Is there any multichannel DAC easy to DIY in 2023?

[meooms]

and BTW, gave it another go with the soldering iron, risking even more than in previous attempts, AND THE DC OFFSET IS GONE!!!!!!!! cant believe it, finally!! i know qfn ics wouldnt be easy to solder the first attempt.... I am happy, i am going to give the DAC section an ok, and move on to the microcontroller work

Congrats! I'm impressed

 

[meooms]

Example from the datasheet. They split and don't even bother writing 114 and 118, that must be all 0:

View attachment 314415

My problem seems to be in the splitting. See voodooless suggestion.

 

[MCH]

First measurement attempt. This is as well the first measurement i ever made with my current gear, and probably the second or third i ever made, so bear with me.
Set up is ES9080 dac xlr, channels 7,8 combined into 1 balanced channel -> motu ultralite line 1 input (the one in the front)
I generated a 1 kHz 96/24 tone in audacity and saved it on a raspberry pi as wav. Playing it with aplay.

question:
I need to play the tone at -15dB because otherwise it clips badly the input of the Motu (the signal is 4 volts). I can't find the way to attenuate the input at the Motu, (@mdsimon2 you know everything about the ultralite )so what you see here is the tone playing at -15 dB. I guess that explains the poor noise result.

THD looks fine to me (THD compensation parameters set to 0, I might play with them later and see what they do @meooms ). Adding -15dB to the -81.2 dB THD+N figure also looks fine, not sure if this is correct though. The signal at 500 Hz, well, no idea, suggestions are welcome.

 

[MCH]

First measurement attempt. This is as well the first measurement i ever made with my current gear, and probably the second or third i ever made, so bear with me.
Set up is ES9080 dac xlr, channels 7,8 combined into 1 balanced channel -> motu ultralite line 1 input (the one in the front)
I generated a 1 kHz 96/24 tone in audacity and saved it on a raspberry pi as wav. Playing it with aplay.

question:
I need to play the tone at -15dB because otherwise it clips badly the input of the Motu (the signal is 4 volts). I can't find the way to attenuate the input at the Motu, (@mdsimon2 you know everything about the ultralite )so what you see here is the tone playing at -15 dB. I guess that explains the poor noise result.

THD looks fine to me (THD compensation parameters set to 0, I might play with them later and see what they do @meooms ). Adding -15dB to the -81.2 dB THD+N figure also looks fine, not sure if this is correct though. The signal at 500 Hz, well, no idea, suggestions are welcome.

View attachment 314506

I am definitely doing something wrong because if i continue decreasing the volume of the dac the noise figure improves significantly. here at -60 dB:

edit: previous capture was wrong. this is the capture with -60dB volume at the DAC

 

[Sokel]

I am definitely doing something wrong because if i continue decreasing the volume of the dac the noise figure improves significantly. here at -60 dB:

View attachment 314555

You don't need any external signal,just use REW's generator and test any signal at any level you like.
You can also use Multitone Analyzer which also has lots of presets too so you measure straight away,plus there's a great thread here about it.

 

[MCH]

You don't need any external signal,just use REW's generator and test any signal at any level you like.
You can also use Multitone Analyzer which also has lots of presets too so you measure straight away,plus there's a great thread here about it.

View attachment 314558

My limitation is that the DAC is not connected to a PC but to a raspberry pi.

I am a bit puzzled about the noise result when the signal is strong (motu at the verge of clipping ). At -60 dB it is at datasheet levels though.

 

[Sokel]

My limitation is that the DAC is not connected to a PC but to a raspberry pi.

I am a bit puzzled about the noise result when the signal is strong (motu at the verge of clipping ). At -60 dB it is at datasheet levels though.

Use ASIO or WASAPI exclusive for your measurements,otherwise a lot of stuff interfere.

 

[MCH]

Use ASIO or WASAPI exclusive for your measurements,otherwise a lot of stuff interfere.

Downloaded asio4all, but when I press "open ASIO control panel" in Rew, it opens Motu cuemix

 

[voodooless]

My limitation is that the DAC is not connected to a PC but to a raspberry pi.

You may want to set it up as a USB gadget device to accept audio from the PC:

Raspberry Pi as USB to I2S adapter

One of the lesser known features of the Raspbery Pi Zero is that you can use its USB port in device mode with the various linux 'gadget' drivers. The audio gadget driver will make the Pi appear to be a USB audio interface when you plug it into a PC, streamer or whatever USB host, with certain...

www.audiosciencereview.com

 

[MCH]

You may want to set it up as a USB gadget device to accept audio from the PC:

Raspberry Pi as USB to I2S adapter

One of the lesser known features of the Raspbery Pi Zero is that you can use its USB port in device mode with the various linux 'gadget' drivers. The audio gadget driver will make the Pi appear to be a USB audio interface when you plug it into a PC, streamer or whatever USB host, with certain...

www.audiosciencereview.com

True, for the measurements I only need 2 channels actually.... (Thinking of skipping the multichannel card all together and use the pi as i2s card)

In either case, are the results posted completely useless or do you think they provide a rough idea of the actual performance?

 

[voodooless]

True, for the measurements I only need 2 channels actually.... (Thinking of skipping the multichannel card all together and use the pi as i2s card)

How are you planning to do multichannel I2S on a Pi anyway? As far as I know, this seems to be quite broken? Maybe newer incarnations of the Pi have this working?

You can make a gadget with as many channels as you like. No need to limit to only 2

In either case, are the results posted completely useless or do you think they provide a rough idea of the actual performance?

If you use the proper test files, the results should be usable. In case you have gain issues, you may also try the volume control of the ESS chip.

 

[mdsimon2]

I need to play the tone at -15dB because otherwise it clips badly the input of the Motu (the signal is 4 volts). I can't find the way to attenuate the input at the Motu, (@mdsimon2 you know everything about the ultralite )so what you see here is the tone playing at -15 dB. I guess that explains the poor noise result.

Based on this it sounds like using are using the XLR mic input on the front. You can add 20 dB of attenuation by hitting the pad button on the front panel, this will allow signal levels up to 20 dBu (7.75 V). Without the pad you are limited to 0 dBu (0.775 V).

The XLR mic input has rather low input impedance at 2.8k, this may make your distortion results worse. Dynamic range on this input is also worse than the rear TRS input, although you can use the preamp gain to dial in the level to close to 0 dBFS on the front input.

Here are some very early measurements I made using the UL Mk5 for reference -> https://www.audiosciencereview.com/...-mk5-review-audio-interface.24777/post-841669.

Michael

 

[MCH]

How are you planning to do multichannel I2S on a Pi anyway? As far as I know, this seems to be quite broken? Maybe newer incarnations of the Pi have this working?

I use a HDMI to multichannel DAC that I hacked to extract the i2s signals. Works pretty good up to 192/24 (see picture a few posts back)

You can make a gadget with as many channels as you like. No need to limit to only 2

I was referring to connect the i2s pins of the pi directly to the DAC. The pi can only output 2 channels via i2s as far as I know.

If you use the proper test files, the results should be usable.

Will look into that

In case you have gain issues, you may also try the volume control of the ESS chip.

Yes, this is how I am currently controlling the volume. Works pretty well too.

 

[voodooless]

I use a HDMI to multichannel DAC that I hacked to extract the i2s signals. Works pretty good up to 192/24 (see picture a few posts back)

Can't you just hook that up to your PC?

I was referring to connect the i2s pins of the pi directly to the DAC. The pi can only output 2 channels via i2s as far as I know.

In theory yes, some people have managed to get multichannel sound, but as far as I know, it's not stable. Other SOCs seem to be more suited for this though.

 

[MCH]

Based on this it sounds like using are using the XLR mic input on the front. You can add 20 dB of attenuation by hitting the pad button on the front panel, this will allow signal levels up to 20 dBu (7.75 V). Without the pad you are limited to 0 dBu (0.775 V).

I knew you would know a solution, thanks!

The XLR mic input has rather low input impedance at 2.8k, this may make your distortion results worse. Dynamic range on this input is also worse than the rear TRS input, although you can use the preamp gain to dial in the level to close to 0 dBFS on the front input.

hm, i don't have a cable to do that, but will look into it

Here are some very early measurements I made using the UL Mk5 for reference -> https://www.audiosciencereview.com/...-mk5-review-audio-interface.24777/post-841669.

thanks!

 

[MCH]

Can't you just hook that up to your PC?

my pc only has usb-c :-/

In theory yes, some people have managed to get multichannel sound, but as far as I know, it's not stable. Other SOCs seem to be more suited for this though.

if all this project works out, i might get a minidsp mch streamer or one of those diyinhk usb to tdm boards, but to be honest, i don't really need this, i do it only for fun, so might stick to my cheapo hdmi solution, it works for now.

Thanks all for all the suggestions, i have plenty of homework to do, and have not started yet with the microcontroller and user interface!

So far i am very happy that the distorsion seem to be at ca. -110 dB (close to datasheet), just need to figure out how to measure the noise properly. not bad for a 20 euros BOM multichannel dac. Ah, and it sounds fantastic!!

 

[MCH]

@meooms , the distorsion coeficient at work :

 

[mdsimon2]

I am definitely doing something wrong because if i continue decreasing the volume of the dac the noise figure improves significantly. here at -60 dB:

Sounds like noise floor modulation to me. I observed something similar with the DIYINHK ESS DACs. Seemed power supply related as shorter wiring from the PS to the DAC board always improved things. However, performance dropped off at higher levels and more channels driven. Your noise pattern looks somewhat similar as well -> https://www.audiosciencereview.com/...s/diyinhk-dac-measurements.23629/post-1082824.

Michael

 

[MCH]

Sounds like noise floor modulation to me. I observed something similar with the DIYINHK ESS DACs. Seemed power supply related as shorter wiring from the PS to the DAC board always improved things. However, performance dropped off at higher levels and more channels driven. Your noise pattern looks somewhat similar as well -> https://www.audiosciencereview.com/...s/diyinhk-dac-measurements.23629/post-1082824.

Michael

But in the example you posted what gets worse is THD, noise remains more or less stable... My case is exactly the opposite.

my tests were with all 8 channels driven, btw. Could try only 2 and see what happens.
Now I am trying to set up a zero 2w in gadget mode so that I can run the tests directly from Ree, let's see what happens. I would like to do windows laptop > pi zero gadget > i2s DAC, this last step is what I need to figure out, let's see if I manage.

 

[mdsimon2]

But in the example you posted what gets worse is THD, noise remains more or less stable... My case is exactly the opposite.

It is true that I saw increases in distortion, but I also saw increases in noise.

With the silent switcher + LDO power supply noise went from -113 dB to -108 dB going from 2 channels driven to 8 channels driven. That was after I had optimized the power supply wiring. Before optimizing the wiring, noise was worse than -100 dB with 8 channels driven. More than the numbers your noise floor shape looks suspicious to me.

Just saying I wouldn't discount the difference in -60 dBFS and -15 dBFS noise floor as measurement artifacts as this point, they look real to me.

Michael