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

It's a pity. If the reclock module can be connected directly to a input terminal of DA chip, theoretically there will be almost no jitter in the digital circuit, and if the reclock module can be integrated with DA chip, the clock signal can be very close to provide the DA chip without the clock jiiter. I think this is the closest to perfect digital circuit design?

Hi ofswitched,

That's a good question.

The key things that decide the jitter performance are the XO itself, power supply and the oscillator circuit. A couple of inches well impedance matched coaxial cable doesn't really add jitter to the MCLK. It still can be looked upon as local clock.

However, for ESS DACs, the MCLK frequency can go up to 100MHz. In this case, RF noise, ground loop noise and cross talk all have to be taken into account to avoid from any degrading of DAC performance.

That's why I still prefer designing the clock as an independent section or as a module. Actually many high end audio system have independent clock section/module designed in this way, though sometimes the clock section could be located at the same big DAC board.

But any way, it's just a personal opinion.

Regards,
Ian
 
In regards to ofswitched’s post. IMHO there is a third and often overlooked position. In general I’m a fan of the modular approach as long as you’re getting better or at least negligible worsening of output.
Here comes the “but”... But, we’re dealing with piHats here. Real estate is limited as well as how much stacking can be permitted.
Let’s assume we all want the “holy” trio. DacHat+I/V board+isolator +reclocker+PSU. That will make a huge stack... there is no way in h*** that will fit in my case.
By cramming as much as possible onto the hats, it won’t build as high and there is a chance to fit the build in a normal case.
 
Hi ofswitched,

That's a good question.

The key things that decide the jitter performance are the XO itself, power supply and the oscillator circuit. A couple of inches well impedance matched coaxial cable doesn't really add jitter to the MCLK. It still can be looked upon as local clock.

However, for ESS DACs, the MCLK frequency can go up to 100MHz. In this case, RF noise, ground loop noise and cross talk all have to be taken into account to avoid from any degrading of DAC performance.

That's why I still prefer designing the clock as an independent section or as a module. Actually many high end audio system have independent clock section/module designed in this way, though sometimes the clock section could be located at the same big DAC board.

But any way, it's just a personal opinion.

Regards,
Ian

I have designed a metal cans for the clock chip. It was slightly affected on the sound, but it was not obvious. This means that RF noise isn't a big issue? The ground loop noise and cross talk need to consider?
 

Attachments

  • 37543.jpg
    37543.jpg
    123.2 KB · Views: 1,012
  • 37544.jpg
    37544.jpg
    153.5 KB · Views: 998
Hi Ian, that is a great success and I am very pleased to see of your smart implementation! I modified these Chinese 9038Q2M board heavily (it is in the ES9038Q2M Board topic, #390, http://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-39.html#post5379006). Because of the size of the board, there is no more space left to step further. But your controller could release some more necessary place (if the on-board controller was removed)! Is it possible to get/buy of your controller for my Chinese board? I would appreciated your help! It is off topic here, but I did try the OPA1612. I don’t know the difference in between OPA1612 and OPA1622, but the Burson V5 is far-far better than my 1612. This difference is more listenable by upgraded power supplies.


I bought this ES9038Q2M DAC last year, just want to confirm my ESS controller works for third party ESS DACs.
Yes, it works.
With my ESS controller, all above issues are fixed. DAC chip-id was recognized as ES9038Q2M. It plays 44.1KHz 16bit music nicely, as well as all other formats. The DoP was also been enables and plays without any problem.
Please see pictures for configuration details.
Some experience to share:

1. u.fl coaxial cables are tested much better than regular jumper wires for input signals. One of the pictures shows how to mount u.fl socket at bottom side of the DAC PCB.
2. The on-board MCU has to be removed in order to use external ESS controller. Two I2C pins can be soldered right on top of the pull-up resistors.
3. When I replace the original OP with OPA1622, I got some improvement on sound quality. It can also drive my Sennheiser 300ohm headphone.
Ian
 
Last edited:
Still following this thread with interest. Thank you.
Over in the group buy for the Signalyst DSCv2 DSD DAC, output transformers are either Lundahl LL1527XL or LL 1684, or a much cheaper Chinese transformer.
In the interests of keeping the total price down, Ian and madds1, would you maybe consider trying these Chinese transformers?
2PCS / 600:10K permalloy audio audio amplifier audio transformer isolator-in Transformers from Home Improvement on Aliexpress.com | Alibaba Group

I have used one brand of Chinese permalloy 10k:10K transformer and found them pretty good. I even did a quick comparison with a Jensen pt-11 that I also have at a ak4497 output and I am not sure I can hear a big difference.
I brought from Taobao but I believe its listed at aliexpress as well (under lgaudio, look for this name under the item details, and they will also specific the primary or secondary inductance and the frequency range). They are normally in the very high inductance range and mine is labelled around 300H (I tried to measure it but its beyond my LCR 100H limit)
Note sometimes those transformers have several grades (price) and the higher the inductance in the spec the high the price.

This is one example:

600:60K permalloy audio isolation transformer, balanced and unbalanced conversion audio isolator-in Transformers from Home Improvement on Aliexpress.com | Alibaba Group
 
Hi Ian, any thoughts when the controller and DAC boards will be ready? I have some thought / questions ...

1. Is a driver available for the volume control of the controller such that I can use the alsa mixer to directly control the volume on the DAC (hardware as opposed to software control) ? I would want to control the volume using a volume control knob as well as the alsa mixer via mpd.

2. Will these boards work with say the Allo Sparky or an odroid?

3. I personally think a modular approach would be best - people can purchase the output boards as they wish. The ability to have a modular board for optical and coaxial (and may be a stereo line in as well will be helpful. By the time a decent PSU is added, it is unlikely to fit into a standard case anyway.

4. My ideal solution would be a high quality all-in-one RPi3 + DAC + amplifier in a nice case. I was considering a Hypex ncore 400 or uCD180HG. I think the DAC output could be fed directly into the hypex modules? A galaxy case with maybe a CNC cut wood front and various acrylic laser cut back panels for the various options?

5. Something like this could possibly surpass the Naim Uniti Atom for less than half the cost with more functionality. Perhaps a second chassis with 3x amps to make this into a high quality 5.1 system for music and movies? I have the ideas but unfortunately not the time, electronic skills or business sense to make this happen!
 
Hi Ian, any thoughts when the controller and DAC boards will be ready? I have some thought / questions ...

1. Is a driver available for the volume control of the controller such that I can use the alsa mixer to directly control the volume on the DAC (hardware as opposed to software control) ? I would want to control the volume using a volume control knob as well as the alsa mixer via mpd.

2. Will these boards work with say the Allo Sparky or an odroid?

3. I personally think a modular approach would be best - people can purchase the output boards as they wish. The ability to have a modular board for optical and coaxial (and may be a stereo line in as well will be helpful. By the time a decent PSU is added, it is unlikely to fit into a standard case anyway.

4. My ideal solution would be a high quality all-in-one RPi3 + DAC + amplifier in a nice case. I was considering a Hypex ncore 400 or uCD180HG. I think the DAC output could be fed directly into the hypex modules? A galaxy case with maybe a CNC cut wood front and various acrylic laser cut back panels for the various options?

5. Something like this could possibly surpass the Naim Uniti Atom for less than half the cost with more functionality. Perhaps a second chassis with 3x amps to make this into a high quality 5.1 system for music and movies? I have the ideas but unfortunately not the time, electronic skills or business sense to make this happen!



This actually comes very close to my current setup.
I have a RPi2,BII with a transformer I/v and nCore 400 mono blocks. Ok, it’s not identical to what you propose but the sonic signature should be in the same ball park.
And my current build is a RPi3, ess DAC (newer dac and better transformer I/v) unit to replace the existing one.
The nCore 400s really are great amps, true giant killers. But as everything HiFi, it’s important to get a good matching.
 
This actually comes very close to my current setup.
I have a RPi2,BII with a transformer I/v and nCore 400 mono blocks. Ok, it’s not identical to what you propose but the sonic signature should be in the same ball park.
And my current build is a RPi3, ess DAC (newer dac and better transformer I/v) unit to replace the existing one.
The nCore 400s really are great amps, true giant killers. But as everything HiFi, it’s important to get a good matching.

I am envious! Once Ian's boards are ready I might just bite the bullet and see if I can put it all together :)