Its AVCC that one would use an opamp buffer supply for.
The other two rails, VCCA and DVCC require very little current although ESS does not specify it under all operating conditions. Best if the two are kept separate if possible. Low noise LDO regulators work just fine for them. Perhaps of some importance is to use R-core transformers for all power supplies since they are best in terms of minimizing coupling of AC power lone noise into critical power supply grounds. One of the reasons people use batteries is to avoid problems like ground coupled noise from the power lines, so wise to keep it to a minimum if using power transformers.
The other two rails, VCCA and DVCC require very little current although ESS does not specify it under all operating conditions. Best if the two are kept separate if possible. Low noise LDO regulators work just fine for them. Perhaps of some importance is to use R-core transformers for all power supplies since they are best in terms of minimizing coupling of AC power lone noise into critical power supply grounds. One of the reasons people use batteries is to avoid problems like ground coupled noise from the power lines, so wise to keep it to a minimum if using power transformers.
All of them are below 100mA, so I don't think there is any problem with Reflektor D. Set you CCS to 150-300mA would be pretty safe.
Good weekend.
Ian
Actually I like the idea of powering the DAC via a separate power supply from the RPi.
I believe DACs prefer linear power supplies while RPis are best served using an SMPS. Check out the RasaPlayer below.
Rasa Player
I ran into this thread while googling for a 9038Pro board that could be connected to the GPIO for I2S (but powered independently). Your logic for using dual 9038Q2M chips instead of a single 9038Pro chip is sound. Thank you for presenting this argument, I never considered this.
AudioPhonics has a full-blown kit using dual 9038Q2M chips
AUDIOPHONICS EVO-SABRE DAC Kit DIY DAC 2xES9038Q2M Fully Balanced & Streamer for Raspberry Pi 4 - Audiophonics
Soncoz also uses 2x9038Q2M chips and has achieved very good results.
SGD1
In effect, the only ESS chip-based DAC that is going to outperform a dual 9038Q2M would be one that has 2 9038Pro chips or maybe a pair of the new ES9068AS chip.
ES9068AS Flagship DAC w/MQA Evaluation Board - Ismosys
Gustard X16 DAC (Digital-to-Analog Converter) – Apos Audio
Until AKM recovers and rebuilds their factories we will have to limit ourselves to ESS and maybe Wolfson and Burr Brown chips.
Last edited:
Just one question. From the spec sheet the 9038Q2M in dual-mono (2CH in parallel) mode has
Output impedance: 387 ohm
DNR differential current mode: 132dB(theoretical)
THD+N differential current mode: -120 dB
The 9038Pro in stereo mode (4 channels in parallel) however has
Output impedance: 50.5 ohm
DNR differential current mode: 137dB
THD+N differential current mode: -122 dB
4 CH in parallel
How did you arrive at a DNR of 140db?
Output impedance: 387 ohm
DNR differential current mode: 132dB(theoretical)
THD+N differential current mode: -120 dB
The 9038Pro in stereo mode (4 channels in parallel) however has
Output impedance: 50.5 ohm
DNR differential current mode: 137dB
THD+N differential current mode: -122 dB
4 CH in parallel
How did you arrive at a DNR of 140db?
It took me a while to read this whole thread and it has been a very steep learning curve.
I was looking to build an RPi4 based streamer connected to a DAC via I2S on the GPIO.
The DAC would also have additional inputs (TosLink, SPDIF Coax, USB, and BT) so I could if I want to connect my Game Console (TosLink), Marantz KI CD Player (Coax), my laptop (USB) and my mobile player (BT) to the same DAC.
Totally enjoying the engineering required for this "project" especially the power supply section. Most other kits I have seen ignore the power supply.
I would be happy to buy the parts and make a copy if you can have the final product available as a kit with the option to choose between the LINEARPI and LIFEPO4?
Could the layout be organised so that it would fit into a box that resembles an audio component (2" high, 17" wide, 16" deep)? Maybe use a ribbon cable to connect the GPIO instead of stacking the boards?
Last edited:
I/V STD MkII
I/V STD MkII new features:
1. High precision low TCR thin metal film SMT resistors.
2. Decoupling capacitance increased to 2000uF for each analog voltage rails.
3. New optional input connector makes it possible to work with other DACs or as an external I/V stage.
4. More room to install discrete op-amplifier boards.
IvStdMkII by Ian, on Flickr
Ian
I/V STD MkII new features:
1. High precision low TCR thin metal film SMT resistors.
2. Decoupling capacitance increased to 2000uF for each analog voltage rails.
3. New optional input connector makes it possible to work with other DACs or as an external I/V stage.
4. More room to install discrete op-amplifier boards.
IvStdMkII by Ian, on Flickr
Ian
Actually, I have McFifo. Don't like the clock board part of it. The FIFO board itself may be okay. If I were going to to use the FIFO board, I would probably design my own clock system.
Regarding the number of channels it has, yes, it supports a lot of channels both for PCM and for DSD. The other part of the overall problem is where the crossover DSP is going to be done. If in a DSP chip then its pretty much limited to PCM at that point. However, many dacs sound better playing DSD256 or DSD512. IMHO that includes the Sabre dacs. One could potentially do the conversion to DSD before the FIFO and after a crossover DSP chip, but the only good quality off-the-shelf hardware converter for that, AK4137, is out of production for now due to the AKM factory fire. Otherwise, its probably best to do all the DSP and DSD conversion in a PC using HQ Player or similar software. In that case a multi-channel USB to I2S board would be needed. Diyinhk may have one, not sure.
Also, IME dual mono dac chips is one of the last and least important things to worry about. Quite good imaging can be produced using a single dac chip if everything else is done right. If everything else isn't done right, then there is not all that much to be gained by dual mono dac chips, as the benefit is much smaller than getting the other stiff right.
Anyway, my point is that one probably ought to really think through the whole system before starting out trying to put something together.
Regarding the number of channels it has, yes, it supports a lot of channels both for PCM and for DSD. The other part of the overall problem is where the crossover DSP is going to be done. If in a DSP chip then its pretty much limited to PCM at that point. However, many dacs sound better playing DSD256 or DSD512. IMHO that includes the Sabre dacs. One could potentially do the conversion to DSD before the FIFO and after a crossover DSP chip, but the only good quality off-the-shelf hardware converter for that, AK4137, is out of production for now due to the AKM factory fire. Otherwise, its probably best to do all the DSP and DSD conversion in a PC using HQ Player or similar software. In that case a multi-channel USB to I2S board would be needed. Diyinhk may have one, not sure.
Also, IME dual mono dac chips is one of the last and least important things to worry about. Quite good imaging can be produced using a single dac chip if everything else is done right. If everything else isn't done right, then there is not all that much to be gained by dual mono dac chips, as the benefit is much smaller than getting the other stiff right.
Anyway, my point is that one probably ought to really think through the whole system before starting out trying to put something together.
Last edited:
@Markw4
McFifo has the McDualXO clock board separate from it. So you are welcome to design a customized clock board for it. That's so-called DIY friendly design. McDualXO is already very good, if you want to make it better, I would suggest you using discrete logic chips to improve IR-drop of the multi-bits chips.
I have both dual mono and signal chip DACs. Dual mono DACs are always better at any point. Especially my ES9038Q2M Dual Mono plus running at true sync clock mode.
Based on my real listening test on my system, the sound quality of ES9038Q2M Dual Mono plus is better than any other ES9038Q2M single chip DACs no matter they are in which mode (async, master or sync). The principle why dual mono is better was discussed many times. I don't think what you are worried about is reasonable. I would suggest doing the real listening test by yourself to confirm.
Regards,Ian
McFifo has the McDualXO clock board separate from it. So you are welcome to design a customized clock board for it. That's so-called DIY friendly design. McDualXO is already very good, if you want to make it better, I would suggest you using discrete logic chips to improve IR-drop of the multi-bits chips.
I have both dual mono and signal chip DACs. Dual mono DACs are always better at any point. Especially my ES9038Q2M Dual Mono plus running at true sync clock mode.
Based on my real listening test on my system, the sound quality of ES9038Q2M Dual Mono plus is better than any other ES9038Q2M single chip DACs no matter they are in which mode (async, master or sync). The principle why dual mono is better was discussed many times. I don't think what you are worried about is reasonable. I would suggest doing the real listening test by yourself to confirm.
Regards,Ian
@iancanada
I did try your dual ES9038Q2M board. Sounded pretty good in synchronous DSD mode, aside from two issues. One issue shown in listening tests was a reduction of stereo imaging compared to at least one single chip ES9038Q2M dac with separate AVCC_L and AVCC_R power supplies. The second issue was noise coupling with boards on either side of it in the stack, the FIFO_Pi and the opamp output stage board. Regarding the latter issue, it looks like shielding and isolation is being improved over time.
I did try your dual ES9038Q2M board. Sounded pretty good in synchronous DSD mode, aside from two issues. One issue shown in listening tests was a reduction of stereo imaging compared to at least one single chip ES9038Q2M dac with separate AVCC_L and AVCC_R power supplies. The second issue was noise coupling with boards on either side of it in the stack, the FIFO_Pi and the opamp output stage board. Regarding the latter issue, it looks like shielding and isolation is being improved over time.
What does real listening test consist of? Is it single blind, double blind, triple blind, level matched between components or something else?
@Markw4
Have you ever tested on my ES9038Q2M Dual Mono Plus which has all the voltage rails separated?
Yes, you are right, the power supplies, the isolation and the shield for EMI noise are all very important for the sound quality of a DAC.
My system uses StationPi to separate RPi and Audio HATs with additional shields. And, I use LifePO4 and ultra capacitor hybrid power supplies for myES9038Q2M Dual Mono Plus DAC. I don't have any of above issues.Please watch my YouTube video for more details:
The Best Sound Quality Audiophile DIY DAC I Built in 2020 - YouTube
DAC section
1. StationPi
2. Raspberry Pi with Volumio
3. FifoPi Q3
4. ReclockPi (optional)
5. ES9038Q2M Dual Mono Plus
6. ESS controller
7. Transform I/V
Power Supply Section
1. LifePO4 MkIII power supply
2. UcAdapter (optional)
3. UcMateConditioner 5V (optional)
4. UcHybrid 3.3V x 4
5. UcConditioner 3.3V
Recently I'm working on a OPA861 Zero feedback balanced I/V stage. The real listening test results are very impressive. It brings the sound quality of the ES9038Q2M Dual Mono Plus DAC to an even higher level.
Any way, all of those improvements work for both dual mono and single chip DAC configuration. However, dual mono DACs still have better sound quality under the same conditions.
Regards,
Ian
Have you ever tested on my ES9038Q2M Dual Mono Plus which has all the voltage rails separated?
Yes, you are right, the power supplies, the isolation and the shield for EMI noise are all very important for the sound quality of a DAC.
My system uses StationPi to separate RPi and Audio HATs with additional shields. And, I use LifePO4 and ultra capacitor hybrid power supplies for myES9038Q2M Dual Mono Plus DAC. I don't have any of above issues.Please watch my YouTube video for more details:
The Best Sound Quality Audiophile DIY DAC I Built in 2020 - YouTube
DAC section
1. StationPi
2. Raspberry Pi with Volumio
3. FifoPi Q3
4. ReclockPi (optional)
5. ES9038Q2M Dual Mono Plus
6. ESS controller
7. Transform I/V
Power Supply Section
1. LifePO4 MkIII power supply
2. UcAdapter (optional)
3. UcMateConditioner 5V (optional)
4. UcHybrid 3.3V x 4
5. UcConditioner 3.3V
Recently I'm working on a OPA861 Zero feedback balanced I/V stage. The real listening test results are very impressive. It brings the sound quality of the ES9038Q2M Dual Mono Plus DAC to an even higher level.
Any way, all of those improvements work for both dual mono and single chip DAC configuration. However, dual mono DACs still have better sound quality under the same conditions.
Regards,
Ian