ES9018K2M, ES9028Q2M, 9038Q2M DSD/I2S DAC HATs for Raspberry Pi

Jump to Latest
O

ofswitched

Banned
Joined 2018
iancanada said:
Hi ofswitched,

That's a good point.

In my dual mono DAC design, two DAC chip share same mclk, I don't think there is any chance having delay in between for different time . Because DAC internal logic are synchronized state machines. If they change state at same rising or falling edge of same clock, there will be no any asynchronous delay in between . But please correct me if I'm wrong.

If each dual mono DAC uses its own local clock, yes, you are right, that will be problem.

Good luck to your project.

Regards,
Ian
Click to expand...

When a signal at the same time into the left and right channel based on different DAC chips, jitter caused by some factors of the two DAC chips individually will no longer be the same time. In fact, even a single chip will have this problem. Not to mention two chips that differ in characteristic.

In dual mono DAC design, it only has a data with good test results, but the human ear does not sound better because the human ear is more sensitive to jitter and less sensitive to distortion.

Different people speak the same sentence, but there are differences with those who originally said it. This is the distortion. However, what it's different from original one is not absolute bad. It's a so-called jitter that the distortion of the sound caused by the DJ intentionally turning the LP. Therefore, anyone will feel that the sound with jitters is bad.
 
Last edited:
D

dutchdiyer

Member
Joined 2017
iancanada said:
Dual mono version ES9038Q2M DAC HAT layout design was done.

Please see the attached PCB layout drawing for details (top side).

Two ES9038Q2M DAC chip will work in dual mono mode. One takes care of left channel, the other for the right channel.

Inside each DAC chip, two cores will be in parallel. The output impedance will be reduced to 387ohm which is not bad. DNR will be increased to 132dB from 129dB theoretically (in differential current mode ).

PCB order placed. I'll have the prototype very soon. I'm looking for some improvement.

Ian
Click to expand...

Hi Ian,

I like your perseverance. I'm really curious what this dual setup will bring.
I see you've also included s/pdif input.
That's nice. It makes the dac more versatile. But maybe you did that before and I missed it.
Keep up the good work.

Regards,
Frank
 
C

Cmdr_DASh

Member
Joined 2006
iancanada

iancanada

Member
Joined 2009
twitter.com/iancanadaTT
Paid Member
ofswitched said:
When a signal at the same time into the left and right channel based on different DAC chips, jitter caused by some factors of the two DAC chips individually will no longer be the same time. In fact, even a single chip will have this problem. Not to mention two chips that differ in characteristic.

In dual mono DAC design, it only has a data with good test results, but the human ear does not sound better because the human ear is more sensitive to jitter and less sensitive to distortion.

Different people speak the same sentence, but there are differences with those who originally said it. This is the distortion. However, what it's different from original one is not absolute bad. It's a so-called jitter that the distortion of the sound caused by the DJ intentionally turning the LP. Therefore, anyone will feel that the sound with jitters is bad.
Click to expand...

Hi Ofswitched,

I understood what you are concerned about. But from what you were saying, they are not particularly related DAC in dual mono or parallel configuration. Basically you are worrying about two things: jitter and skew. The first thing can be improved by introducing good clock through FIFO or running DAC in sync mode. But the second thing is mainly decided by silicon material and processing, though the tolerance of skew can still be reduced in some degree by using good power supply.

However the advantages of paralleling DAC or running them in dual mono configuration to parallel more are very significant.

Actually a lot of top of the line DACs are using the paralleling and dual mono technology to improve performance. For example, PCM1704 was originally designed in mono (single channel) with two DACs internally in parallel. PCM63 too.

http://www.dddac.de/files/PCM1704.pdf
http://www.sowter.co.uk/dacinfo/PCM63.pdf

ESS also recommends mono mode for high-end audio applications.

In my next post, I’ll explain the principle of why sound quality can be improved significantly by paralleling DAC technology.

I’m audiophile myself. I believe in real listening test too. I’ll get my dual mono and single ES9038Q2M DAC HATs compared as soon as I get the prototype. I’ll update with the result.

Regards,
Ian
 
iancanada

iancanada

Member
Joined 2009
twitter.com/iancanadaTT
Paid Member
Principle: Why sound quality can be significantly improved by paralleling DACs

This is a widely used signal processing principle.

For a given current output mode DAC, the output current contains two parts:
Is: output current of signal (RMS)
In: output current of noise (RMS)

Thus, the SNR (in dB) of this DAC can be calculated as:
SNR=10*log( square(Is)/square(In) )
=20*log( Is/In )

If we put two DACs in parallel, the output current will be summed together.

Because the signal is correlated and determined, the new signal output current of two DAC in parallel will be:
Isp=2*Is , (RMS)

While, the noise is uncorrelated and white random, in this case, the noise output current of two DAC in parallel will be:
Inp=In* SquareRoot(2), (RMS)

So, when we re-do the SNR math for two DAC in parallel, the new SNR will be:

SNRp=10*log( square(Isp)/square(Inp) )
=10*log(2* square(Is)/square(In) )
=20*log( Is/In ) + 20*log( SquareRoot(2) )
= SNR + 3dB

Now, the conclusion we got is that, when parallels two current output mode DAC together, the SNR will be improved by 3dB than original the signal DAC.

Does 3dB really matter for a DAC? Yes, it’s a small number but it does matter. Because when we convert 3dB into percentage, it equals to 30%! So, it means, by paralleling two DACs together, the performance will be improved by 30%! The number can still be doubled if we paralleling them once again.

However, in the real world, the number could be slightly less than 3dB due to other limitations, such as jitter and skew and other time related things.

Let’s take ES9038Pro for example:

The DNR of ES9038Pro is 137dB when it’s configured in stereo mode. But when ES9038Q2M is configured into dual mono mode by paralleling both left and right channels together in each DAC chip, the DNR will be boost to 140dB. 140db equals exactly to 137dB + 3dB.

Now we know why sound quality can be significantly improved by paralleling DACs.

Have a good weekend.
Ian
 
O

ofswitched

Banned
Joined 2018
iancanada said:
Hi Ofswitched,

I understood what you are concerned about. But from what you were saying, they are not particularly related DAC in dual mono or parallel configuration. Basically you are worrying about two things: jitter and skew. The first thing can be improved by introducing good clock through FIFO or running DAC in sync mode. But the second thing is mainly decided by silicon material and processing, though the tolerance of skew can still be reduced in some degree by using good power supply.

However the advantages of paralleling DAC or running them in dual mono configuration to parallel more are very significant.

Actually a lot of top of the line DACs are using the paralleling and dual mono technology to improve performance. For example, PCM1704 was originally designed in mono (single channel) with two DACs internally in parallel. PCM63 too.

http://www.dddac.de/files/PCM1704.pdf
http://www.sowter.co.uk/dacinfo/PCM63.pdf

ESS also recommends mono mode for high-end audio applications.

In my next post, I’ll explain the principle of why sound quality can be improved significantly by paralleling DAC technology.

I’m audiophile myself. I believe in real listening test too. I’ll get my dual mono and single ES9038Q2M DAC HATs compared as soon as I get the prototype. I’ll update with the result.

Regards,
Ian
Click to expand...

You still don't really understand what I mean, but many expensive DAC, Wadia, wyred4sound or Northstar, do not use two chip even ESS recommends mono mode for high-end audio applications.

In order to achieve mono mode, you must use the FPGA it will generate a lot of jitter. This is another reason.

PCM1704 or PCM63 is not as good as ess9018 in performance, so if mono is used, performance can be improved, and the additional jitter generated by the mono architecture itself can be ignore, but not for 9018 or 9038.
 
Last edited:
iancanada

iancanada

Member
Joined 2009
twitter.com/iancanadaTT
Paid Member
ofswitched said:
You still don't really understand what I mean, but many expensive DAC, Wadia, wyred4sound or Northstar, do not use two chip even ESS recommends mono mode for high-end audio applications.

In order to achieve mono mode, you must use the FPGA it will generate a lot of jitter. This is another reason.

PCM1704 or PCM63 is not as good as ess9018 in performance, so if mono is used, performance can be improved, and the additional jitter generated by the mono architecture itself can be ignore, but not for 9018 or 9038.
Click to expand...

Hi Ofswiched,

ESS DACs don't need any FPGA to achieve dual mono configuration. And dual mono configuration doesn't introduce any jitter additional to stereo configuration. That's why ESS recommends dual mono for high-end application.

PCM1704 and PCM63 are the best R-2R DAC chip had ever made. Both of them have been discontinued. The only reason is that they were too expensive to make.

Regards,
Ian
 
Last edited:
O

ofswitched

Banned
Joined 2018
iancanada said:
Hi Ofswiched,

ESS DACs don't need any FPGA to achieve dual mono configuration. And dual mono configuration doesn't introduce any jitter additional to stereo configuration. That's why ESS recommends dual mono for high-end application.

PCM1704 and PCM63 are the best R-2R DAC chip had ever made. Both of them have been discontinued. The only reason is that they were too expensive to make.

Regards,
Ian
Click to expand...

The chip is a small cost for assembled DAC cost. If mono mode is better, why are Wadia, wyred4sound or Northstar not used? At the same time, they are not to use very expensive clock chip either. Obviously, the sound is beautiful or not, not because of the test data.

The resistors in the chip can not be high-precision. DAC using the R-2R DAC chip will not be as good as soekris. In contrast, the well-designed 9018 DAC has far better sound quality than soekris. Why?

Some people claim that low-cost R-2R sounds good, because such low-cost R-2R makes the sound obscure, so that in fact the ugly looks like the beauty.
 
Last edited:
O

ofswitched

Banned
Joined 2018
iancanada said:
Hi Ofswiched,

ESS DACs don't need any FPGA to achieve dual mono configuration. And dual mono configuration doesn't introduce any jitter additional to stereo configuration. That's why ESS recommends dual mono for high-end application.

PCM1704 and PCM63 are the best R-2R DAC chip had ever made. Both of them have been discontinued. The only reason is that they were too expensive to make.

Regards,
Ian
Click to expand...

YGGDRASIL DAC is one of R-2R chip DAC. Maybe you can try to design a R-2R DAC with Analog Devices AD5791BRUZ IC?
 
M

matt_garman

Member
Joined 2008
Paid Member
ofswitched said:
DAC using the R-2R DAC chip will not be as good as soekris. In contrast, the well-designed 9018 DAC has far better sound quality than soekris. Why?
Click to expand...

Green is a far better color than blue. Why?


ofswitched said:
Some people claim that low-cost R-2R sounds good, because such low-cost R-2R makes the sound obscure, so that in fact the ugly looks like the beauty.
Click to expand...

Where can you find a low cost r2r dac chip?


ofswitched said:
YGGDRASIL DAC is one of R-2R chip DAC. Maybe you can try to design a R-2R DAC with Analog Devices AD5791BRUZ IC?
Click to expand...

Note that chip isn't actually intended for audio use. Obviously it can be made to work for audio, and based on all the good press, it can sound good. But, based on what I read, Schiit did a fair amount of work in the digital domain to convert PCM to a format that chip can understand.
 
abraxalito

abraxalito

Member
Joined 2007
iancanada said:
That's a interesting idea. I'll have a look at it.
Click to expand...

I'm not really convinced its such a good idea, mainly because that ADI DAC is designed for relatively low frequency operation, the datasheet seems rather coy about its dynamic performance (most importantly, settling time) which is important for audio. Its inherently glitchy so needs a de-glitcher following it. Its also pretty expensive as DAC chips go.
 
iancanada

iancanada

Member
Joined 2009
twitter.com/iancanadaTT
Paid Member
ofswitched said:
The chip is a small cost for assembled DAC cost. If mono mode is better, why are Wadia, wyred4sound or Northstar not used? At the same time, they are not to use very expensive clock chip either. Obviously, the sound is beautiful or not, not because of the test data.

The resistors in the chip can not be high-precision. DAC using the R-2R DAC chip will not be as good as soekris. In contrast, the well-designed 9018 DAC has far better sound quality than soekris. Why?

Some people claim that low-cost R-2R sounds good, because such low-cost R-2R makes the sound obscure, so that in fact the ugly looks like the beauty.
Click to expand...

Hi Ofswitched,

All Wadia flagships were designed in mono block and parallel configuration:

Wadia 861SE/861/781i/581/381i: PCM1704K * 4

Wadia 860: PCM1702K *4

Wadia 9: PCM63K * 8

Regards,
Ian
 
You must log in or register to reply here.
Top Bottom