Okay, some new stuff to report with DPLL bandwidth. By very carefully positioning some ferrites in the area sort of between the dac chip and the I2S connector, I can get DPLL bandwidth stable down to 3 out of 10. If with the ferrites in the same state, I change DPLL bandwidth back to 5 out of 10, the brightness and distortion is mostly gone. In fact, I can hear reverb tails on vocals and instruments easier than I can with DAC-3. In some ways it still isn't quite as refined as DAC-3, but maybe in that very particular way it is even better. I would guess it is just the effect of extremely low jitter.
Where I found an previously unknown problem is that we have a new desktop PC in the listening area that replaced the old desktop in the same location (the DAW) and it is the one that usually drives DAC-3. The modded dac is driven from a laptop. Turns out that new desktop PC is very near the open top of the steel case where the modded dac is sitting. The new desktop also has a big clear window in the side to one can see inside, but it also apparently radiates out more noise. Therefore last night when I turned it on to compare the modded dac with DAC-3 I was disappointed that I wasn't closer, and that the modded dac was still too distorted. HaHa, it was the noise it was picking up from the desktop. I will leave that desktop computer off for now and consider the modded dac shielding problem as something to leave for down the road. Right now I need to better understand, if nothing else, what kind of ferrite material is best and how to find position it, and keep its function stable. If I can do that I might almost have a world class dac at some level. Maybe not Stereophile A+, but maybe Stereophile something. Wow!
I haven't even looked at harmonic distortion compensation yet either. Just have to figure out more about how the ferrites for final jitter control works and how to stabilize it long term and I will be happy.
Where I found an previously unknown problem is that we have a new desktop PC in the listening area that replaced the old desktop in the same location (the DAW) and it is the one that usually drives DAC-3. The modded dac is driven from a laptop. Turns out that new desktop PC is very near the open top of the steel case where the modded dac is sitting. The new desktop also has a big clear window in the side to one can see inside, but it also apparently radiates out more noise. Therefore last night when I turned it on to compare the modded dac with DAC-3 I was disappointed that I wasn't closer, and that the modded dac was still too distorted. HaHa, it was the noise it was picking up from the desktop. I will leave that desktop computer off for now and consider the modded dac shielding problem as something to leave for down the road. Right now I need to better understand, if nothing else, what kind of ferrite material is best and how to find position it, and keep its function stable. If I can do that I might almost have a world class dac at some level. Maybe not Stereophile A+, but maybe Stereophile something. Wow!
I haven't even looked at harmonic distortion compensation yet either. Just have to figure out more about how the ferrites for final jitter control works and how to stabilize it long term and I will be happy.
Last edited:
Thank you Mark,
makes me curious about DAC output filter results...how do you think about ferrite bread at ams1117 input?
@Eziitis,
looking at your proposal (thanks for it) it comes close to my idea, to place a ferrite bread between output cap (1uF) and filter caps of clock line to avoid too high output cap for LDO and get a steeper filter though, correct? How important is RDC of the bread here?
makes me curious about DAC output filter results...how do you think about ferrite bread at ams1117 input?
@Eziitis,
looking at your proposal (thanks for it) it comes close to my idea, to place a ferrite bread between output cap (1uF) and filter caps of clock line to avoid too high output cap for LDO and get a steeper filter though, correct? How important is RDC of the bread here?
Markw4,
I would highly recommend shielded coax wires and ufl connectors. I had a lot of trouble with unshielded I2S wires in the past, even channel switching or cracks, same length as you use. With shielded coax wires or at least with a grounded copper film around, the topic was solved. Same with a clock wire for MCLK signal.
I would highly recommend shielded coax wires and ufl connectors. I had a lot of trouble with unshielded I2S wires in the past, even channel switching or cracks, same length as you use. With shielded coax wires or at least with a grounded copper film around, the topic was solved. Same with a clock wire for MCLK signal.
Thanks for all your excellent work and shared posts, Mark!
I am part of the many silence ones that read your thread everyday, but who are happy (for now) with their music / DAC and hence not building anything. In my case all priorities have been anyway shifted... since a new born has arrived.
However, despite having nothing to share or any feedback to provide - and I know you are looking for them, so sorry for that! -, I thought I might just post to let you know that as en engineer I appreciate very much all your efforts, your constant progresses and even the DAC comparisons you are carrying out. All this is indeed very interesting to read, on a daily basis!
Please keep us posted and all the best for all your work
Claude
I am part of the many silence ones that read your thread everyday, but who are happy (for now) with their music / DAC and hence not building anything. In my case all priorities have been anyway shifted... since a new born has arrived.
However, despite having nothing to share or any feedback to provide - and I know you are looking for them, so sorry for that! -, I thought I might just post to let you know that as en engineer I appreciate very much all your efforts, your constant progresses and even the DAC comparisons you are carrying out. All this is indeed very interesting to read, on a daily basis!
Please keep us posted and all the best for all your work
Claude
I haven't tried either one of those things so far. Right now I I'm not so sure much effort needs to be put into dac output filtering before I/V. There would probably be trade offs there. One, two, or three caps without any resistors or inductors might be the least harmful, or might not help. Adding any resistance is likely to increase distortion though a different mechanism than RF induced distortion. We don't necessarily want to increase one source of distortion to reduce a different source of distortion. Have to be careful about that.
Truly, right now if I could just figure out exactly what is going on with the ferrite experiments on the I2S side of the dac and get the effects highly stabilized, I could call this dac done and it could be a great dac in its own right. No need to bother with dac output filter in that case, the dac doesn't sound too bright or too distorted. Makes me think that isn't so much of a problem. Or, maybe I would come back to it later, but the I2S side ferrite thing is much more important to work on. Makes a big difference.
Regarding your question about a ferrite bead for AMS1117, I am only using my AMS regulator for the MCU at this time. Don't think it needs any mods to it for that use. However, if someone wanted to keep using AMS1117 to power more things then maybe a bead would help. Can't say that it wouldn't. Try it and see is about all you can do. Also, exactly what it would do would depend on the type of ferrite material you happen to have. Some types are better for noise suppression than others.
Last edited:
Yeah, okay. Right now I seem to be okay with gold connectors and ribbon cable with every other conductor a ground. Takes some mods to the boards to add ground pin, and maybe swap out pin headers for gold ones. But, I agree with you that ufl should work well. The cables should just be routed together so as not to create ground loops.
Hi Claude,
Good of you to post and let us know of your interest in the project and its continuing progress. Now that the baby has arrived, maybe you could convince your wife to work and let you stay home and take care of the baby. That way you could work on modding a dac while the baby is napping.
But, even if that wouldn't work out for you, we are happy to have you here reading along and taking an interest in what we are doing. Best of luck, and wishes of great happiness for you and your growing family.
Cheers,
Mark
New chinese ES9038Q2M board
Hi, have any of you looked at this new design chinese ES9038Q2M board?
I could not find it anywhere in this long thread.
Lusya ES9038 Q2M I2S DSD Decoder Coaxial Fiber input DAC decoding board For hifi amplifier audio A1 011-in Digital-to-Analog Converter from Consumer Electronics on Aliexpress.com | Alibaba Group
ES9038Q2M ES9038 Q2M I2S DSD Optical Coaxial Input Decoder DAC Headphone Output HiFi Audio Amplifier Board With OLED Display-in Digital-to-Analog Converter from Consumer Electronics on Aliexpress.com | Alibaba Group
Hi, have any of you looked at this new design chinese ES9038Q2M board?
I could not find it anywhere in this long thread.
Lusya ES9038 Q2M I2S DSD Decoder Coaxial Fiber input DAC decoding board For hifi amplifier audio A1 011-in Digital-to-Analog Converter from Consumer Electronics on Aliexpress.com | Alibaba Group
ES9038Q2M ES9038 Q2M I2S DSD Optical Coaxial Input Decoder DAC Headphone Output HiFi Audio Amplifier Board With OLED Display-in Digital-to-Analog Converter from Consumer Electronics on Aliexpress.com | Alibaba Group
Attachments
Yes. I ordered one to try it. The report was brief. It was junk for a number of reasons. It has a SMPS on it to make 5v, and it sounded awful. But that isn't the only problem. There is no volume control, and no way to select filters. No SPDIF or TOSLINK inputs. AVCC power is dirtier than for most other boards. Very poor layout of output stage with no ground plane under it, and ground for output stage only returns to dac chip through SMPS area of board, making for a very long and noisy loop, etc. Bottom line is: not recommended.
Thanks for asking though. We try to keep up with new boards in case a good one happens to come along.
volume control etc requires a dedicated board with LCD, available for 10EUR more. and apparently, there are both SPDIF and TOSLINK inputs ... at least connectors for them. power and output are substandard of course, but please show me a board for 40EUR which has them?
Sorry, you are right about the input. There is one button on the board to cycle through the inputs. Regarding the price, I think I always paid around $39 for a Chinese dac board for my own use, but I didn't care if it could be upgraded to a display. If this one is $10 cheaper I still would not recommend it, because of worse sound quality and lack of features, and because I don't think it is as suitable for modding as the ones we use. But, just because I don't recommend it doesn't mean you can't recommend it if you want.
Hi eziitis,
Understood. I think I would say there are two main reasons I would recommend the boards we usually use. (1) As strange as it may seem, IMHO, they have the best designed layout, and the best ground plane (which I rely on a lot). (2) All the instructions and pictures we have for how to mod one use that layout, and the ones with displays can be apparently be register tweaked without pin lifting or tiny trace soldering (things people don't think they can do).
While I think the boards sound awful in the stock condition, and while modding one is a whole lot of work, there are still some people interested in modding a dac, and other people just interested to read about it. And for me, there are still puzzles to understood. So, here we are. Then once in a while someone points out an uncommon Chinese Q2M dac and asks if we know about it. So far, the answer has been yes, or if not then I try to get my hands on one to check out. Again, so far, nothing better to recommend than the ones we currently use, IMHO.
Understood. I think I would say there are two main reasons I would recommend the boards we usually use. (1) As strange as it may seem, IMHO, they have the best designed layout, and the best ground plane (which I rely on a lot). (2) All the instructions and pictures we have for how to mod one use that layout, and the ones with displays can be apparently be register tweaked without pin lifting or tiny trace soldering (things people don't think they can do).
While I think the boards sound awful in the stock condition, and while modding one is a whole lot of work, there are still some people interested in modding a dac, and other people just interested to read about it. And for me, there are still puzzles to understood. So, here we are. Then once in a while someone points out an uncommon Chinese Q2M dac and asks if we know about it. So far, the answer has been yes, or if not then I try to get my hands on one to check out. Again, so far, nothing better to recommend than the ones we currently use, IMHO.
yes, right, I also have the one discussed here moded to no limits. but this is definitely the last Chinese board I bought. if one wants to push it to the limits you are pursuing, time, money and all the effort will be better spent building the DAC from scratch. PCB manufacturing is pretty cheap nowadays.
It is interesting in that I have thought the same as you say, and at multiple points along the way.
However, there is a problem (at least one). I keep finding out new things that I don't fully understand. If I don't understand them I can't design a board to make sure they are optimal. Also, there are aspects to how the current version works that depends on layout. Developing a board for a very high performance dac (which this is getting to be) would probably take more than one iteration. A well equipped measurement lab would be appropriate too. It would be helpful right now for some of this RF stuff. I wish I could measure jitter coming out of AK4137. I wish I could measure what my finger, copper foil, and or a bag of ferrites is doing exactly and make sure a board could provide a level of sound quality at least as good as the best I can get during experiments here.
Anyway, I have said I am not going to do a PCB. Somebody else can do it, but I would be willing to provide whatever help and support might be needed to get it working well. As far as intellectual property goes, everything here in the thread is public domain. Anyone can make a PCB or whole dacs and sell them if they want. You want to quit your job and work for yourself, my friend? Please do, in fact, please be my guest.
Last edited:
hi, there,
i confirm this driver is working, just change the clock from 100mhz to 49.152mhz.
GitHub - VinnyLorrin/ES9038Q2M-Linux_Driver
via i2c control hardware volume, play pcm from 8000 all the way to 352800, dsd 64, 128, 256 through i2s.
also another volumio plugin could work on q2m configurations, that will lead to the full potential of the q2m dac.
i confirm this driver is working, just change the clock from 100mhz to 49.152mhz.
GitHub - VinnyLorrin/ES9038Q2M-Linux_Driver
via i2c control hardware volume, play pcm from 8000 all the way to 352800, dsd 64, 128, 256 through i2s.
also another volumio plugin could work on q2m configurations, that will lead to the full potential of the q2m dac.
Attachments
Last edited:
Hello,
I am now working on the SMD mod. for my 2nd DAC board (V1.07). I de-soldered the onboard opamp components of voltage mode to modify it to the new differential outputstage for current mode. I promised to post a picture of it, so here it is. I have added the connection points in the schematics where I split up the onboard differential stage to the below-board iv-stage (same as in my post #1289).
Clock is also changed already to Crystek and LT6655 Vref is placed on a small breakoutboard which gave me some headage to connect the 8 pin chip to a 10 bin breakoutboard without pin interconnections... happy to not have frieded my LT6655 after all this heat-cycles...poor small bug.. ;-)
Next will be I/V stage and 3,3V regs for clock and VCCA.
By the way, should I better take a seperate 3,3V reg for SPDIF amp-stage? - it is currently running with the 3,3V AMS1117 reg. rail for MCU and DVDD.
Happy modding!
I am now working on the SMD mod. for my 2nd DAC board (V1.07). I de-soldered the onboard opamp components of voltage mode to modify it to the new differential outputstage for current mode. I promised to post a picture of it, so here it is. I have added the connection points in the schematics where I split up the onboard differential stage to the below-board iv-stage (same as in my post #1289).
Clock is also changed already to Crystek and LT6655 Vref is placed on a small breakoutboard which gave me some headage to connect the 8 pin chip to a 10 bin breakoutboard without pin interconnections... happy to not have frieded my LT6655 after all this heat-cycles...poor small bug.. ;-)
Next will be I/V stage and 3,3V regs for clock and VCCA.
By the way, should I better take a seperate 3,3V reg for SPDIF amp-stage? - it is currently running with the 3,3V AMS1117 reg. rail for MCU and DVDD.
Happy modding!
Attachments
Last edited:
Hi freezebox,
Thank you for the picture with nice labeling!
Regarding the question of how to best power the SPDIF and TOSLINK receiver electronics on the dac board, my guess would be that the stock AMS1117 power is okay. The way that Sabre dacs decode SPDIF is designed to be independent of jitter. That being the case, I don't think those inputs need special power regulation.
Also, I haven't used those inputs much since getting the AK4137 boards. The more expensive version has its own SPDIF and TOSLINK inputs that it can upsample to 11.2MHz DSD, so not much need to use those inputs on the dac board.
In addition, during my efforts to make progress with jitter control using I2S I decided to disconnect power to the dac board SPDIF and TOSLINK inputs. That is because it seemed like there might be some interaction between the SPDIF amplifier on the dac chip and the ferrites I am experimenting with. It seemed like noise in the SPDIF amp might have been coupling into what I am working on and thus having some confounding influence on experiments.
I now have adjusted ferrites for the third time to achieve DPLL stability at a setting of 2 out of 10, after having fully removed and replaced them each time. Therefore, the effects are repeatable. But, adjusting the ferrite positioning to effect stability at 2 out of 10 is very critical, maybe sometimes as little as .001" which I am adjusting using an adjustable dial indicator holder to position a plastic stick touching against a sensitive spot on the bag of ferrites. Unfortunately, I can't tell exactly what is being affected.
I did notice that pressing against the 22uf tantalum filter caps at DVCC and VCCA pins on the side of the dac by the SPDIF and TOSLINK inputs has a noticeable effect on DPLL stability. I can sweep through it to some extent by pressing on the caps with my fingernail or a long pointy plastic stick. I am thinking that maybe going to more mechanically stable caps could help remove another confounder.
In one sense the extreme sensitivity in affecting DPLL stability is almost silly at this point. However the sonic advantages of what appears to be extreme jitter reduction are worth trying to understand. I am getting rid of remaining brightness in the output sound and hearing details in cymbals from a CD recording very similar to DAC-3. In other words, it appears I am getting jitter down around where it must be in DAC-3, but doing it by another means that works with 100MHz clock on the dac chip. Very interesting, is all I can say. The question is can I figure out how to get practical control of it and keep it stable enough over time?
From doing experiments here, it is clear that careful control of RF fields in the vicinity of the dac chip is critical to optimal sound quality. Trying to do that with a 2-layer PCB may end up being impractical for everyday use. A 6-layer board would probably start to make a lot more sense from the perspective of seriously trying to make a great dac. But, I am willing to continue to try to see if I can come up with something that diyer's can use with dac the board we are working on. Don't know how it will turn out exactly, but I will try to keep you all posted on developments.
Last edited: