Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Digital Line Level DACs, Digital Crossovers, Equalizers, etc.

ES9038Q2M Board
ES9038Q2M Board
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 13th November 2018, 06:10 PM   #3111
Markw4 is offline Markw4  United States
diyAudio Member
 
Join Date: Jul 2016
Location: California
Did an interesting experiment recently. Don't have measurements yet though, its on my list to do pretty soon. One issue I have been trying to look into is why the 2nd modded dac has much more H3 than the 1st modded dac. On the face of it, there didn't seem an obvious reason. Used the same AK4137, same external power supplies, etc. All looked to be the same.

Since the 2nd modded dac has sockets for the AVCC opamp and the output stage opamps, I looked around and found an OPA1612 on an SMD to DIP adapter. So I tried swapping it in for one of the output stage opamps. Tried for I/V and for differential summing. In both cases, it didn't sound bright at all like LME49720, but it didn't sound right either. It sounded like the dynamics were compressed quite a bit. So, why did the two different types of opamps sound so different from one another, and neither sounded quite right?

Decided to try an experiment with the +-15 power rails. Thought the might be the most salient commonality that potentially could affect the opamps differently. I gathered up all the big film caps I have, and soldered them together in two groups. There is nominally about 110uf of total capacitance in each group of film caps. Connected one group between +15 and ground, and the other group similarly connected to the negative rail. This is in addition to several thousands of uf of electrolytic filter caps per rail, and pairs of tantaum and ceramic caps at each opamp power pin. Pictures attached below.

As I said above, I haven't measured anything yet, and haven't even finished subjective experiments, but there is a change for the better in sound quality with the LME49720 opamps. Haven't tested with OPA1612 yet. The excessive brightness is less now, maybe by about half, and it sounds smoother (less distorted, I think). I expect when I do measure I will likely find I don't need -150 counts of H3 compensation to null H3 to a minimum, meaning odd harmonic distortion is probably reduced.

Obviously, I have some more work to do on this, but thought it might be of interest to thread participants without having to wait longer for the final conclusions.

EDIT: Forgot to mention, for this experiment I replaced the LME49720 opamps with LME49860 opamps. That is because the latter opamps are on SMD adapters which are easier to swap in and out without bending pins. However, the types of opamps in this case are exactly the same except for maximum voltage ratings. They sound the same.
Attached Images
File Type: jpg FilmCapBank.jpg (451.6 KB, 122 views)
File Type: jpg FilmCapBank2.jpg (673.5 KB, 122 views)

Last edited by Markw4; 13th November 2018 at 06:19 PM.
  Reply With Quote
Old 13th November 2018, 10:03 PM   #3112
occip is offline occip  France
diyAudio Member
 
Join Date: Mar 2018
Quote:
Originally Posted by Markw4 View Post
@eslei,


Back here occip said he did an AVCC mod: ES9038Q2M Board Not sure if that was a board that got the HD setting above. Some more here: ES9038Q2M Board

Maybe occip can clarify for us a little what mods he did, which of them helped the most, and what he thinks about the sound quality at the end of the modding.
Hi
My first board was a green board
First Mod : LT3045 after +12V switched supply : Bad high harmonic (- 90db H3) + noise
Second Mod: add 1800uf on avcc pins: less harmonics + same noise + more bass
Third Mod: added lme49720 for avcc supply : less harmonics + less noise
Fourth Mod: linear supply +12v with lm317 + big capacitors: same harmonics + lower noise on low frequency ( 50Hz , 100hz ...)
Fifth Mod: H3 register set to F800 : less harmonics on H3 ( -96db)
Sixth Mod : OPA swap for Muses 8920D for audio path ( stock voltage mod)
lower harmonic ( only on 1 channel ...) + good feeling on hearing

For my ears , the first mod is to have a linear supply with a lot of capacitors ( at least 10000uf) then add 1800Uf on avcc pins. you will have a very cost effective solution, other mods will cost much more without substantial gain.

Tweaking H3 is good for benchmark but not really hearable, if your volume level is less than -5 dbfs.

But with 99$ you could have higher sound quality
I discovered the Tone board feeded by usb cable ( on board xmos + low jitter oscillators ) and with a very low harmonics and noise for 99$ . I was very skeptical at first but after measurement I found that the guys at Khadas have engineered a exceptionnal piece of hardware !Take a look at audio science review. For me this board has the best price/sound quality ratio and the XMOS software is tweakable ( I'm running my own firmware on it now
Please note that the H3 reg is near 0 on the tone board ( current mode buffer is used);

I will be very happy to see RMAA measurement in order to compare to few values I posted on this thread before.

It would be good to measure the improvement of all the mods showed on this thread, less talks more numbers an Asus xonar sound card is a good start to measure with RMAA

Occip

Last edited by occip; 13th November 2018 at 10:05 PM.
  Reply With Quote
Old 13th November 2018, 11:07 PM   #3113
Markw4 is offline Markw4  United States
diyAudio Member
 
Join Date: Jul 2016
Location: California
Regarding numbers, I don't have Asus xonar, and the only test I can do at the moment is HD @1kHz. That is absolutely insufficient to characterize a dac, as anyone can see looking at the set of measurements performed on dacs at Stereophile. Even at that, Stereophile doesn't measure everything that some people can hear, at least according to Scott Wurcer's friend, Martin Mallinson, VP of engineering at ESS. (Who is also the guy with all the patents.)

So, like I have said before, if anyone has all the test equipment, is in the local area here, and would like to measure my dacs, then sure, please do. Be very interesting to see. So far, I haven't gotten any volunteers.

EDIT: Doesn't surprise me that tweaking H3 with a voltage mode output stage doesn't help sound quality much, if any. Just demonstrates that numbers are not the whole story, particularly static HD tests for a dac.

Last edited by Markw4; 13th November 2018 at 11:11 PM.
  Reply With Quote
Old Yesterday, 10:15 PM   #3114
Markw4 is offline Markw4  United States
diyAudio Member
 
Join Date: Jul 2016
Location: California
Tried a couple of things today. Put three OPA1612 opamps in the output stage. The differences between the way they sound relative to the sound of LME49720 (or LME49860) are much reduced with the film caps still connected up to the power rails (~110uf per rail). In theory, these particular different opamp types (OPA1612 vs LME49720) shouldn't sound any different from one another in an output stage like this. That tells me the +-15v power probably still isn't good enough, or as good as I require it to be. Of course, there could be some other issue making them sound different such as how they are loaded by the output stage passive components, but I'm not done checking to see of there are any more power related issues before considering other possibilities.

Thinking some more about power issues, maybe I should have gone with LDOs for the final 15v regulators rather than lower performance linear regulators and lots of filter caps. Thing is if I were to use LDOs for +-15v, I would need to put them on the output stage board (close to the loads) and remove excess other caps that would degrade LDO performance. It would still be appropriate to take steps to keep LDOs as low noise at low frequencies as reasonably possible.

Unfortunately, we are not using multi-layer PCBs with power planes. The distributed capacitance and low series impedance in that case would probably ideal for the most high performance output stage.

Maybe at some point I will try an LDO approach. I can turn up the voltage on my external +-15 supply to give LDOs a little regulation headroom. That's the thing about working this way performing experiments instead of using a room full of test equipment. Sometimes it would be really nice to have the test equipment. Can't justify the cost though for as infrequently as it would be used. Instead I have to do the manual labor to try different things. Of course, if we had a few more people at the stage of modding where I am at then we could better divide and conquer some of these issues. In the meantime think I will try a little more good quality filtering and see what happens with that.

More or less around the same time I was trying opamps, I was also experimenting with DPLL bandwidth. With gold flashed connectors on the I2S lines, that helps a lot to get one confounder considerably attenuated. At one point found that strategically placing a finger on the SMD filter caps at each side of the dac chip allowed me to get DPLL bandwidth much lower without audible evidence of instability. However, as I have seen before, when I came back a little later and tried putting my fingers in the same places I had much less luck with it. My takeaway is that I decided to clean up the digital power going into the dac chip, so it will get its own dedicated regulator and maybe some upgraded caps for the 1.8v internal regulator. I know that issue came up for discussion before, but I don't think we had a clear idea of what we expected messing with it to fix. I'm still not sure I can fix anything by working on that area, but at least I know that I am trying to help stabilize minimal DPLL bandwidth settings. If I can affect it for some improvement then great, I will know why a fix is needed there. If no change in symptoms, I still won't know if there is any reason to do work on those power supply areas. Either way, it's okay. I do it all for you guys, as I don't need a better dac myself. But, I do see a need for people to have good dacs which unfortunately remain expensive and hard to make, so here we are trying to remedy that problem a little as best we can. Also, maybe we can contribute a little to knowledge of what affects what with these Sabre dac design choices. Always good to have a better understanding of that type of thing.

Last edited by Markw4; Yesterday at 10:35 PM.
  Reply With Quote
Old Today, 12:00 AM   #3115
redjr is offline redjr  United States
diyAudio Member
 
Join Date: May 2011
Location: Danbury, CT
Quote:
Originally Posted by Terry Demol View Post
Rick,

Are you converting everything to DSD in 4137 or playing in whatever the
source format is?

Thanks,

T
T - Are you referring to the TerraBerry DAC? If so, at the moment I've got it set to up-convert incoming signals to DSD. I have roon set to 'DSD over PCM' (DoP) for playback strategy. I've just started on the other ES9038Q2M project (this thread topic) that will include an AK4137 reclocker board too.

Rick
__________________
redjr

Last edited by redjr; Today at 12:19 AM.
  Reply With Quote
Old Today, 01:52 AM   #3116
Mikett is online now Mikett  Canada
diyAudio Member
 
Join Date: Aug 2003
Location: Toronto
Quote:
Originally Posted by Markw4 View Post
Tried a couple of things today. Put three OPA1612 opamps in the output stage. The differences between the way they sound relative to the sound of LME49720 (or LME49860) are much reduced with the film caps still connected up to the power rails (~110uf per rail). In theory, these particular different opamp types (OPA1612 vs LME49720) shouldn't sound any different from one another in an output stage like this. That tells me the +-15v power probably still isn't good enough, or as good as I require it to be. Of course, there could be some other issue making them sound different such as how they are loaded by the output stage passive components, but I'm not done checking to see of there are any more power related issues before considering other possibilities.

Thinking some more about power issues, maybe I should have gone with LDOs for the final 15v regulators rather than lower performance linear regulators and lots of filter caps. Thing is if I were to use LDOs for +-15v, I would need to put them on the output stage board (close to the loads) and remove excess other caps that would degrade LDO performance. It would still be appropriate to take steps to keep LDOs as low noise at low frequencies as reasonably possible.

Unfortunately, we are not using multi-layer PCBs with power planes. The distributed capacitance and low series impedance in that case would probably ideal for the most high performance output stage.

Maybe at some point I will try an LDO approach. I can turn up the voltage on my external +-15 supply to give LDOs a little regulation headroom. That's the thing about working this way performing experiments instead of using a room full of test equipment. Sometimes it would be really nice to have the test equipment. Can't justify the cost though for as infrequently as it would be used. Instead I have to do the manual labor to try different things. Of course, if we had a few more people at the stage of modding where I am at then we could better divide and conquer some of these issues. In the meantime think I will try a little more good quality filtering and see what happens with that.

More or less around the same time I was trying opamps, I was also experimenting with DPLL bandwidth. With gold flashed connectors on the I2S lines, that helps a lot to get one confounder considerably attenuated. At one point found that strategically placing a finger on the SMD filter caps at each side of the dac chip allowed me to get DPLL bandwidth much lower without audible evidence of instability. However, as I have seen before, when I came back a little later and tried putting my fingers in the same places I had much less luck with it. My takeaway is that I decided to clean up the digital power going into the dac chip, so it will get its own dedicated regulator and maybe some upgraded caps for the 1.8v internal regulator. I know that issue came up for discussion before, but I don't think we had a clear idea of what we expected messing with it to fix. I'm still not sure I can fix anything by working on that area, but at least I know that I am trying to help stabilize minimal DPLL bandwidth settings. If I can affect it for some improvement then great, I will know why a fix is needed there. If no change in symptoms, I still won't know if there is any reason to do work on those power supply areas. Either way, it's okay. I do it all for you guys, as I don't need a better dac myself. But, I do see a need for people to have good dacs which unfortunately remain expensive and hard to make, so here we are trying to remedy that problem a little as best we can. Also, maybe we can contribute a little to knowledge of what affects what with these Sabre dac design choices. Always good to have a better understanding of that type of thing.
Well I finally got my Super Regulator built up and I am using the PCM1794 as a test vehicle before installing it on the 9028PRO.
Why did I wait so long to do this. WOW! I cannot believe the difference it makes. Now as a repeat preface on the 1794 DAC, the +- 15V feeds both the IV as well as an LDO to supply the DAC with 5V for AVCC. What appears to have happened is that this superfast with outrageous low impedance power supply really helps the LDO. So could I separate the Super Reg from the AVCC? No. But man, the way it breathes power into this simple board and the depth and image is way beyond what I had expected. There is simply NO comparison in this case to using a 317/337 or even a Sulzer power supply driven by 5534s. I was using an AD817 on the Super Reg, then I switched to the LME49710s thinking that they're both approximately the same bandwidth but so about the same/ Not so the LME49710 indeed sound a lot more focused and the depth increased. The biggest difference is skin tight drums. Guitar strumming is no longer a blur. That the Super Reg will have a place in the 9028 Pro.....yes definitely. Next, I have to acquire a low voltage OP amp to use in a super reg as well to drive the 9028PRO AVCC. I finally found it. Jan Didden suggested an AD8031 that he uses 80Mhz. Nice. I have a spare super reg to generate the 3.3V and I am thinking of placing the LT6655 right in there for the reference voltage. Remember the super reg, uses the regulated voltage to power itself and the reference. Kind of illogical but it works and is very high performance.
The other thing I am also noticing is that certain op amps are more sensitive to the power supply to which it is attached. The AD797 needs a good power supply like the Super Reg. With that, I have no reservations about its performance now. With the 317/337 and the Sulzer you sensed something was lacking in imaging and energy. That is no longer the case.

So in my view a lot of things could make the sound of the DAC change and some of them have no "real explanation" currently though of course it exists but we don't know why. Any ideas why all DACs from different brands sound different? It becomes pretty obvious....and frightening that different production batches can easily sound different.

Kind of fearful that this 1794 the way it is now actually might sound better than my 9028pro at this stage. Oh well..we'll see when I start putting on the regs and clocks etc.
  Reply With Quote
Old Today, 10:58 AM   #3117
janos_904 is offline janos_904  Hungary
diyAudio Member
 
Join Date: Mar 2009
I did suggested to use Jung-Didden superregs approx a half year ago here. You can hear a massive improvement when powering the output stage at +-15 V. Also, Salas regulators (BiB, SSLV1.3, Reflector) are very good to power the clock and the digital patrs. Salas regs are far better than any of 1763 LDO based stuff what you are using now.

https://linearaudio.nl/superregs
Check the bottom of the page: Updated for version 2.3

Quote:
Originally Posted by Markw4 View Post
Thinking some more about power issues, maybe I should have gone with LDOs for the final 15v regulators rather than lower performance linear regulators and lots of filter caps. Thing is if I were to use LDOs for +-15v, I would need to put them on the output stage board (close to the loads) and remove excess other caps that would degrade LDO performance. It would still be appropriate to take steps to keep LDOs as low noise at low frequencies as reasonably possible.

Last edited by janos_904; Today at 11:03 AM.
  Reply With Quote
Old Today, 12:03 PM   #3118
Markw4 is offline Markw4  United States
diyAudio Member
 
Join Date: Jul 2016
Location: California
Quote:
Originally Posted by janos_904 View Post
You can hear a massive improvement when powering the output stage at +-15 V.
Thing is, I notice that DAC-3 gets by fine without using superregs. There are no heat sinks, no SCR, nor other stuff needed to make a superreg. Instead there is a SMPS, and looks like some LDOs. One of these days I should probably look at the insides of it close up and make note all the branding codes on the SMD parts. Might be able to figure out a little more of what they decided to use. Of course, I still think a multi-layer board with power planes is probably best. In addition of course to whatever regulators are used. Selection of good filter caps is important too, but I don't see film caps used in DAC-3.

Anyway, I know there has to be more than one way to get sufficient performance, and I am probably pretty close right now. I am going to make a few small changes then check opamp differences in the circuit again. I don't want there to be any difference, or at least keep to a very small minimum. And, it should sound just as good with either type of opamp, even if very slightly different.
  Reply With Quote
Old Today, 02:51 PM   #3119
Mikett is online now Mikett  Canada
diyAudio Member
 
Join Date: Aug 2003
Location: Toronto
Thing is....what if the Super Reg can and does improve over the DAC3 when used in a hobbyist DAC ? Is it possible that this point is within reach.........?
I can assure you that 20+ years ago when this reg was used on preamps...that was the case..it was better than anything commercially available.

The other thing that I noted is that a Super Reg in front of the LDO LT1761-5 appears to help immensely....well not appears but DOES. In the DAC3 they are possibly using a very high performance SMPS.....and if that is used in front of the LDO then voila. We know that a high performance SMPS is good. But Benchmark might have done a great job at that.....hence the IV also benefits as well. Well I am sure they've done that. The influence of the PS cannot be doubted and its interaction with various op amps is at this point...pick and choose based on eventual sound.

If we can take real close look at the items that Benchmark is using on the board and we remove the IV circuits we can pretty much figure out what is left for the AVCC supply. They claim it is buffered......that might provide another hint. I tried but the provided pics was not high res enough. I was not able to make out the marking on the devices.

Last edited by Mikett; Today at 02:56 PM.
  Reply With Quote

Reply


ES9038Q2M BoardHide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
My Cambridge 500SE, needs an output board (to replace internal DAC Board) slbender Digital Source 1 11th July 2018 03:43 AM
JLSounds XMOS USB board + AK4396 board + Direct DSD PCB ravid Swap Meet 1 29th December 2016 12:36 PM
Unknown Board DC12V TPA3116 amplifier board DIY L163-9 AUS Ben Class D 0 8th December 2016 05:05 AM
FS: Ian FIFO i2s-to-PCM board, Single XO clock board trung224 Swap Meet 3 26th November 2016 03:18 PM
Lost 1 Zv4 ckt board..looking to buy 1 circuit board Blues Pass Labs 1 22nd March 2004 11:09 PM


New To Site? Need Help?

All times are GMT. The time now is 02:58 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio
Wiki