• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

4 channel DIY digital interface.... with an Amazon Basics HDMI extractor????

ok, after a few unexpected issues, i have the 4 channels up and running!!

here is the final result (unless for the power supply filtering caps that i have yet to solder). I made it to fit on top of the audio extractor:

1692113264172.png

not my best soldering job....

here is the pin tapping job, it was as difficult as i expected, but managed to get it done:

1692113949527.png


As explained before, i left the cables long and messy on purpose to test how robust the i2s connection is. As is, it goes to 96kHz/24bits but not higher than that. I know wm8804 has a different mode for 176 and 192kHz so it might well be that i need to do something else to access that mode. It is my first time working in hardware mode with this chip so i am not 100% sure if i am doing it correctly.

I also mentioned that i had experienced channel swapping, that of course is a no go for multichannel applications. I went to ask to diyaudio and thanks to @HenrikEnquist and @phofman i could find out that the channel issues happened when there were buffer underruns at the source. Seems that it is not clear why this happens, it could be a device (the extractor) issue, a driver issue... in any case, using camilladsp is very easy to avoid buffer underruns at the cost of a little latency, so i set the parameters that i always use in my main system (chunksize 4096, target level 4096) and i have been now running it 24+ hours without issues. Btw, if anyone is using a raspberry pi to output audio via HDMI to an AVR and has experienced such issues, i think they would like to know.

In parallel, as the parts i needed took very long to arrive, i have been working on a second hdmi -> spdif project, but this time with 8 channels (4 via toslink+2Aes/ebu+2 to chose coax or AES/ebu) and much easier to solder. More to come!

1692114311454.png
 
Last edited:
Short update:
I have a workaround for the channel swap issue. It is a bit too long to explain step by step here but if someone is doing multichannel via HDMI with CamillaDSP and experiences it, just let me know and I will explain it.
 
Hi, I want to share a reflection with all of you: if our end goal is reconstruct a great analog signal and simply control the volume in the Analog domain,
why we do not leverage the fact that HDMI give to us DSD and with a low pass filter we have the Analog Signal?

We can do that with a simply Low Pass Filter (LPF) see this RT Audio Design Pure DSD Converter. This Amazon HDMI Extractor give already to us the analog signal, but (I guess) that the LPF used is pretty basic based on this Amazon Basics 4K HDMI Extractor Review.

If that is true, I think we can build a really low cost HDMI to DSD to Analog Signal simply using a better LPF (see here examples). Let me know what to you think!
 
Wake me up when you have any measurements of these solutions that are at least halfway decent.
Hi, have you already tried this approach or simply you do not like any unconventional solution?

Pure DSD Converter
  • Digital input : USB only DSD
  • Audio files: SF and DFF
  • Resolution:
  1. DSD Native DSD64 (2.8224MHz and 3.0720MHz)
  2. DSD128 (5.6448MHz and 6.1440MHz) and
  3. DSD256 (11.2896MHz and 12.2880MHz) on Linux and MAC OS
  • Digital output: SPDIF only PCM mode 24bit/192Khz
  • Analog outputs :1 XLR (5Vpp) 1 RCA(2,5Vpp)
  • Phase shift: <0,2 degrees
  • Frequency response: 10 Hz -70 kHz +/-0.5 dB @ 0 dBU
I think that if we obtain some similar from an Amazon Basics HDMI Extractor plus a better LPF could be a good deal for many of us
 
Last edited:
Hi, have you already tried this approach or simply you do not like any unconventional solution?
I don't see the point. The performance will be bad, so why bother?

simpler != better
 
@voodooless
This got me thinking
Just purely theoretically: we can convert PCM to DSD in real-time with Jriver
We output that via USB (or spdif) to a device that could convert that to analog
(not the one above for 2000 EUR but one with realistic pricing or even DIY)

The PCM-DSD conversion in Jriver I guess is audibly transparent and if the DSD-analog conversion is done in a way that it is realistically priced (or even cheap) and measures excellently...

Would that be nonsense?
 
Ik will probably work, but it will need very careful handling of the analog electronics. So until I see measurements confirming that this performs well, I'm not that interested.
 
Looking around, I found this project STAR Pure DSD DAC able to convert the DSD signal to analog without noise and without a DAC chip. The measurement are quite interesting:

DSC3_Int_12MHz.jpg
Thanks for digging that up!

Yeah, so as you can see, this isn't very impressive. 74 SINAD and very poor SNR :rolleyes:. We're not even approaching Redbook.
 
Thanks for digging that up!

Yeah, so as you can see, this isn't very impressive. 74 SINAD and very poor SNR :rolleyes:. We're not even approaching Redbook.
I also did some digging
Apparently most of the DACs just process DSD by converting it to PCM.....ESS DAC chips for example do this
However there are other chips (like AKM and Rohm) that have a direct DSD mode

See for example the Topping E30 II with dual AK4493S chip

image.png


The signal path is much shorter with DSD signal if direct mode is enabled (vs PCM or vs DSD without direct mode)
(having no volume control is not an issue since volume can be controlled on the PC in Jriver for example)

This DAC was measured here on ASR too - I am just wondering if it was also measured with DSD input signal in DSD direct mode, would there be a measurable difference?
FWIW I have sent an email to Topping to ask them about the same

This is getting interesting.... :)
 
Last edited:
What’s the point of a multilevel delta sigma modulator if you’re not going to use it? Yes it works, and performance isn’t even bad as far as I can see, but why? Just use the PCM path, that’s what the chips were designed for.
 
to be honest, no particular reason, I am just trying to broaden my horizon and my knowledge on things
this topic just simply got my interest; I see potential in it
Like using Jriver's real-time DSD encoding capability and then having a DAC that has a much shorter signal path (vs PCM) - to me this is intriguing
I might be totally wrong though.... one day I might get an E30 II and just listen to it if I hear any difference :)
 
Hi ppataki, based on my knowledge, one of the best DSD Pure amplifier is the SM-SX. In attach the full documentation with the schema
 

Attachments

  • dxsx1h.pdf
    4 MB · Views: 93
Hi ppataki, based on my knowledge, one of the best DSD Pure amplifier is the SM-SX. In attach the full documentation with the schema
Thanks @666Andromeda this looks really interesting!
I wonder why this concept was abandoned...

According to Wikipedia, DSD is actually a PDM signal which actually is a generalized version of PWM

Since PWM can be amplified (see Class D amps), theoretically speaking a DSD amp would be 'easy' to design
 
I DIYed a HDMI to 4-group AES/EBU box, which supports one HDMI input, one HDMI output, and 4 AES/EBU outputs (7.1 channels)View attachment 340303
Saw it yesterday for the first time. Is it your project? Cool and congrats.

Question for you. Are you using an Explore transceiver chip? If yes, have you tried it on Linux? Have you experienced channels swapping?

Thanks for sharing your project.
 
Saw it yesterday for the first time. Is it your project? Cool and congrats.

Question for you. Are you using an Explore transceiver chip? If yes, have you tried it on Linux? Have you experienced channels swapping?

Thanks for sharing your project.
Yes, ADI's Explore transceiver chip is used. It has not been tested on Linux-based PCs. The tested device types are roughly as follows: Windows-based PC hosts, Raspberry Pi 4, hard disk players, Blu-ray players, etc. , no abnormality occurred when switching channels
 
Would be cool to have the same using USB input and four SPDIF optical outs (like the RME Digiface but I guess much cheaper and maybe even better)
 
Back
Top Bottom