ProtoDAC TDA1387 X8 project

[Tim Curtis]

Here is another option for installing the yml file. Just tested on my end and no issues.

Code:

# Install file
wget -q "https://moodeaudio.org/test/ProtoDAC VC.yml.zip"
unzip "ProtoDAC VC.yml.zip"
sudo cp "ProtoDAC VC.yml" /usr/share/camilladsp/configs/
chmod 0666 "/usr/share/camilladsp/configs/ProtoDAC VC.yml"

# Confirm format and permissions
pi@moode:~ $ file "/usr/share/camilladsp/configs/ProtoDAC VC.yml"
/usr/share/camilladsp/configs/ProtoDAC VC.yml: ASCII textpi

@moode:~ $ ls -l "/usr/share/camilladsp/configs/ProtoDAC VC.yml"
-rw-rw-rw- 1 root root 1188 Jun 15 07:18 '/usr/share/camilladsp/configs/ProtoDAC VC.yml'

After setting the config in CamillaDSP but before playback confirm the volume proxy is running

Code:

pi@moode:~ $ sudo systemctl status mpd2cdspvolume.service 
● mpd2cdspvolume.service - Synchronize MPD volume to CamillaDSP
     Loaded: loaded (/lib/systemd/system/mpd2cdspvolume.service; disabled; vendor preset: enabled)
     Active: active (running) since Thu 2023-06-15 07:18:44 EDT; 11min ago
   Main PID: 154438 (python)
      Tasks: 1 (limit: 1009)
        CPU: 1.134s
     CGroup: /system.slice/mpd2cdspvolume.service
             └─154438 python /usr/local/bin/mpd2cdspvolume --pid_file /var/run/mpd2cdspvol/mpd2cdspvol.pid --volume_state_file /var/lib/cdsp/camilladsp_volume_state --config>

Jun 15 07:18:44 moode systemd[1]: Started Synchronize MPD volume to CamillaDSP.

Be aware that this config results in -3dB lower volume to prevent clipping so when turning it on/off there will be difference in loudness i.e., Software volume will be louder.

You can also see the Camilla volume in action and examine the filter by opening the Pipeline editor.

ETA: Also turn on "Display dB volume" in the MPD options section of Audio Config.

[Skip Pack]

That procedure worked nicely. Slight differences in some of the replies, but all went to completion and it's now running with the CDSP volume control and sounds good. I've had issues over the years getting stuff into the Moode filesystem from my Windows machine, so picking the file up with wget was nice.

Am I correct in assuming that the dithering done in this case makes the difference whenever 24 or 32 bit values are sent to the ProtoDAC? 

Appreciate your help very much.

Skip

[Tim Curtis]

That procedure worked nicely. Slight differences in some of the replies, but all went to completion and it's now running with the CDSP volume control and sounds good. I've had issues over the years getting stuff into the Moode filesystem from my Windows machine, so picking the file up with wget was nice.

Am I correct in assuming that the dithering done in this case makes the difference whenever 24 or 32 bit values are sent to the ProtoDAC? 

Appreciate your help very much.

Skip

Dithering is only applied to 16 bit samples because at that bit depth it can help reduce any noise in the lowest bit thats correlated to the audio signal and possibly audible.

For 24 or 32 bit samples dithering is unnecessary because any noise in the lowest bit would be at -144dB or lower and thats way lower than the noise floor of any audio device and way way way lower than the threshold of human hearing.

ETA: The TDA1387 chips used in the ProtoDAC are 16-bit DAC's and thus they simply truncate > 16 bit samples. An option is to set Camilla resampling or MPD/SoX resampling to specify 16-bit and let these components do the truncation/conversion instead. I've tested this on the few > 16 bit albums in my collection and can't hear any difference so I go for simplicity and just let the DAC chips do it.

Dithering
https://www.soundguys.com/what-is-dither-23700/

[hifinet]

Recently it  was suggested I might consider the ProtoDAC kit as a replacement for my Allo Piano DAC with reclocker rig. Would prefer to have it in a case but I have not seen pics of completed DAC in a case. If anyone has case options please post. Much appreciated.

That procedure worked nicely. Slight differences in some of the replies, but all went to completion and it's now running with the CDSP volume control and sounds good. I've had issues over the years getting stuff into the Moode filesystem from my Windows machine, so picking the file up with wget was nice.

Am I correct in assuming that the dithering done in this case makes the difference whenever 24 or 32 bit values are sent to the ProtoDAC? 

Appreciate your help very much.

Skip

Dithering is only applied to 16 bit samples because at that bit depth it can help reduce any noise in the lowest bit thats correlated to the audio signal and possibly audible.

For 24 or 32 bit samples dithering is unnecessary because any noise in the lowest bit would be at -144dB or lower and thats way lower than the noise floor of any audio device and way way way lower than the threshold of human hearing.

ETA: The TDA1387 chips used in the ProtoDAC are 16-bit DAC's and thus they simply truncate > 16 bit samples. An option is to set Camilla resampling or MPD/SoX resampling to specify 16-bit and let these components do the truncation/conversion instead. I've tested this on the few > 16 bit albums in my collection and can't hear any difference so I go for simplicity and just let the DAC chips do it.

Dithering
https://www.soundguys.com/what-is-dither-23700/

There should be an audible difference between converting from 24 bits to 16 bits by simple truncation compared to 24 to 16 bit conversion with dither (preferred).

[michaelagiles]

Recently it  was suggested I might consider the ProtoDAC kit as a replacement for my Allo Piano DAC with reclocker rig. Would prefer to have it in a case but I have not seen pics of completed DAC in a case. If anyone has case options please post. Much appreciated.

That procedure worked nicely. Slight differences in some of the replies, but all went to completion and it's now running with the CDSP volume control and sounds good. I've had issues over the years getting stuff into the Moode filesystem from my Windows machine, so picking the file up with wget was nice.

Am I correct in assuming that the dithering done in this case makes the difference whenever 24 or 32 bit values are sent to the ProtoDAC? 

Appreciate your help very much.

Skip

Dithering is only applied to 16 bit samples because at that bit depth it can help reduce any noise in the lowest bit thats correlated to the audio signal and possibly audible.

For 24 or 32 bit samples dithering is unnecessary because any noise in the lowest bit would be at -144dB or lower and thats way lower than the noise floor of any audio device and way way way lower than the threshold of human hearing.

ETA: The TDA1387 chips used in the ProtoDAC are 16-bit DAC's and thus they simply truncate > 16 bit samples. An option is to set Camilla resampling or MPD/SoX resampling to specify 16-bit and let these components do the truncation/conversion instead. I've tested this on the few > 16 bit albums in my collection and can't hear any difference so I go for simplicity and just let the DAC chips do it.

Dithering
https://www.soundguys.com/what-is-dither-23700/

There should be an audible difference between converting from 24 bits to 16 bits by simple truncation compared to 24 to 16 bit conversion with dither (preferred).

I am working on getting up to speed on camilla to try this but in the meantime I wondered if using SoX todo the 16 bit conversion would also have the advantage of adding dither? The SoX documentation I looked at suggested that dither needed to be specified.  Does the implementation in moode incorporate dither in this conversion?
I seem to hear a difference if I use SoX vs just letting the DAC truncate but "better "?
Thanks again to hifinet!

[Tim Curtis]

As far as dither goes, MPD's Software volume control does it. No need for SoX to do dithering.
https://github.com/MusicPlayerDaemon/MPD...Volume.cxx

Alternatively the CamillaDSP config that I posted performs dither.

If u want to use the CamillaDSP config just follow the instructions in the earlier post
https://moodeaudio.org/forum/showthread....5#pid47205

[michaelagiles]

As far as dither goes, MPD's Software volume control does it. No need for SoX to do dithering.
https://github.com/MusicPlayerDaemon/MPD...Volume.cxx

Alternatively the CamillaDSP config that I posted performs dither.

If u want to use the CamillaDSP config just follow the instructions in the earlier post
https://moodeaudio.org/forum/showthread....5#pid47205

Thanks Tim!
I am not using the MPD volume control so that doesn't apply.
But I wanted to try down sampling to 16 bits rather than just letting the DAC truncate. So I am trying to use SoX to do this and hopefully SoX will also add dither this process.
Will experiment with Camilla when I have time .

[hifinet]

I am working on getting up to speed on camilla to try this but in the meantime I wondered if using SoX todo the 16 bit conversion would also have the advantage of adding dither? The SoX documentation I looked at suggested that dither needed to be specified.  Does the implementation in moode incorporate dither in this conversion?
I seem to hear a difference if I use SoX vs just letting the DAC truncate but "better "?
Thanks again to hifinet!

I have yet to compare truncating 24 bit recordings to dithering using Moode software volume, SoX or Camilla.

The (presumably) dithered CD (16/44.1) Classic Records from the HDAD set sounds better than the 24/192 file truncated with ProtoDAC. To my ear, the CD sounds more realistic in terms of the sound of the instruments in the hall.

I also discovered this when editing and converting ввс AAC files of live radio broadcasts. It was the first time using Audacity. Initially, I imported the AAC file into Audacity, and edited the sections I wanted, and then exported to flac. Audacity converts from AAC to 32 FP for editing. I didn't realize that it wasn't dithering the file when it was exported to 16/48 flac. I went on for a few weeks this way. The edited files seemed to be a bit disappointing compared to a file of the entire program converted to flac using Foobar2000. 

I finally realized that the file wasn't properly dithered when exported to flac by Audacity. Using the Foobar2000 converted AAC to flac files imported into Audacity, 16/48 is maintained and apparently the dithering or whatever secret sauce ввс is using. The edited files then sound identical to the flac file of the entire program. It is amazing that the 16th bit can have such an effect. 

Trucation is loss of information, since the 8 LSBs are simply thrown away. Dithering uses the information in the 8 LSBs to correct the 16th bit. Subjectively, it is subtle, but most noticeable in terms of the realism of the sound of acoustic instruments in a hall.

I hear a similar effect with the electrolytic power supply decoupling caps. The small Panasonic FR caps sound best to my ear (looking forward to hearing the tiny Panasonic NHG). The image and soundstage is very sharp and well defined. But, they are less forgiving than the other larger caps, which seem to sound a bit softer. The Elna Silmic II sound really good, except that they distort the image and soundstage. Some instruments are just too big and out of proportion. This could be a good thing for some types of music.

[hifinet]

I received a question regarding power. 

ProtoDAC receives 5VDC power from the Raspberry Pi GPIO by default. 

It is possible to have ProtoDAC powered from a separate supply by removing pins 2 and 4 from the GPIO header for ProtoDAC. This is essentially irreversible once the header is soldered in place. Power would then be applied to the 5V and GND pads between the RCA jacks.

[TheOldPresbyope]

I received a question regarding power. 

ProtoDAC receives 5VDC power from the Raspberry Pi GPIO by default. 

It is possible to have ProtoDAC powered from a separate supply by removing pins 2 and 4 from the GPIO header for ProtoDAC. This is essentially irreversible once the header is soldered in place. Power would then be applied to the 5V and GND pads between the RCA jacks.

Or perhaps do a little surgery on a so-called stacking header (example) for a reversible solution?

Regards,
Kent