ASR Open Source Streamer Project

[Skeptischism]

since the alwinner only supports 192kHz IIRC, its probably 8 channels TDM @ 48kHz

 

[phofman]

https://linux-sunxi.org/images/b/b4/Allwinner_A64_User_Manual_V1.1.pdf page 621 - 8 slot TDM, max. 192kHz samplerate. Allwinner may not be the best long-term option to pick, its support looks pretty sub par https://linux-sunxi.org/A64#GPL_Violations , plus the A64 is rather outdated today.

 

[Skeptischism]

Oh its definitely not a long term solution; just what I have onhand until the ROCK64 gets here.

 

[buz]

https://linux-sunxi.org/images/b/b4/Allwinner_A64_User_Manual_V1.1.pdf page 621 - 8 slot TDM, max. 192kHz samplerate. Allwinner may not be the best long-term option to pick, its support looks pretty sub par https://linux-sunxi.org/A64#GPL_Violations , plus the A64 is rather outdated today.

Allwinner is probably the least friendly of the Chinese media SoCs. It seems that Google is a force for good there, the chromebook vendors seem to be a lot more compliant than they used to be, even gasp MTK

 

[voodooless]

The RockPro64 would be even better probably. It also offers 4 I2S data lines but has quite a bit more CPU power. It also takes NVMe SSD's. No wifi by default though, which is a bit of a letdown.

 

[notabenem]

I was analyzing the ROckPro64 for more than an hour, but I could not find the 4x I2S lines. Where have you seen that?

 

[voodooless]

I was analyzing the ROckPro64 for more than an hour, but I could not find the 4x I2S lines. Where have you seen that?

Yeah, I can imagine you didn't find it I had a hard time as well. But here is the pinout of the 40-pin connector. While the labels are somewhat obscure and contain typo's, it clearly shows 4 SDO lines.

 

[somebodyelse]

I was analyzing the ROckPro64 for more than an hour, but I could not find the 4x I2S lines. Where have you seen that?

https://files.pine64.org/doc/rockpro64/rockpro64_v21-SCH.pdf is another place to find the details - the Pi-2 Bus is on page 29.

 

[phofman]

Yeah, it's weird that Pine64 does not publish the Pi2 pinout for their RockPro64. I found only a proposed table at https://forum.pine64.org/showthread.php?tid=5710&pid=37612#pid37612 The older Rock64 has the documentation complete. Sad and makes one cautious...

Edit: somebodyelse - great catch, thanks!

 

[notabenem]

That's a great find. That means we have two quite powerful solutions at our disposal, one even with a PCIe slot. And it's supported until 2023. That's one extra year more than the Rock64.
Now let's hope those actually work and are stable.

 

[somebodyelse]

Yeah, it's weird that Pine64 does not publish the Pi2 pinout for their RockPro64. I found only a proposed table at https://forum.pine64.org/showthread.php?tid=5710&pid=37612#pid37612 The older Rock64 has the documentation complete. Sad and makes one cautious...

Edit: somebodyelse - great catch, thanks!

Pine64 are a bit of an odd company - the software and much of the documentation in the wiki are community generated rather than company produced. The forum post @voodooless linked has a better image of the pinout than the wiki, but nobody has updated it. There's sometimes better info in forum posts than in the wiki, or in the docs of projects and individuals using the boards. Several of their products are closer to an organized group buy than a traditional product - preorders for the next batch, with an estimated delivery date, available until that batch sells out, then an indeterminate wait until the next run. A multichannel DAC compatible with their boards may pique their interest.

 

[buz]

A multichannel DAC compatible with their boards may pique their interest.

I was going to say they might well be convinced to manufacture such a thing. Possibly even an assembled streamer including it...

 

[Skeptischism]

The RockPro64 would be even better probably. It also offers 4 I2S data lines but has quite a bit more CPU power. It also takes NVMe SSD's. No wifi by default though, which is a bit of a letdown.

Rockpro64 is what I mean and what i'm getting, sorry. Too many similar named products in my head. I found the pinout fairly easily for the pro IIRC, it was the A64 I had to find somewhere else ... lol. https://forum.pine64.org/attachment.php?aid=1318

 

[Skeptischism]

and this perhaps. think this was for the rockpro64

 

[notabenem]

Interestingly, from a KODI point of view, the Rock64 boards (and PINE64 in general) do not have the best reputation: Android API changes, bits and pieces that do not work reliably with the given linux kernel, video decoding quirks, etc. This may not bother us too much though for a headless streamer.

 

[Skeptischism]

I was going to say they might well be convinced to manufacture such a thing. Possibly even an assembled streamer including it...

i'm not sure I would trust them to assembel the streamer for us, let alone a multichannel dac and it will still need some work from us regardless i'm sure. I mean sure, an RPI compute module 4 type footprint could tempt me.

 

[Skeptischism]

and while I'm not one to go overboard on linear PSU for a streamer, I would certainly look into higher quality SMPS with low stray fields and lower noise than standard for my own board. one thing that could be cool would be if they could provide isolators. i'll be doing isolated i2s, with master clock to send back to it, but chatting with them about suggested isolation for gpio could be interesting. since its GPIO, going to some troublethere will pay some dividends just keeping noise out of our dac.

 

[phofman]

IMO when designing the DAC board from scratch it would be a pitty to leave out the I2S/I2C isolators, they are quite inexpensive. Plus they could break the potential ground loops should the streamer be connected via USB to some PC/NTB or monitor + class I amps (+ class II PSU for the DAC).

Any plans for the I2S clock circuit?

 

[Skeptischism]

well for my uses, yes, definitely isolated, but I was planning on that as a separate sub board, with any additional MCU. either including the clock, or possibly driven by a clock located on the dac board, with its power supply there too and sent back over SMA to the clean side of the reclocking. (not sure on the trade offs there, to be continued after investigation)

MCLK over SMA, the rest probably FPC or u.fl. not planning any kind of additional fifo, as I cant live with the latency; I want ability to play all video sources including streams. I know they can be delayed to match the audio with a bit of work, but i'd rather do without and just try and follow best practices for FF reclock, with minimal delay and allow using one set of filters for video and one, with more intense FIR, for music only where delay isnt a big deal. I want to divide and send to ADC as well, to allow use as a measurement rig.

as far as specifics for the XO/reclock circuit, no, not yet, thats the next module i'm working on after the first round of dac boards I send off. initially both boards will have placement for the XO or XO's and only after the isolation, isolated power supply and control hardware is looking more finalized, will I decide on whether i'm doing 3 boards, or 2, with slotted PCBs and isolated islands.

So for this i2s input DAC, I guess that would mean the difference between 2 boards and 1, rather than 3 boards or 2 for my wider project.

 

[Skeptischism]

But I will probably ease off on the specifics in public, until I have something more concrete in my hands. Happy to discuss with you in PM @phofman , just the wider project is not something I really want to document on a public forum ahead of time.