solutions by measuring and listening at the line level output.
My own position in this thread has been to the effect that since most people here don't seem to be equipped for measuring state of the art DAC circuits, that we should try to make use of proven designs that others have already measured. We can listen of course, and I think we all do. But, listening alone without measuring is fraught with problems arising from human perceptual biases. I could say more about that, but will leave off here.
Last edited:
The only way how to reach any point here is to take a soldering tool, build a circuit, measure it, listen to it. People over here have done this and share their results. It is pretty unclear what kind of device are you aiming to construct utilizing this rather primitive board?
Yes, I agree, my soldering irons are at the ready and I have accumulated a pile of parts, what's holding me up is I'm still waiting for one or two key parts to arrive on the slow boat from China.
When all the necessary parts have arrived, which will hopefully be in the next week or so, I shall start a new thread for the build.
I suppose I should give some details of what I intend to do with this project, which I shall do when I start my own thread. In short I aim to build a DAC to connect to my PC in order to play my music collection (which is almost all FLAC files) through my HiFi setup which currently consists of a Yamaha DSP-E800 which is doing double duty as both DAC (it has a 24/96 DAC built in) and pre-amp, I have a Linn Roomamp as power amp. My initial goal is to better the SQ I am getting using the DSP-E800's DAC, I can then try to make improvements in the SQ by tweaking and upgrading my initial build.
I also intend to build a pre-amp so I can completely replace the DSP-E800, but the DAC comes first. There's nothing wrong with the SQ of my current setup, in fact it sounds very good to my ears, a marked improvement over my previous Marantz PM66SE integrated amp which died on me a couple of months ago. So really, I'm doing this project more as a fun hobby than as a desire for better SQ and I am well aware that I might not be able to achieve much of an SQ improvement due to the limitations of my speakers, their placement and the room itself, which are factors that I am not really able to make any changes to, other than an upgrade of the speakers, which are Mission 731s I've owned since 1994 and have always been very happy with as they are very detailed and don't seem to have much if any colouration. They aren't very bass-y, but living in a terraced house with neighbours either side, this is something I am not overly bothered about, hence I have never added a sub.
When all the necessary parts have arrived, which will hopefully be in the next week or so, I shall start a new thread for the build.
I suppose I should give some details of what I intend to do with this project, which I shall do when I start my own thread. In short I aim to build a DAC to connect to my PC in order to play my music collection (which is almost all FLAC files) through my HiFi setup which currently consists of a Yamaha DSP-E800 which is doing double duty as both DAC (it has a 24/96 DAC built in) and pre-amp, I have a Linn Roomamp as power amp. My initial goal is to better the SQ I am getting using the DSP-E800's DAC, I can then try to make improvements in the SQ by tweaking and upgrading my initial build.
I also intend to build a pre-amp so I can completely replace the DSP-E800, but the DAC comes first. There's nothing wrong with the SQ of my current setup, in fact it sounds very good to my ears, a marked improvement over my previous Marantz PM66SE integrated amp which died on me a couple of months ago. So really, I'm doing this project more as a fun hobby than as a desire for better SQ and I am well aware that I might not be able to achieve much of an SQ improvement due to the limitations of my speakers, their placement and the room itself, which are factors that I am not really able to make any changes to, other than an upgrade of the speakers, which are Mission 731s I've owned since 1994 and have always been very happy with as they are very detailed and don't seem to have much if any colouration. They aren't very bass-y, but living in a terraced house with neighbours either side, this is something I am not overly bothered about, hence I have never added a sub.
You really can’t expect the Mission 731 to allow you to hear the SQ improvement. You should get a high end headphone with a good head amp. I had a pair of Mission 700 for years and know what the Mission sounds like. On the other hand, even a stock form Chinese ES9038Q2M board without I/V will be Hugh improvement over the 20+ years old Yamaha DSP-E800. DAC technology marches in leaps and bonds.
It is perfect ok to ask questions, but you must have your own goal.
Last edited:
The problem is, I simply hate headphones, I have never been able to enjoy listening to music on headphones.
I also have a pair of Jamo Euro 7B speakers, they are probably not any better than the Missions, I haven't listened to them for a few years but my memory is that they are more bass-y but less detailed than the Missions - good for loud rock music but overall, less good than the Missions for more refined types of music.
I suppose I should look into a speaker upgrade, anyone care to recommend some models that I could find secondhand around the 50-70ukp price range?
I also have a pair of Jamo Euro 7B speakers, they are probably not any better than the Missions, I haven't listened to them for a few years but my memory is that they are more bass-y but less detailed than the Missions - good for loud rock music but overall, less good than the Missions for more refined types of music.
I suppose I should look into a speaker upgrade, anyone care to recommend some models that I could find secondhand around the 50-70ukp price range?
Hey guys, awhile back a few people expressed interest in Arduino access to DAC registers and in particular to allow adjustment of harmonic distortion compensation: http://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-142.html#post5452170 And of course, other interesting things could be done too such as possibly trying master mode I2S. At the time I was also thinking about a way that might work to do that with something like SRC4392 where it could use the DAC master clock as its reference clock so the two clocks would be perfectly in sync. Then ASRC in the DAC could be turned off. Might work with XMOS or Amanero boards too although the DAC might need a clock swap in that case, or then again maybe not, depending on how one wanted to architect the clocking.
May I ask if anybody ever gave that a try or maybe did something similar, or am I the only one here to delve into that aspect of DAC modding?
May I ask if anybody ever gave that a try or maybe did something similar, or am I the only one here to delve into that aspect of DAC modding?
There are ES9023 add-on boards for the Raspberry Pi which might be able to do something along those lines so studying those may yield answers.
ES9023 I2S vers DAC Sabre 24bit/192KHZ Analogique for Raspberry PI top | eBay
ES9023 I2S vers DAC Sabre 24bit/192KHZ Analogique for Raspberry PI top | eBay