DAC HAT ES9038Q2M vs ES9028Q2M
Since I designed both ES9038Q2M and ES9028Q2M DAC HAT. The next, I'm gonna get them compared side by side. Hopefully I will find something interesting.
ES9038Q2MES9028Q2M by Ian, on Flickr
Ian
Since I designed both ES9038Q2M and ES9028Q2M DAC HAT. The next, I'm gonna get them compared side by side. Hopefully I will find something interesting.
ES9038Q2MES9028Q2M by Ian, on Flickr
Ian
Last edited:
ES9028Q2M\ES9038Q2M Test set up
ES9028Q2M\ES9038Q2M Test set up
1. Raspberry Pi 2 model B with Volumio, Moode or piCorePlayer or any other player software.
The reason I still prefer Pi 2 is that I don't like the noisy Wifi on Pi 3. It doesn't stop working even it is disabled.
Please let me know if anybody knows how to 100% shut down the Pi 3 wifi module.
2. FIFO reclock with built-in isolator and integrated clock board. To reach higher sound quality, I'll use 3.3V LifePO4 battery to power the isolated clock section. Crystek CCHD957 45.1584/49.1520MHz XOs will also be used for the test.
3. ES9028Q2M and ES9038Q2M I2S/DSD DAC HATs to evaluate. Both DAC HAT will be setup into true sync mode. MCLK will be fed from FIFO via u.fl coaxial cable.
To reach best possible sound quality, 3.3V LiFePO4 battery will also be used to replace the on board LT3042 ultra-low noise regulator.
4. Universal ESS controller. it will be connected to non-isolated GPIO on the FIFO board to make it working as an isolated controller to those DAC HAT.
Besides the true sync mode, all other settings will be kept as factory default. Volume will be set to 0dB.
Both Apple remote and control knob encoder can be used for the controller for now. Software based control will also be available when the driver is finished.
5. I/V stage: LL1674 or LL1544A transformer I/V. I'll discuss why I like transformer I/V stage later.
ES90389028Q2MSETUP3 by Ian, on Flickr
ES90389028Q2MSETUP2 by Ian, on Flickr
Ian
ES9028Q2M\ES9038Q2M Test set up
1. Raspberry Pi 2 model B with Volumio, Moode or piCorePlayer or any other player software.
The reason I still prefer Pi 2 is that I don't like the noisy Wifi on Pi 3. It doesn't stop working even it is disabled.
Please let me know if anybody knows how to 100% shut down the Pi 3 wifi module.
2. FIFO reclock with built-in isolator and integrated clock board. To reach higher sound quality, I'll use 3.3V LifePO4 battery to power the isolated clock section. Crystek CCHD957 45.1584/49.1520MHz XOs will also be used for the test.
3. ES9028Q2M and ES9038Q2M I2S/DSD DAC HATs to evaluate. Both DAC HAT will be setup into true sync mode. MCLK will be fed from FIFO via u.fl coaxial cable.
To reach best possible sound quality, 3.3V LiFePO4 battery will also be used to replace the on board LT3042 ultra-low noise regulator.
4. Universal ESS controller. it will be connected to non-isolated GPIO on the FIFO board to make it working as an isolated controller to those DAC HAT.
Besides the true sync mode, all other settings will be kept as factory default. Volume will be set to 0dB.
Both Apple remote and control knob encoder can be used for the controller for now. Software based control will also be available when the driver is finished.
5. I/V stage: LL1674 or LL1544A transformer I/V. I'll discuss why I like transformer I/V stage later.
ES90389028Q2MSETUP3 by Ian, on Flickr
ES90389028Q2MSETUP2 by Ian, on Flickr
Ian
Ian,
Sorry, I missed this. Busy with work last week and then I was AWOL from Audio stuff doing my other hobby, model aviation, late last week and the weekend.
Each setup was run on exactly the same front-end, power supplies, and music selections. Time between listening to one, then the other was typically 10-15 minutes, about as fast as I could swap things over. I did give the amps some warm-up time (I put them into standby to swap, so they warm up pretty fast if not in standby long).
I really liked the Joe R / Madds1 / Rod Coleman setup on both output transformer setups I've tried so far. After listening some over the weekend, I feel it still needs a little more aggressive filtering. So I will swap in the trimming cap 1000pF from 470pF. I will do that soon, but I have another setup going into that system for some listening trials for a few days, so I won't be listening to your DAC card / output transformers setup again until likely this weekend.
I got similar great service from K&K. I'm very interested in how you like the LL1544A transformers. I was comparing their schematics to the Onetics, LL1570XL, and LL1684 I have here and your other transformer, the LL1674. The key difference is the ability to strap the LL1544A into 1:4, which could be VERY useful in allowing a lower R input resistor, pushing it closer to current mode.
What I've heard so far playing with this, it is definitely a winning combo.
Also VERY excited about your FIFO board!!!!!
Finally, I am TOTALLY with you using the RPi 2B for audio over the RPi 3 with built-in WiFi. ALL of my RPi setups use RPi 2Bs for that reason.
Ian, do take a look at PiCorePlayer. I believe it has a selectable function to turn off the RPi 3's Wifi system, though I haven't confirmed this as I don't have any RPi 3's. Of course, it is based on the Logitech LMS/Squeezebox setup which requires a server running somewhere, typically on another computer for best sonics, though you can run one on the Pi to serve local content. BUT likely a quick look and some searches will point you to the right configuration settings or script needed to turn off the RPi 3's WiFi.
OTOH, my reference digital source is an SDTrans384 SD card player which is feeding a somewhat modified Soekris 1021 DAM DAC. The super-tweakers at TirNaHiFi have found a different SD card player (having tried 3 so far) that they have working very well and it is now their reference setup:
Computer Audio Blasphemy, nige tries Another SD Card Player - Page 13 - Tir Na HiFi
I have one here to try and a 2nd on the way. BUT the most interesting thing is that they are playing with a WiFi-enabled Toshiba FlashAir SD card and not finding any sonic degradation with that card compared to other cards.
I haven't played with a FlashAir much except to confirm it works ok in my SDTrans384 (which can be very picky about SD cards... only about 60%-70% of the brands / models I've tried work in it). Their goal is to use that card for part of a more sophisticated GUI & control system than what comes with their SD Card player:
SD player UI development - Page 11 - Tir Na HiFi
I would assume that the WiFi setup on these SD Cards is pretty low-power, which would aid in reducing any sonic impacts.
AND with an SD card to micro adapter, it could be used as the boot card for the RPi, possibly providing another network connection for control signals. I use these adapters to host full-sized 512Gb SD Cards in the micro SD card slots on my low-power LMS server computers (ZOTAC ZBOX PI320-W2) and store my main selections from my library on those cards.
Greg in Mississippi
ES9028Q2M\ES9038Q2M Test set up
1. Raspberry Pi 2 model B with Volumio, Moode or piCorePlayer or any other player software.
The reason I still prefer Pi 2 is that I don't like the noisy Wifi on Pi 3. It doesn't stop working even it is disabled.
Please let me know if anybody knows how to 100% shut down the Pi 3 wifi module.
2. FIFO reclock with built-in isolator and integrated clock board. To reach higher sound quality, I'll use 3.3V LifePO4 battery to power the isolated clock section. Crystek CCHD957 45.1584/49.1520MHz XOs will also be used for the test.
3. ES9028Q2M and ES9038Q2M I2S/DSD DAC HATs to evaluate. Both DAC HAT will be setup into true sync mode. MCLK will be fed from FIFO via u.fl coaxial cable.
To reach best possible sound quality, 3.3V LiFePO4 battery will also be used to replace the on board LT3042 ultra-low noise regulator.
4. Universal ESS controller. it will be connected to non-isolated GPIO on the FIFO board to make it working as an isolated controller to those DAC HAT.
Besides the true sync mode, all other settings will be kept as factory default. Volume will be set to 0dB.
Both Apple remote and control knob encoder can be used for the controller for now. Software based control will also be available when the driver is finished.
5. I/V stage: LL1674 or LL1544A transformer I/V. I'll discuss why I like transformer I/V stage later.
ES90389028Q2MSETUP3 by Ian, on Flickr
ES90389028Q2MSETUP2 by Ian, on Flickr
Ian
The 100MHz clock signal will be converted so as not to generate extra clock jitter?
The local 100MHz clock was disabled by jumper switch when works at true sync mode with MCLK from FIFO.
Ian
Local 100MHz clock will cause interference if there is still power supply. The jumper switch will cut off the power of local 100MHz clock?
Can further explain how FIFO is integrated?
Hi Greg,
That's great approach, I'll do similar listening test. I'll focus on difference between ES9028Q2M and ES9038Q2M, as well as LL1544A and LL1674. I'll post some details then.
I'm figuring out the schematics of LL1544A at 1:4 configuration. I may need your advise before I start the test.
Regards,
Ian
When local clock is disabled, the output will be in tri-state. Normally it doesn't affect FIFO MCLK quality because the trace is very short less then half inch. However if you are really worry about it, you can remove the local 100MHz XO. Should be no difficult.
The FIFO is an all in one solution. It has everything integrated, I2S/DSD FIFO + DoP Decoder + Isolator + clock board. It's still under developing. I'll have more details later.
Regards,
Ian
ES9028Q2M\ES9038Q2M Test set up
1. Raspberry Pi 2 model B with Volumio, Moode or piCorePlayer or any other player software.
The reason I still prefer Pi 2 is that I don't like the noisy Wifi on Pi 3. It doesn't stop working even it is disabled.
Please let me know if anybody knows how to 100% shut down the Pi 3 wifi module.
2. FIFO reclock with built-in isolator and integrated clock board. To reach higher sound quality, I'll use 3.3V LifePO4 battery to power the isolated clock section. Crystek CCHD957 45.1584/49.1520MHz XOs will also be used for the test.
3. ES9028Q2M and ES9038Q2M I2S/DSD DAC HATs to evaluate. Both DAC HAT will be setup into true sync mode. MCLK will be fed from FIFO via u.fl coaxial cable.
To reach best possible sound quality, 3.3V LiFePO4 battery will also be used to replace the on board LT3042 ultra-low noise regulator.
4. Universal ESS controller. it will be connected to non-isolated GPIO on the FIFO board to make it working as an isolated controller to those DAC HAT.
Besides the true sync mode, all other settings will be kept as factory default. Volume will be set to 0dB.
Both Apple remote and control knob encoder can be used for the controller for now. Software based control will also be available when the driver is finished.
5. I/V stage: LL1674 or LL1544A transformer I/V. I'll discuss why I like transformer I/V stage later.
ES90389028Q2MSETUP3 by Ian, on Flickr
ES90389028Q2MSETUP2 by Ian, on Flickr
Ian
Very impressive!
Cant' wait to see measurement and reviews from your beta testers.
Like everyone else, I too am looking forward to your results.
Upon seeing the setups people are talking about here in the thread, I wonder where we start shooting ourselves in the proverbial foot rather than improve audio.
In a digital setup only one thing should matter. A stable clock and available data for the clock to pop. How the data got to the memory is irrelevant.
Just some food for thought.
Naturally a great psu is always a good thing. And minimizing distortion picked up from the surroundings.
Upon seeing the setups people are talking about here in the thread, I wonder where we start shooting ourselves in the proverbial foot rather than improve audio.
In a digital setup only one thing should matter. A stable clock and available data for the clock to pop. How the data got to the memory is irrelevant.
Just some food for thought.
Naturally a great psu is always a good thing. And minimizing distortion picked up from the surroundings.
Hi gentlemen, will this setup be some kind of ultimate Pi-I2S DAC board despite all Pi related I2S critics ? (I'm not an engineer but I've read the whole thread.. just asking).
I want to build a player and still struggling if I should use a small form factor PC with a Pink Faun I2S Card and Foobar2000 on Windows OR a Pi based solution like this here, still staying on I2S.
I want to build a player and still struggling if I should use a small form factor PC with a Pink Faun I2S Card and Foobar2000 on Windows OR a Pi based solution like this here, still staying on I2S.
Hi Ian i agree with you about using rpi2B because of vifi. Which version of Rpi2B (1.1-32bit or 1.2-64bit) do you use and what kernell settings that are best performance in moode? Which vesion of moode do you use. I have still 3.84 and some droputs in real time kernell vith 1.1 and no dropouts on 1.2.
Thank you
Thank you
Vortex> Not speaking for anyone else here but rarely will you find any “ultimate” anything in a small form factor.
I do however suspect Ian will get good enough given the constraints.
Good enough as in most people will never be able to detect any real discernible difference compared to cost-no-object builds.
I do however suspect Ian will get good enough given the constraints.
Good enough as in most people will never be able to detect any real discernible difference compared to cost-no-object builds.
My 2 cents (actually more like $4!!!)
I think you have seriously oversimplified what is important in digital audio.
I'd add at at a minimum:
- low jitter clock and data waveforms at the input to the D-to-A conversion process
- good clocks (needed for the previous factor)
- attention to clock and data transmission routes (ditto)
- low noise on the signal, grounds, and power feeds (needed for the above and below)
- very high bandwidth/low noise power feeds to allow the digital processing circuits to operate as designed AND without interfering with each other and other connected circuits
- reducing/eliminating electical noise introduced from circuits outside of the actual digital processing system
Isolators, reclockers, multiple low-noise/wide bandwidth power supplies are some techniques to deal with these factors, among many others.
This week I did a set of listening tests with a RPi DAC Hat setup (RPi -> IsolatorPi -> Allo Kali reclocker -> Dial DAC) where each board in the stack had a separate supply. My purpose was to compare various quality power supply options.
The lowest level supplies were modified Jameco 5V wall warts feeding the isolator and reclocker boards and the DAC board being fed from a Silent Switcher powered by another modified Jameco 5V to provide the +-15V needed by the Dial DAC.
The mid level used AC supplies (Hammond 229-series transformers / DSEP30-06A diodes / Jensen 4-pole capacitors / John Swenson transformer ringing snubbers) to feed the isolator and reclocker boards through ADM715x-based preregulators. I started this level using the Jameco/Silent Switcher to power the DAC card at first and then replaced that combo with another similar +-AC supply feeding a pair of shunt regulators (similar to Salas, but somewhat simpler).
The highest level retained the shunt-regulated +-15V AC supply, but replaced the AC supplies feeding the isolator and reclocker with 2 Uptone Audio LPS-1s.
In each level the RPi was powered from a somewhat beefed up K&K Audio 12 watt Low Voltage Power Supply.
Each step up was a clear improvement and NONE of them would I ever term 'shooting myself in the proverbial foot'. Try this for yourself, I bet you'd like it.
The best thing is that Ian's DAC here, especially with his up-coming isolator/reclocker, either addresses all of these (and other) factors OR provides you with the flexibility to address them yourself.
I suggest that a setup such as this is only as good as the implementation. Setups sharing power supplies between the RPi and the subsequent DAC card will always (in my experience) perform at a level below one with separate supplies. Adding isolation (which will require at least 2 supplies) and reclocking each provides benefits. Quality of power supplies matter. There are many 'good' RPi I2S-fed DAC setups out there today... the Dial DAC I mentioned above, Mamboberry (they have a new model that should be interesting), Allo.com's Piano 2.1, Boss 1.2, and upcoming Katana, TAUDAC are all ones I'd look at today along with Ian's DAC. BUT none of them will perform at their best if you use a cheapo RPi SMPS to power the whole stack. While that option is available on most, 'ultimate' will mean something better. If you spend time looking under the hood of many of the highly-regarded DACs that use chip D-A converters, you'll see that much of what is added to make them perform at a higher level are attention to power supplies, input signal integrity, and a good output stage. AND those factors have to be addressed in the non-chip DACs (like the PS Audio DirectStream, Chord, & Soekris FPGA DACs) to make them reach the higher ranks.
You can also do a similar setup with a small-form-factor PC and something like the Pink Faun I2S card (which looks to be a very well-engineered solution). I've done something similar with good results. BUT you still need to pay attention to power supplies to move it to the 'ultimate' level. IMHO, as long as I don't want / need to re-sample / up-sample, I don't feel I'm giving anything up going to an RPi-based setup over my previous motherboard / modified Juli@ I2S board with 13 linear power supplies (including a linear ATX-24/P-4 supply).
In this area, we agree. Ultimate takes significant infrastructure, mostly in the form of power supplies, but also techniques to improve signal integrity such as Ian's upcoming isolator / reclocker board. Yes, I have some very good sounding RPi-based digital setups, with several of them using DAC HATS such as Ian's here. BUT the power supplies (and related music server, networking gear, and power supplies for them) take up a lot of real-estate. Honestly though, the input and DAC circuitry does not need to be much larger than Ian's stack pictured above for even an 'ultimate' setup... it is the power supplies feeding them that move them to a larger form factor.
My 2 cents!
Greg in Mississippi
I think you have seriously oversimplified what is important in digital audio.
I'd add at at a minimum:
- low jitter clock and data waveforms at the input to the D-to-A conversion process
- good clocks (needed for the previous factor)
- attention to clock and data transmission routes (ditto)
- low noise on the signal, grounds, and power feeds (needed for the above and below)
- very high bandwidth/low noise power feeds to allow the digital processing circuits to operate as designed AND without interfering with each other and other connected circuits
- reducing/eliminating electical noise introduced from circuits outside of the actual digital processing system
Isolators, reclockers, multiple low-noise/wide bandwidth power supplies are some techniques to deal with these factors, among many others.
This week I did a set of listening tests with a RPi DAC Hat setup (RPi -> IsolatorPi -> Allo Kali reclocker -> Dial DAC) where each board in the stack had a separate supply. My purpose was to compare various quality power supply options.
The lowest level supplies were modified Jameco 5V wall warts feeding the isolator and reclocker boards and the DAC board being fed from a Silent Switcher powered by another modified Jameco 5V to provide the +-15V needed by the Dial DAC.
The mid level used AC supplies (Hammond 229-series transformers / DSEP30-06A diodes / Jensen 4-pole capacitors / John Swenson transformer ringing snubbers) to feed the isolator and reclocker boards through ADM715x-based preregulators. I started this level using the Jameco/Silent Switcher to power the DAC card at first and then replaced that combo with another similar +-AC supply feeding a pair of shunt regulators (similar to Salas, but somewhat simpler).
The highest level retained the shunt-regulated +-15V AC supply, but replaced the AC supplies feeding the isolator and reclocker with 2 Uptone Audio LPS-1s.
In each level the RPi was powered from a somewhat beefed up K&K Audio 12 watt Low Voltage Power Supply.
Each step up was a clear improvement and NONE of them would I ever term 'shooting myself in the proverbial foot'. Try this for yourself, I bet you'd like it.
The best thing is that Ian's DAC here, especially with his up-coming isolator/reclocker, either addresses all of these (and other) factors OR provides you with the flexibility to address them yourself.
I suggest that a setup such as this is only as good as the implementation. Setups sharing power supplies between the RPi and the subsequent DAC card will always (in my experience) perform at a level below one with separate supplies. Adding isolation (which will require at least 2 supplies) and reclocking each provides benefits. Quality of power supplies matter. There are many 'good' RPi I2S-fed DAC setups out there today... the Dial DAC I mentioned above, Mamboberry (they have a new model that should be interesting), Allo.com's Piano 2.1, Boss 1.2, and upcoming Katana, TAUDAC are all ones I'd look at today along with Ian's DAC. BUT none of them will perform at their best if you use a cheapo RPi SMPS to power the whole stack. While that option is available on most, 'ultimate' will mean something better. If you spend time looking under the hood of many of the highly-regarded DACs that use chip D-A converters, you'll see that much of what is added to make them perform at a higher level are attention to power supplies, input signal integrity, and a good output stage. AND those factors have to be addressed in the non-chip DACs (like the PS Audio DirectStream, Chord, & Soekris FPGA DACs) to make them reach the higher ranks.
You can also do a similar setup with a small-form-factor PC and something like the Pink Faun I2S card (which looks to be a very well-engineered solution). I've done something similar with good results. BUT you still need to pay attention to power supplies to move it to the 'ultimate' level. IMHO, as long as I don't want / need to re-sample / up-sample, I don't feel I'm giving anything up going to an RPi-based setup over my previous motherboard / modified Juli@ I2S board with 13 linear power supplies (including a linear ATX-24/P-4 supply).
In this area, we agree. Ultimate takes significant infrastructure, mostly in the form of power supplies, but also techniques to improve signal integrity such as Ian's upcoming isolator / reclocker board. Yes, I have some very good sounding RPi-based digital setups, with several of them using DAC HATS such as Ian's here. BUT the power supplies (and related music server, networking gear, and power supplies for them) take up a lot of real-estate. Honestly though, the input and DAC circuitry does not need to be much larger than Ian's stack pictured above for even an 'ultimate' setup... it is the power supplies feeding them that move them to a larger form factor.
My 2 cents!
Greg in Mississippi
Last edited:
Greg> Who am I to argue. That post was indeed borderline article material.
I think we agree on most things but use very different language. I totally agree that my previous post was a very simplified description. To be perfectly frank, if I tried using complicated language I would probably just embarrass myself.
I believe in simplicity where possible. I usually try to work with basic physics as much as possible. Keeping it simple leaves less room for errors. I am an amateur after all.
I also believe in the law of diminishing returns. The trick lies in finding a good balance.
I think we agree on most things but use very different language. I totally agree that my previous post was a very simplified description. To be perfectly frank, if I tried using complicated language I would probably just embarrass myself.
I believe in simplicity where possible. I usually try to work with basic physics as much as possible. Keeping it simple leaves less room for errors. I am an amateur after all.
I also believe in the law of diminishing returns. The trick lies in finding a good balance.