We also brought up and discussed the bypassing of internal filters in the Katana thread.
It's known since a long time that if you're after the last bit of sound improvement
you can't neglect looking at these ON-DAC filters.
It's possible to bypass them on the Sabre. You need a special firmware offering that feature. You need to feed 352/384 material then.
And then you'll also loose this or that feature (VC and mute if I recall it correctly) .
* I bypass the filters on my current Piano2.1 HAT (by running it at 352/384)
And I like it.
* And perhaps Allo one day offers a tweakers firmware for Katana that bypasses the filters too
* Many Soekris users are using custom filters - another way of doing it
If I recall correctly there were also bypass discussions to do the job on the transport.
And perhaps Ian will also look into it for his DACs.
If it's worth the hassle, I don't know.
Looking into a better power supply solution for sure has higher priority.
It's known since a long time that if you're after the last bit of sound improvement
you can't neglect looking at these ON-DAC filters.
It's possible to bypass them on the Sabre. You need a special firmware offering that feature. You need to feed 352/384 material then.
And then you'll also loose this or that feature (VC and mute if I recall it correctly) .
* I bypass the filters on my current Piano2.1 HAT (by running it at 352/384)
And I like it.
* And perhaps Allo one day offers a tweakers firmware for Katana that bypasses the filters too
* Many Soekris users are using custom filters - another way of doing it
If I recall correctly there were also bypass discussions to do the job on the transport.
And perhaps Ian will also look into it for his DACs.
If it's worth the hassle, I don't know.
Looking into a better power supply solution for sure has higher priority.
Last edited:
* I was just confirming what Ian also has found on a 9028. There seems to be a pattern.
* And I am by no means a HW expert. I do have a solid understanding of the stuff that's
been talked about. And I do know how to use a soldering iron and a scope to tweak my own stuff.
* DSD: I would ask myself then why DSD would sound better - if it's converted back to PCM anyhow.
I am not questioning that it sounded better in your setup. But Ian has said already that you didn't
seem to run the DAC in its best possible config for PCM anyhow.
Still your input is valuable. Certainly it's something worth to be looked into. Ian agreed on that too.
I don't think DSD is converted back to PCM exactly. It is never interpolation filtered the same way. Why some might say it is converted to PCM could be because when it is converted back to analog, the modulator is run in multi-bit mode, not single bit, and multi-bit operation at that point is the same as with PCM. However, the oversampling ratio appears to be higher.
No need for spreading opinions. Just look it up. The Sabres are no native DSD DACs.
And I also said "some kind" of PCM. I'm well aware that they use some kind of tricks to cope with it.
And these tricks will still allow digtial VC to work.
Last edited:
I don't want to step on anyone's toes but I feel I should share my disagreement about the 4137 resampler. It was the greatest of hopes but it turned out it was the greatest of disapointments. It sounds spectacualr at first but if you compare it to really good DACs with R2R chips or something like a Dcs it sounds detailed, fluid, spacious but unnatural. The last DAC tested here that used it and was a real technological marvel was the Analog Domain DAC1. At euro 20K not cheap either.
I didn't mean to imply that you misspoke about how they work.
I meant that what they do may be better than native DSD in some ways despite not being pure, and it is enough different from how they do PCM to account for why DSD can sound better than PCM. That's all.
Not sure that the experience with you mention can be completely attributed to the AK4137 itself. The symptoms you describe seem to me like things I have heard before from a dac, but not due to 4137.
Since I don't have a lab full of test equipment (in retirement) what I do is very simple in some ways. The gauge of dac progress I have that is fixed is a Benchmark DAC-3. It is a very accurate as dacs go, although better may come along at some point. Anyway, it is a major piece of my lab equipment. If I get closer to its sound, that is what I will define as the right direction and if the opposite, then the opposite.
With the particular dac I am working on now AK4137 is getting me closer. I know there may be other paths to get closer, but this is the one I am investigating right now and is based on components that are readily available in the market without having to order large quantities or customize a lot of ASIC firmware.
Looks like it can do rather well compared to many dacs on the market, and do so at a fairly low cost. And low cost is a goal in this case. Anyway, let's give it some time and let Ian have a go at it. No need to reach conclusions about how that will turn out at this point. Be interesting to see what he thinks after he has a chance to take a look at it.
I totaly agree. When you have chance to hear hi-end dacs you know how music can be heard. Music is not 1khz 0.0001% thd. Is much more.
How should i described...my experiences...all digital processing, upsampling converting to no original format is similar as you will flow excellent wine, maybe not so perfect clean golden colour trought wine filter. At every process wine is more and more clean and transparent but at the end you drink wine similar taste to water. All taste, sweetnes, nice smell ...is gone.
After some while of listening i feel empty and i get headache because of unatural detailed dry sound with full of sharp impulses...
But everyone has his own taste.
@androa76
I have similar experience.
I always tend to listen to the original format. For example, if the music is 44.1KHz, I'll just stick with it. I will not do any up-sampling to higher Fs. Because no up-sampling process is lossless. The up-sampling process will add or lose something to the music. In either case the music will be changed and no longer what it was. Music is a kind of art, it contains the information that the musician and the recording studio want to show to the audiences. If the style is changed, those information will be changed too. Somebody may say that most of modern DAC has internal up-sampling. It's true.But it's the nature and the style of the DAC that we are using, I'll just take it. BTW, I found the original format always sounds the best on ESS DACs.
Another example is the story sync and async clock mode. One time, I switched my ESS DAC back to async mode (the default mode with 100MHz clock) just want to feel how bad the sound was. But to my surprise, the sound was not that bad, actually it has beautiful high range and deep low range. The whole music seemed to be extended and became vivid. But when I switch back to the sync mode (with 90/98 MHZ clock), I suddenly realized what's the difference. The difference is not something superficial, it's something inside. Sync mode has much better sense on stage, depth, 3D feelings, instruments position focusing and atmosphere. It has nothing to do with frequency range and THD. It's something more related to the phase concept.
Again, I'm not against adding DSP or ASRC before DAC. Though they could make some change to the music, but the flavor they added may also has some improvement or enhancement to the sound quality. However, like or dislike, it will be highly up to personal preference. That's why I said it has to be optional and easy to switch back to original format to decide if you really like the changes. The last off topic story is that, I like the taste of matsutakes(kind of mushroom). So each time when I cook beef soup, I put some of them. My wife said it's not the original taste of beef soup. And I said, it doesn't matter, I just love the new taste. it's delicious.
BTW, I like what you said: “Music is not 1khz 0.0001% thd. Is much more.... ”
Enjoy DIY and feel the music!
Regards,
Ian
I have similar experience.
I always tend to listen to the original format. For example, if the music is 44.1KHz, I'll just stick with it. I will not do any up-sampling to higher Fs. Because no up-sampling process is lossless. The up-sampling process will add or lose something to the music. In either case the music will be changed and no longer what it was. Music is a kind of art, it contains the information that the musician and the recording studio want to show to the audiences. If the style is changed, those information will be changed too. Somebody may say that most of modern DAC has internal up-sampling. It's true.But it's the nature and the style of the DAC that we are using, I'll just take it. BTW, I found the original format always sounds the best on ESS DACs.
Another example is the story sync and async clock mode. One time, I switched my ESS DAC back to async mode (the default mode with 100MHz clock) just want to feel how bad the sound was
. But to my surprise, the sound was not that bad, actually it has beautiful high range and deep low range. The whole music seemed to be extended and became vivid. But when I switch back to the sync mode (with 90/98 MHZ clock), I suddenly realized what's the difference. The difference is not something superficial, it's something inside. Sync mode has much better sense on stage, depth, 3D feelings, instruments position focusing and atmosphere. It has nothing to do with frequency range and THD. It's something more related to the phase concept.Again, I'm not against adding DSP or ASRC before DAC. Though they could make some change to the music, but the flavor they added may also has some improvement or enhancement to the sound quality. However, like or dislike, it will be highly up to personal preference. That's why I said it has to be optional and easy to switch back to original format to decide if you really like the changes. The last off topic story is that, I like the taste of matsutakes(kind of mushroom). So each time when I cook beef soup, I put some of them. My wife said it's not the original taste of beef soup. And I said, it doesn't matter, I just love the new taste. it's delicious
.BTW, I like what you said: “Music is not 1khz 0.0001% thd. Is much more.... ”
Enjoy DIY and feel the music!
Regards,
Ian
@androa76
I have similar experience.
I always tend to listen to the original format. For example, if the music is 44.1KHz, I'll just stick with it. I will not do any up-sampling to higher Fs. Because no up-sampling process is lossless. The up-sampling process will add or lose something to the music. In either case the music will be changed and no longer what it was. Music is a kind of art, it contains the information that the musician and the recording studio want to show to the audiences. If the style is changed, those information will be changed too. Somebody may say that most of modern DAC has internal up-sampling. It's true.But it's the nature and the style of the DAC that we are using, I'll just take it. BTW, I found the original format always sounds the best on ESS DACs.
Another example is the story sync and async clock mode. One time, I switched my ESS DAC back to async mode (the default mode with 100MHz clock) just want to feel how bad the sound was
Again, I'm not against adding DSP or ASRC before DAC. Though they could make some change to the music, but the flavor they added may also has some improvement or enhancement to the sound quality. However, like or dislike, it will be highly up to personal preference. That's why I said it has to be optional and easy to switch back to original format to decide if you really like the changes. The last off topic story is that, I like the taste of matsutakes(kind of mushroom). So each time when I cook beef soup, I put some of them. My wife said it's not the original taste of beef soup. And I said, it doesn't matter, I just love the new taste. it's delicious
BTW, I like what you said: “Music is not 1khz 0.0001% thd. Is much more.... ”
Enjoy DIY and feel the music!
Regards,
Ian
Under the architecture of delta sigma, the analog signal will be very similar to the DSD, so whether you do up-sampling to higher Fs, the DA chip will be up-sampling internally , but if the input signal to the DA chip has done the up-sampling, it will not be done again. This is why DSD signals are easier to process for DA chip. Since up-sampling is inevitable, it depends on whether the external up-sampling is better than the DA chip, but most external up-samplings are not well done, but you cannot said the up-sampling is a bad idea. Otherwise, who is the expensive Vivaldi Upsampler to sell to? Vivaldi Upsampler | dCS
Last edited:
There are many things that affect the sound of a digital device, we have discussed that ad nauseum. My experience (and I am relieved that there are others) comes from a project that a lot of smart people worked on and who WANTED to make it sound good, nothing was spared. But in the end the 4137 has an algorithm that was classified as very good for mid-range devices, at first listen it is quite good and many people will be in love with the sound but not for the top level devices.
I was just hoping that sharing my experience will provide the option of switching it off, nothing more, I am just not interested in any device that relies on it.
I was just hoping that sharing my experience will provide the option of switching it off, nothing more, I am just not interested in any device that relies on it.
Well... We all can debate for days about this but in the end, it is the listener's ears (brain) that decide if she/he likes it or not. Too bad the RPi2-3 cannot output native DSD on I2S and that we have to rely on is DoP which is limited in most cases to 128... Same applies here, if we have a bunch of native DSD material, we need to re-sample to PCM or DoP 64-128 and thus cannot listen to it natively.
I'm 100% ok with Ian's implementation just it would have been great that 1-Pi support native DSD over i2s and 2-ES90XX can process DSD stream natively. That being said, I won't feel bad about it and it'll still sound amazing I'm sure.
Do
I'm 100% ok with Ian's implementation just it would have been great that 1-Pi support native DSD over i2s and 2-ES90XX can process DSD stream natively. That being said, I won't feel bad about it and it'll still sound amazing I'm sure.
Do
Understood AK4137 is not expected to be the path to a high end dac. The question I have been working on is how to make a low cost, entry level dac so more people can access better sound than they can now.
For those wanting the very best sound quality I don't think there is any way around it costing more than a few hundred dollars, which unfortunately leaves out many people from being able to experience a move upward.
Starting from ES9038Q2M and a building with high quality circuitry can only get one so far. At the cost of around $15 for another chip, AK4137 does sound better than not in most respects. Without it there is other distortion that appears to be associated with ES9038Q2M PCM interpolation filtering, which is not top quality for the most discerning listeners. It is possible to to do better there with ES9028PRO or ES9038PRO, the latter for sure having an option to use more taps if limited to stereo operation.
For those wanting the very best sound quality I don't think there is any way around it costing more than a few hundred dollars, which unfortunately leaves out many people from being able to experience a move upward.
Starting from ES9038Q2M and a building with high quality circuitry can only get one so far. At the cost of around $15 for another chip, AK4137 does sound better than not in most respects. Without it there is other distortion that appears to be associated with ES9038Q2M PCM interpolation filtering, which is not top quality for the most discerning listeners. It is possible to to do better there with ES9028PRO or ES9038PRO, the latter for sure having an option to use more taps if limited to stereo operation.
Last edited:
See this thread by Bunpei of SDTrans384 fame:
Applying GaN transistor as a replacement of SiC rectifier
Greg in Mississippi
Ian,
In my testing of your single-ES9038Q2M prototype, I found the Onetics 600:600 transformer to be the best sounding so far, bettering the Lundahl LL1570XL. I HAVE NOT yet tried the Lundahl LL1684's I have here, largely because the quality of the sound with the Onetics over the LL1570XLs. That included better dynamics.
From casual listening, that setup did not seem to have less strong dynamic performance than my 2 Twisted Pear ES90x8Pro setups, both with one of their active output stages.
Shall I send the Onetics up to you for a listen? While I mounted them on a DIY protoboard, they do you the same plug-connection as your custom boards. All you'll need to do is arrange for output connectors.
In the meantime, as I get some audio time freeing up from another pressing project, I'll assemble a board for the LL1684's so I can give them a whirl.
Greg in Mississippi
Actually, I was thinking before about how to run an SRC4392 instead of an AK4317 in synchronized mode, but the same scheme should probably work the same for either chip. The way I was thinking of doing it would not require me to change the existing 100MHz clock, since I can divide it down and use it to clock the SRC. Although it would not be using a standard sample rate it should still work fine. I wrote a short explanation of the idea in the other thread in this post: ES9038Q2M Board - Page 230 - diyAudio
By that means, they could run in master mode (synchronous), or use ASRC, whichever is configured by register programming. Should be doable so long as there is still a separate clock that can be enabled for AK4137. Might use a NB3L553 to buffer the ES9038Q2 master clock output so I could disable that signal and enable another clock for the SRC, if wanting to run in ASRC mode. So, maybe nice to be able to switch it either way in software is all I was thinking. Certainly would make for quick comparisons or perhaps some ability to use synchronous or ASRC depending on what works best for a very high jitter incoming source.
Yes, I have been interested. The thing that puzzles me is that Allo Katana runs in synchronous mode and so far my modded dac using ASRC has sounded as good or better. I do have to tweak the Q2M DPLL bandwidth register to get the best sound quality though. Also, the Benchmark DAC-3 here always sounds better than my modded dac and Katana, and DAC-3 runs with ASRC enabled. It may be doing upsampling of some kind though. However it works it seems to show there is some way to make ASRC give excellent sound quality. Maybe the actual problem is figuring out how they do it.
<SNIP>
I've had a similar experience to those of Ian and Soundcheck on synchronous reclocking of ESS DACs. I've taken 3 different DACs (a slightly earlier ES9023 Mamboberry than the one Soundcheck has, Ian's ES9018K2M DAC prototype, and Ian's ES9028Q2M DAC prototype) and converted them to Sync mode, in each case using an Allo.com Kali FIFO reclocker to provide the appropriate clock signal. In each case, I found the SQ took a step toward a more natural, closer to the source experience. The SQ with the ESS ASRC is a bit more 'technocolor' than real in my experience, though the difference is MUCH smaller in the later chips (ES902x & ES903x versions).
Also, I wouldn't compare the Katana in master-sync mode to the DAC-3 in ASRC mode to generalize about the superiority of a feature in one over the other. As Soundcheck said above "Bottom line. All this crap is much too complex to draw any generic conclusions." There are too many differences between the implementations of those two DACs to compare them and say that a feature of one is better than that of the other. BUT when I've modified a ESS-chip-based DAC from ASRC to Sync with a roughly equivalent clock, I have preferred Sync mode.
Ian's ES9028Q2M DAC with his controller is so far unique for us DIY'ers, in that it can use the 'pure sync' mode available in the later chips which turns off the DPLL. This has been discussed some in the Twisted Pear Buffalo-IIIPro DAC boards main thread and those who have tried it have been very happy with the results... I'll be very happy when the release firmware for their boards enabling that functionality!
Still, the latest ESS chips with their latest filters and processing make the difference between ASRC & Sync modes smaller than with the earlier generations.
Finally, Mark, I haven't looked at the AK4317 datasheet in awhile, but I believe (if I'm not confusing it with another ASRC chip) that it can be run in synchronous mode with a DAC, using a single clock for both, similar to how Ians' FIFO can do the same.
Greg in Mississippi
Sure. Don't know of any reason why that couldn't work. I would still like a dac that can run with ASRC when I want to play music or other sound from a different clock domain. What if I want to watch a movie instead of listen to a recording of a concert? I can't make the DVD player or the computer sync to the dac clock. I need real time, non-delayed sound for some use cases.
So, it would be nice if the sync mode (ASRC or SYNC) could be controlled by software. Most SRC chips allow for two clocks, one for 44.1 and its multiples, and one for the 48kHz set. The clocks are for determining the output sample rate, so one clock could be used for whatever the sweet spot sample rate of the dac is (just convert all inputs to that sample rate if using ASRC mode), and the other clock input could be used as in input for sync mode clocking.
There are various ways to do that, but using the MCLK output from the dac chip to clock the SRC guarantees that setup and hold times should be in the correct phase for what the dac needs. Then it would be possible to change that clock rate using Clock_Gear if desired. However, although the above may be a potentially interesting way to do sync mode (seems to me anyway), it is no doubt not the only way. Ian's existing way might work fine.
However, I don't know if conversion to DSD can be done synchronously. I also don't know if it's even necessary to go to DSD to get better sound quality if certain other things are done such as external interpolation filtering in an ASIC. The thing about DSD with AK4137 is that it is an easy way to get better sound quality at low cost with an off the shelf item. If custom DSP can be used then that would be a whole new ball game to think about.
Last edited:
By the way, some info about how DSD is handled in Sabre dacs here: Introducing the Buffalo III-SE-Pro 9028/9038 - Page 15 - diyAudio
It explains that DSD doesn't go through the PCM oversampling filter, which is probably why it sounds better, IMHO. That's why it would make sense if wanting to use PCM rather than DSD, to use an external higher quality "oversampling filter" which is the same thing as the interpolation filter. It is the filter for which there are 7 or so built-in choices, none of which ever sounds exactly right. Part of making it sound better 'the DAC-3 way' involves clocking an SRC4392 with a 27MHz clock to convert all PCM to 211kHz, then operate the Q2M clock at 30MHz, a frequency that leaves some frequency space for wider transition band for the interpolation filter that is calculated in Spartan 6. It also leaves some room for Q2M to use ASRC to minimize any remaining jitter at that point. So long as the 30MHz clock is ultra low jitter, and Q2M is configured for the tightest DPLL bandwidth, SQ can be excellent even though there is still ASRC.
For the most part, it appears that people complaining about the sound of ASRC are not using smart firmware that tries to minimize DPLL bandwidth as much as possible for PCM, and for DSD (two different settings). It does help, although there appears to possibly be more room for improvement with DSD, given that the default setting is rather loose for that mode.
It explains that DSD doesn't go through the PCM oversampling filter, which is probably why it sounds better, IMHO. That's why it would make sense if wanting to use PCM rather than DSD, to use an external higher quality "oversampling filter" which is the same thing as the interpolation filter. It is the filter for which there are 7 or so built-in choices, none of which ever sounds exactly right. Part of making it sound better 'the DAC-3 way' involves clocking an SRC4392 with a 27MHz clock to convert all PCM to 211kHz, then operate the Q2M clock at 30MHz, a frequency that leaves some frequency space for wider transition band for the interpolation filter that is calculated in Spartan 6. It also leaves some room for Q2M to use ASRC to minimize any remaining jitter at that point. So long as the 30MHz clock is ultra low jitter, and Q2M is configured for the tightest DPLL bandwidth, SQ can be excellent even though there is still ASRC.
For the most part, it appears that people complaining about the sound of ASRC are not using smart firmware that tries to minimize DPLL bandwidth as much as possible for PCM, and for DSD (two different settings). It does help, although there appears to possibly be more room for improvement with DSD, given that the default setting is rather loose for that mode.
I did some ESS DAC HAT comparison test a couple of months ago.
https://www.diyaudio.com/forums/pc-...i2s-dac-hats-raspberry-pi-31.html#post5440848
Based on the actual listening test. I found the ES9038Q2M dynamic performance and quality was not as good as ES9028Q2M. The output impedance of ES9038Q2M is 774 ohm higher than ES9028Q2M's 403ohm. I suspect that is the main reason.
So I designed this ES9038Q2M dual mono DAC HAT. In this configuration, the internal impedance of ES9038Q2M will become 387 ohm, lower than ES9028Q2M.
I'll do a comparison test between signal and dual mono configuration ES9038Q2M DAC HAT very soon to see if there is any improvement.
ES9038Q2MpiDualHAT_2 by Ian, on Flickr
Good weekend.
Ian
Hey Ian,
How does the dual mono 9038q2m compare to 9028q2m ? Any thoughts yet?
Also, curious to know if anyone has considered an I/V using TPA6120 current feedback amp - it is differential input and can handle low impedance loads so that would allow multiple pole passive filtering. Any disadvantages?
@Hugh Jazz;
CFB has higher bandwidth than VFB, so seems suitable for I/V stage. Don't know why few people did so.
I designed a balanced OPA861 I/V stage months ago. I'm very happy with the sound quality. It's transconductance amplifier similar to CFB. May be it's worth to give a try.
Are you interested in trying it by yourself? If so, please let me know if my "hold" board can be some helps.
Regards,
Ian
CFB has higher bandwidth than VFB, so seems suitable for I/V stage. Don't know why few people did so.
I designed a balanced OPA861 I/V stage months ago. I'm very happy with the sound quality. It's transconductance amplifier similar to CFB. May be it's worth to give a try.
Are you interested in trying it by yourself? If so, please let me know if my "hold" board can be some helps.
Regards,
Ian