Hi Mark,
Thank for all your work and teaching mods on this dac card.
Speaking for myself, it is a matter of complexity, but you already know that
For upsampling files, I use sox in Moode and a kali recloacker between the RPI3 and the 9038q2m board.
Do you think using SRC4392 will improve SQ comparing to software upsampling + Kali?
About the LME49600 amp, I looked at your board picture and I don't figure how it is connected to the dac board, so i think it is out of my league for now. I was thinking that the LME49600 board was replacing the opamp on the card, but it looks that it is not the case
Hello Mark,
same for me. I do upsampling from all formats (mainly 44.1 kHz) to 96kHz before convolving with FIR room correction filters in my Raspberry Pi. Then I give this out to the DAC via I2S. Same question as Terry - would a SRC have a big benefit? Is there the need of programming the SRC or is it all adjustable by on-board switches (e.g. 96kHz I2S in -> 192 kHz I2S out)?
As I do not listen to headphones the LME49600 is not interesting for me.
Thank you for your great Job here!
same for me. I do upsampling from all formats (mainly 44.1 kHz) to 96kHz before convolving with FIR room correction filters in my Raspberry Pi. Then I give this out to the DAC via I2S. Same question as Terry - would a SRC have a big benefit? Is there the need of programming the SRC or is it all adjustable by on-board switches (e.g. 96kHz I2S in -> 192 kHz I2S out)?
As I do not listen to headphones the LME49600 is not interesting for me.
Thank you for your great Job here!
I've been having fun with this 9038q2m board. Make no mistake. It has been a great learning experience that has whet my appetite. Now if I were to start again I think I have found a potential superior candidate. ES9028 ES9028PRO DAC Q8 Ver2.2 HiFi Audio Decoder PCM384K / DSD128 | eBay
The schematic for the board can be found when you look at the assembled in case version. Scroll through the pictures of the listing towards the end. Q8 Ver2.2 ES9028 DAC ES9028PRO HiFi Audio Decoder Finished in case | eBay
Now considering the cost of the assembled board versus the SMPC unit, yet it is more costly. However consider the value. You get first a 9028pro which has 4 channels tied together for better SNR. You get a linear shunt PS for the analog built in. Now look at the PS section for the digital side. You get three 3 term regs 317s plus 3 LT1763s as well as another 5 ADP regulators. Some digital lines are filtered three times. Finally I looked at the output section and the topology uses the same IV circuit as ESS recommends pretty much or can be easily changed to match. It also appears to use some audio recommended caps on the analog side as well. The only other thing is that it does not come with the OP amp section for the 9028 but I wonder if the opamp can drive the 9028 avcc?
Now the more interesting thing is that when they throw in a case and transformer, the deal is even more attractive. How does it sound? I don't know but I am tempted to pull the trigger.
This one appears to have potential to get closer to the DAC3 maybe?
The schematic for the board can be found when you look at the assembled in case version. Scroll through the pictures of the listing towards the end. Q8 Ver2.2 ES9028 DAC ES9028PRO HiFi Audio Decoder Finished in case | eBay
Now considering the cost of the assembled board versus the SMPC unit, yet it is more costly. However consider the value. You get first a 9028pro which has 4 channels tied together for better SNR. You get a linear shunt PS for the analog built in. Now look at the PS section for the digital side. You get three 3 term regs 317s plus 3 LT1763s as well as another 5 ADP regulators. Some digital lines are filtered three times. Finally I looked at the output section and the topology uses the same IV circuit as ESS recommends pretty much or can be easily changed to match. It also appears to use some audio recommended caps on the analog side as well. The only other thing is that it does not come with the OP amp section for the 9028 but I wonder if the opamp can drive the 9028 avcc?
Now the more interesting thing is that when they throw in a case and transformer, the deal is even more attractive. How does it sound? I don't know but I am tempted to pull the trigger.
This one appears to have potential to get closer to the DAC3 maybe?
Last edited:
For those already software upsampling, that should be fine. No benefit from adding hardware to do it in that case, assuming that one has selected good sounding software for the task.
Experience has shown that upsampler sound quality can't be predicted by looking at specs alone, is all. Then again, I may be pickier than most about sound quality, perhaps more of a curse than a useful ability.
Regarding convolution with room correction filters, personally, I have never had much luck with convolutional EQ (or reverb, for that matter). The sound quality seems to get more damaged than helped by convolvers I have tried. Also, it usually sounds better to use the minimum amount EQ that sounds acceptable rather than trying for ruler-flat correction. Automated room correction algorithms often overdo or under-do it.
If I wanted to do room correction EQ it would probably be done by adding VST plugin non-convolutional EQ after some experimentation to find the best sounding one readily available.
Experience has shown that upsampler sound quality can't be predicted by looking at specs alone, is all. Then again, I may be pickier than most about sound quality, perhaps more of a curse than a useful ability.
Regarding convolution with room correction filters, personally, I have never had much luck with convolutional EQ (or reverb, for that matter). The sound quality seems to get more damaged than helped by convolvers I have tried. Also, it usually sounds better to use the minimum amount EQ that sounds acceptable rather than trying for ruler-flat correction. Automated room correction algorithms often overdo or under-do it.
If I wanted to do room correction EQ it would probably be done by adding VST plugin non-convolutional EQ after some experimentation to find the best sounding one readily available.
The SRC board I am recommending has switches to set the desired upsampling. As is, it always upsamples as necessary to produce the switch-selected output sample rate. So, it means you set it once and forget it. Also, it has coaxial, TOSLINK, and I2S outputs all available at the same time. Just plug into the one you want.
Later on, if one wanted to use the SRC4392 chip on the board to do other things or in some different configuration, then the chip could be programmed with a small microcontroller.
Regarding the LME49600 headphone amp, if one has a clean, accurate, low-distortion amp of some kind then that should suffice. I was just thinking that if someone was using most class D amplifiers or any other less-than-very-low-distortion amp then it can be hard to tell exactly how well one's DAC sounds. Whatever is the weakest link in the reproduction chain may tend to dominate in producing not-very-good sound quality for the whole system. In that case, it can sometimes be hard to know what the problem is. Based on specs alone, one may be misled into thinking everything is really good when in reality it is not. That can happen more easily than one might expect. The best insurance is to have a very, very low distortion headphone amp or another amp you can rely on for doing test listening.
The headphone amp connects to the DAC the same as any other amp would. The input of the amp plugs into the output jack of the DAC. That way the headphone amp could be used for listening to the DAC or any other line-level audio source.
Later on, if one wanted to use the SRC4392 chip on the board to do other things or in some different configuration, then the chip could be programmed with a small microcontroller.
Regarding the LME49600 headphone amp, if one has a clean, accurate, low-distortion amp of some kind then that should suffice. I was just thinking that if someone was using most class D amplifiers or any other less-than-very-low-distortion amp then it can be hard to tell exactly how well one's DAC sounds. Whatever is the weakest link in the reproduction chain may tend to dominate in producing not-very-good sound quality for the whole system. In that case, it can sometimes be hard to know what the problem is. Based on specs alone, one may be misled into thinking everything is really good when in reality it is not. That can happen more easily than one might expect. The best insurance is to have a very, very low distortion headphone amp or another amp you can rely on for doing test listening.
The headphone amp connects to the DAC the same as any other amp would. The input of the amp plugs into the output jack of the DAC. That way the headphone amp could be used for listening to the DAC or any other line-level audio source.
Last edited:
@Mikett, You may well have a point there regarding a choice of starting point. Don't know about the quality of all the components they used. The upgrades I did to my 1.06 board use really good parts and really good circuits which would be expected to probably produce a better sound quality than the board you are considering as it is without modification.
Since it is already a lot closer to good from the get-go, one may be tempted to leave it that way as a matter of convenience. Given the way the 9028 is being used on that board, in theory, it shouldn't sound any better than the Q2M on the board we have been modifying in this thread.
Also, noticed the schematic doesn't show the controller which is an essential part of the whole. However, having the schematic is a good thing.
I would say that with that board if you wanted to get it the sound quality up the where my modified Chinese DAC is, it would need:
Low-jitter clock upgrade.
AVCC power supply upgrade.
Linear power supply low-noise mods (diodes, regulators, etc.).
SRC4392 board.
Output stage passive component upgrades (we haven't talked about this and no one has asked, but we probably should talk about it).
Low distortion LME49600 headphone amp for test listening.
USB to SPDIF board (if desired).
Things we might get to here you might not be able to do without you
going through the non-disclosure process I am trying to get through:
Try master mode I2S with jitter correction off.
Intersample over fix for 9028 implemented via SRC4392.
EDIT: Something I am not sure about with the board you are considering would be ground planes and trace routing. The pictures I posted of the mod layout I used shows use of a ground plane and reasonably rational layout and wire routing. With a finished board, it can be really helpful if the designer did a good job, and a lot of rework if it was done unwisely. (If anyone else considering doing mods and would like to talk about layout considerations we could probably discuss that here.)
EDIT 2: Regarding the use of a 9038PRO with tied-together channels, the schematic they posted doesn't show that. It shows extra channels not connected to anything. Not sure they have enough IV stages on the board for all the potential 9038 output current either if outputs tied-together.
My overall opinion would be that I agree with you it would probably be a better PCB for starting with. Where I guess I would have some reservations would be in the potential total cost of the project for a final sound quality equivalent to my modified DAC.
Since it is already a lot closer to good from the get-go, one may be tempted to leave it that way as a matter of convenience. Given the way the 9028 is being used on that board, in theory, it shouldn't sound any better than the Q2M on the board we have been modifying in this thread.
Also, noticed the schematic doesn't show the controller which is an essential part of the whole. However, having the schematic is a good thing.
I would say that with that board if you wanted to get it the sound quality up the where my modified Chinese DAC is, it would need:
Low-jitter clock upgrade.
AVCC power supply upgrade.
Linear power supply low-noise mods (diodes, regulators, etc.).
SRC4392 board.
Output stage passive component upgrades (we haven't talked about this and no one has asked, but we probably should talk about it).
Low distortion LME49600 headphone amp for test listening.
USB to SPDIF board (if desired).
Things we might get to here you might not be able to do without you
going through the non-disclosure process I am trying to get through:
Try master mode I2S with jitter correction off.
Intersample over fix for 9028 implemented via SRC4392.
EDIT: Something I am not sure about with the board you are considering would be ground planes and trace routing. The pictures I posted of the mod layout I used shows use of a ground plane and reasonably rational layout and wire routing. With a finished board, it can be really helpful if the designer did a good job, and a lot of rework if it was done unwisely. (If anyone else considering doing mods and would like to talk about layout considerations we could probably discuss that here.)
EDIT 2: Regarding the use of a 9038PRO with tied-together channels, the schematic they posted doesn't show that. It shows extra channels not connected to anything. Not sure they have enough IV stages on the board for all the potential 9038 output current either if outputs tied-together.
My overall opinion would be that I agree with you it would probably be a better PCB for starting with. Where I guess I would have some reservations would be in the potential total cost of the project for a final sound quality equivalent to my modified DAC.
Last edited:
Hi,
What do you think of this 9038 pro card?
It's almost the same price.
Does yours has a better implementation/design?
NEW hifi TOP ES9038 ES9038PRO DAC decoder assembled board + TCXO 0.1PPM + remote control + option USB XMOS XU208 or Amanero dans Amplificateur de Electronique sur AliExpress.com | Alibaba Group
Last edited:
Ordered this the other day after seeing someone mention it on this thread
NEW ES9038Q2M Super DAC + Output Cattle | eBay
Will post measurements and impressions once I have it on hand
NEW ES9038Q2M Super DAC + Output Cattle | eBay
Will post measurements and impressions once I have it on hand
AVCC should have something better than only a voltage regulator. A linear super-regulator would be better.
There is some allowable voltage range for AVCC. I think it can be as high as 4.0 volts. The important things are that it is very stable, very low-noise, and low impedance across a across a wide range of frequencies.
The AVCC opamp circuit recommended by ESS has input filtering to keep the voltage at some stable value within the allowable range, good-opamp-grade low-noise behaviour, includes lots of feedback to keep output impedance low, and is aided by a big enough output cap to help filter out LF noise below the opamp noise corner. In other words, there are reasons ESS recommended what it did. If more current is needed for newer DACs, an opamp should still be used to keep it highly regulated and be located right next to the DAC chip as much as possible to help keep noise pickup and trace resistance issues at a minimum. One or more pass transistors may be needed to supply higher current needs. Essentially, the AVCC inputs on the DAC chip have no PSRR at all and directly affect output sound quality.
Last edited:
Mark,
http://www.analog.com/media/en/technical-documentation/data-sheets/3042fb.pdf
Pretty impressive.
http://www.analog.com/media/en/technical-documentation/data-sheets/3042fb.pdf
Pretty impressive.
Regarding some RF linear regulators, some concerns:
1) Maybe not so easy to DIY as there is a ground-pad/heatsink-pad on the bottom that would probably need to be cooked onto a board. Some people may be equipped to do it though.
2a) I had some 'funny' experiences with Silent Switcher which uses TI RF linear regulators. There was definitely leakage from input to output of noise from upstream switching supplies. That includes battery operated 5v USB power supplies which use switching to produce a regulated 5v output from whatever the battery voltage is.
2b) Same SS, had some issues even when using an upstream linear supply. DAC output was a bit bright in a way that did not sound quite right as though it needed a treble tone control I could turn down a little. In other words, it sounded like a little low level higher order or odd order harmonic distortion to me. Not sure at that point if it was characteristic of the DAC itself or something else until I connected the DAC to a very clean linear supply instead. It then became clear that particular not-quite-right aspect of sound quality (or the lack thereof) was associated with SS.
The SS observations may have to do with the particular implementation, possibly with the particular RF regulator parts or other parts on the boards, or perhaps some mix of things. Don't know.
My suspicion would be there is probably no reason why RF regulators should not work fine and in fact, I am pretty sure Benchmark uses that technology in places. Until I am sure I know how to make them work right and how to implement for the application at hand I am a little reluctant to recommend them to others. If you get it all worked out, by all means, please share.
1) Maybe not so easy to DIY as there is a ground-pad/heatsink-pad on the bottom that would probably need to be cooked onto a board. Some people may be equipped to do it though.
2a) I had some 'funny' experiences with Silent Switcher which uses TI RF linear regulators. There was definitely leakage from input to output of noise from upstream switching supplies. That includes battery operated 5v USB power supplies which use switching to produce a regulated 5v output from whatever the battery voltage is.
2b) Same SS, had some issues even when using an upstream linear supply. DAC output was a bit bright in a way that did not sound quite right as though it needed a treble tone control I could turn down a little. In other words, it sounded like a little low level higher order or odd order harmonic distortion to me. Not sure at that point if it was characteristic of the DAC itself or something else until I connected the DAC to a very clean linear supply instead. It then became clear that particular not-quite-right aspect of sound quality (or the lack thereof) was associated with SS.
The SS observations may have to do with the particular implementation, possibly with the particular RF regulator parts or other parts on the boards, or perhaps some mix of things. Don't know.
My suspicion would be there is probably no reason why RF regulators should not work fine and in fact, I am pretty sure Benchmark uses that technology in places. Until I am sure I know how to make them work right and how to implement for the application at hand I am a little reluctant to recommend them to others. If you get it all worked out, by all means, please share.
LT3042 is not the sort of thing that goes into the silent switcher. It contains a hq opamp and a current source to generate a reference voltage. The MHz bandwidth refers to the opamp, not to any switching frequency, since there is none.
You might be able to build something better with an AD797 and discrete components, but I don't think it will be easy.
You might be able to build something better with an AD797 and discrete components, but I don't think it will be easy.
Okay, sorry for not checking more carefully. It might work fine in that case.
Noise does seem to go up in the lower audio frequency range that humans are sensitive to. They stop specifying at 10Hz. Don't know if it matters or not for this application (say, for AVCC).
Probably worth trying it to find out. Of course, presumably one would want to have the ESS circuit available for comparison. Measure and listen both, would be my suggestion. Obviously, if one were considering mass production there would be a lot of things to consider, test, Muntz, etc. For one-off DIY projects, it might make more sense just to do what one knows will work and to the extent practical not mess too much with what is already working. Am I being too lazy?
Noise does seem to go up in the lower audio frequency range that humans are sensitive to. They stop specifying at 10Hz. Don't know if it matters or not for this application (say, for AVCC).
Probably worth trying it to find out. Of course, presumably one would want to have the ESS circuit available for comparison. Measure and listen both, would be my suggestion. Obviously, if one were considering mass production there would be a lot of things to consider, test, Muntz, etc. For one-off DIY projects, it might make more sense just to do what one knows will work and to the extent practical not mess too much with what is already working. Am I being too lazy?
If I seem a bit distracted it is because the ES9038Q2M datasheet arrived today. Lots to think about there. It also means I am no longer free to talk about certain requirements of the chip or any deductions I might be able to make in relation to that. Sorry. I tried to put that off as long as possible, but I think we are at the point where it might be fruitful to investigate what can be done to help with the project given more information about the chip. Without yet having studied exactly what was supplied, my expectation would be along the lines that what I could do would probably be limited to things like closed software and or hardware mods implemented without any explanation of why. In other words, things like one might find in an undocumented commercial product. We will have to see how it turns out as time goes by.
You can still ask questions about things not directly related datasheet topics. We can talk about things that are already public knowledge too. I just have to be careful not say things that are in the datasheet or are derivative of it and not publically known.
Also, I would mention that many of the things we have been talking about up until now are public knowledge and have even been mentioned in this thread before. It is a long thread and there are potentially a lot of details to remember, especially if one is new to some of this stuff. It can take some time for everything to make sense and start to fall into place.
Also, if one is familiar with datasheets for programmable devices what is most commonly seen are electrical specifications and register specifications. You could download an AK DAC datasheet if you would like to see what such a thing looks like.
Other than talking about AVCC regulation we really haven't been getting very much into a lot of electrical specifics. And for that, we have been talking in relation to information made public by downloads on the ESStech website.
My concern is mostly in relation to what I can and can't say about new things I would like to try.
Also, I would mention that many of the things we have been talking about up until now are public knowledge and have even been mentioned in this thread before. It is a long thread and there are potentially a lot of details to remember, especially if one is new to some of this stuff. It can take some time for everything to make sense and start to fall into place.
Also, if one is familiar with datasheets for programmable devices what is most commonly seen are electrical specifications and register specifications. You could download an AK DAC datasheet if you would like to see what such a thing looks like.
Other than talking about AVCC regulation we really haven't been getting very much into a lot of electrical specifics. And for that, we have been talking in relation to information made public by downloads on the ESStech website.
My concern is mostly in relation to what I can and can't say about new things I would like to try.
Hi Markw4,
I own a blue pcb version that works, sound flat (as user sbf2 said on post 162), and come without 1.2v regulator from factory, same with a different blue version of a friend, looking to the chinese seller pictures seems that v1.07 is build in the same way, that is why I´ve asked, looks odd...
I own a blue pcb version that works, sound flat (as user sbf2 said on post 162), and come without 1.2v regulator from factory, same with a different blue version of a friend, looking to the chinese seller pictures seems that v1.07 is build in the same way, that is why I´ve asked, looks odd...
