ASR Open Source Streamer Project

[voodooless]

This thing actually comes quite close conceptualy: https://a.aliexpress.com/_mN72BcT .

 

[Skeptischism]

hehe, wow, sure ... I suppose.

clock distribution over long traces on 2 layer board, through cheap molex copy pin headers. no local clocks on the dac cards, just on the one central DSP card .... soooo far away from the dac, no local regulation on any of the cards either from what I can see, just 2 basic LM317 type regs on the main board. that thing would radiate RFI like you wouldn't believe. I can hear ground loops just looking at it. plus at the same time, it seems unnecessarily complex. you could fit all of that functionality and more on a single 100 x 100mm PCB. only on aliexpress could you find something so unnecessarily complex, for so little money and designed so badly. exactly the sort of junk I wont compete with I wouldnt recommend sending that to Amir for testing

I know you aren't suggesting it as a real alternative mate, just conceptually, i'm smiling here, so please dont take offense, none is intended.

I totally feel you though. i'm sure you could cobble together a solution from stuff around the web, but it will be a matter of finding something that has good quality software support and interface for the DSP too. oh and doesnt release the magic smoke in short order.

Just because you can, doesnt mean you should

 

[voodooless]

I wouldnt recommend sending that to Amir for testing

I think it would actually be a valuable lesson not to buy such things unseen and untested.

I know you aren't suggesting it as a real alternative mate, just conceptually, i'm smiling here, so please dont take offense, none is intended.

None taken . I’m pretty adapt at smelling BS if I see it. So indeed, it was meant as an illustration, not as an actual recommendation.

Personally, I’d rather have a single good quality night integrated solution. But since I’m not the only one participating in this topic, and needs are rather diverging, I thought to explore some of the possible options.

What could work however: ESS USB option for 2 channel operation (It’s already included), and an extension board with more channels, that could be optionally used.

 

[Skeptischism]

Yep

I think it would actually be a valuable lesson not to buy such things unseen and untested.



None taken . I’m pretty adapt at smelling BS if I see it. So indeed, it was meant as an illustration, not as an actual recommendation.

Personally, I’d rather have a single good quality night integrated solution. But since I’m not the only one participating in this topic, and needs are rather diverging, I thought to explore some of the possible options.

What could work however: ESS USB option for 2 channel operation (It’s already included), and an extension board with more channels, that could be optionally used.

Fair enough.

yes, that could work, but as mentioned, would make it a closed project, all under NDA and even the registers would likely have to be obfuscated by any control software and would never be able to be included in any linux builds/packages. that's if they would even sell less than 1000 of them to us. I dont know, maybe someone in the EU could shoot ismosys an email about it? they seem to be more friendly than Shaw (north america and Australasia)

You would think it was an air to air missile targeting system the way they carry on.

 

[voodooless]

There might be several ways to get a hardware project of the ground. From ASR Kickstarter to maybe a joint effort with a company that already has similar products and has access to ESS resources. Audiophonics comes to mind, but there might be others.

But I’ll keep repeating myself some more: before we do anything we should first agree on the actual project goals. Up until now, it’s still a lot of ideas all over the place.

 

[Skeptischism]

Yeah I 100% agree you guys need to do that. This was my attempt to stir that up. Maybe i'll be finished my xmos before you have that outline . there is no guarantee the ESS USB PHY will even work with Rpi. might be easier to get the non MQA version as well ES***80 vs ES***81.

Anyway, i'll pop my head in every now and then.
Good luck herding cats

 

[somebodyelse]

That was a Kickstarter project a while ago. I think it only does 48kHz.

From what I remember looking into it before, it also has a nasty habit of losing sync and rotating channels - your sub channel suddenly appearing on the tweeter may not end well.

 

[mdsimon2]

Interesting. I had seen the 9080Q on the ESS website recently and had wondered about the cost. The fact that the evaluation board is so much less than the 9026/9028/9038 pro 8 channel evaluation boards bodes well for a decently priced 8 channel commercial offering (although the spec’d performance is not as good as those chips).

It says that it is currently back ordered (checked Mouser and Shaw but did not see it there). It also looks like you may need to establish a business account to order from Ismosys. If it pops up at Mouser/Shaw for a similar price I will buy one as a learning exercise.

Michael

Update on this, Shaw wants $750 for the evaluation board (over double the Ismosys price!) and Ismosys won't sell to someone in the US.

Michael

 

[Skeptischism]

Yup, if ESS cared about how they were perceived in the wider community, they would get rid of Shaw. $750USD for that eval board of a <$5 chip that has so little on it and average performance .... wow, they're having a laugh. the ismosys price is already too expensive for what it is theres nothing to it, thats the whole point of the chip; there's nothing to its implementation

 

[gingerbeer]

...

But I’ll keep repeating myself some more: before we do anything we should first agree on the actual project goals. Up until now, it’s still a lot of ideas all over the place.

Agree completely. Going back to Amir's opening post, I've had a go at summarising what has unfolded over the last 16 pages, for the 2 channel and 8 channel options.

TLDR:

• 2 channel is largely solved with Moode/CamillaDSP and existing hardware that can be done for low cost yet high SINAD, high functionality.
• multi (8) channel is not easily available - high cost, multiple boxes. ASR streamer has the opportunity to offer something new to the market.

1. Built on a low cost, readily available hardware platform like RPi. Ethernet and Wifi support of course.

• 2-Channel. Multiple commercial products available. Single/multiple boxes of varying complexity, can be low cost with high SINAD and high functionality.
• 8-Channel. Nothing readily available, no single box elegant solution available at any price. High SINAD Solutions are high cost and clunky, if even available (DEQX $$$; Okto Dac Pro is not available to purchase; overpriced ESS evaluation boards are being considered as potential solutions, etc.) This is where I see the ASR Streamer offering something new to the market, and where I vote for our efforts to focus.

2. End-point support for Roon and Airplay. You call can suggest others.

• 2-Channel - achieved, Moode, Volumio and others
• 8-Channel - Sort of achieved - DSP (e.g.#4 below) will convert a 2 channel stream into >2 outputs. Native support for multi channel audio from a source I don't think is available (other than JRiver or Roon? not an expert)

3. Ability to download parametric EQ or FIR filters for speaker and room EQ. Integration with other tools should be provided for automatic Room EQ.

• 2-Channel - CamillaDSP has ability for downloading pre-establshed profiles for headphones etc. Downloading REW or FIR developed elsewhere is possible. Developing them from the RPi itself is more awkward, as running what is a graphical tool from a potentially headless Rpi or similar is not straight forward. There isn't a "plug in mic and press calibrate" wizard
• 8-Channel - Same achievement/limitations as 2 channel.

4. Support for active speakers using software crossovers and multiple external DACs. Support for sub-woofer outputs using independent DAC or whatever is in RPi if it is any good. If not, some kind of low cost HAT DAC.

• 2-Channel - Not Applicable / Not achieved as multiple USB DACs will drift over time.
• 8-Channel - This would be relatively easy to achieve, as CamillaDSP and some ALSA tutorials would make this very accessible.

5. GUI in a web-browser for configuration of above.

• 2-Channel - achieved in Moode with inbuilt Camilla DSP
• 8-Channel - achieved, same as 2 channel

6. Support of optional display for cool animations/VU meters. I like explore very low latency implementation of this to keep better sync with music.

• 2-Channel - I believe Moode has some work being done on external screens etc.
• 8-Channel - same as 2 channel.

7. Cloud/automatic firmware update service with proper security fixes and implementation.

• 2-Channel - achieved, Moode can self update.
• 8-Channel - achieved, same as 2 channel

8. Availability of a complete solution in a case ready to go (sans DACs of course).

• 2-Channel - achieved, multiple solutions available (with or without DACs)
• 8-Channel - no complete solution available.

9. No need for internal UI to browse music and such. All of that will be driven by the client app elsewhere. Indeed the device should run headless out of the box. Per #5, web UI should be there for configuration.

• 2-Channel - achieved via Moode, Volumio, etc
• 8-Channel - same as 2 channel for 2 channel audio. As above, not fully familiar with the state of play for native multichannel audio.

 

[notabenem]

So what are our options for connecting >4 DAC channels to a processing platform (be it R-PI, x86, or else)? It looks like

• I2S on a R-PI may or may not be an option (see #289)
• multiple 2-channel DAC through USB is not an option because of clock drift.
• N-channel DAC through a single USB is an option (basically any external DAC offering more than N channels)
• miniDSP USB->i2S. May also work. Can do 8channels with 48kHz (https://www.minidsp.com/images/documents/Product Brief - USBstreamer.pdf)
• PCI bus: easy on x86 SBCs, but also possible on a R-PI (by replacing the USB host controller). when using the compute board 4 with its PCI Express slot on the IO board

EDIT: corrected I2S.

 

[Skeptischism]

I didnt say i2s on rpi wasnt an option ....

 

[Skeptischism]

it does need clarification from developers for more than 4 channels. Not sure on who to contact for that. ahh, actually, is @phofman here on this forum? he is probably one of the people to ask

 

[somebodyelse]

So what are our options for connecting >4 DAC channels to a processing platform (be it R-PI, x86, or else)? It looks like

• multiple 2-channel through USB is not an option because of clock drift.
• I2S on a R-PI is not an option (see #289)
• N-channel DAC with a single USB is an option (basically any external DAC offering more than N channels)
• miniDSP USB->i2S. May also work. Can do 8channels with 48kHz (https://www.minidsp.com/images/documents/Product Brief - USBstreamer.pdf)
• PCI bus: easy on x86 SBCs, but also possible on a R-PI (by replacing the USB host controller)

For PCIe from a Pi you don't need to resort to hardware mods - just get a Compute Module 4 IO board which has a standard socket.
USB to multiple spdif outputs is also an option, although uncommon and strangely expensive. Might be a viable Tone1 mod.
USB->I2S/TDM on linux SBCs may be about to get more reliable and able to work with Windows - patches may just need a rebase to be accepted, tested so far on Pi and BeagleBone Black which can do TDM (and incidentally has an AVB-capable network port...)
I guess we're ignoring Dante/Ravenna/AES67/AVB because of cost of compatible hardware. GentooPlayer has Ravenna output support.

 

[Skeptischism]

Yup, without designing our own baseboard for rpi4 compute module, the I/O board would be the best way forward. Dante would at the minimum appear to need some glue logic to adapt the non standard i2s output. a thread on DIYA not long ago tried to use the output with some AKM dacs and gave up after a while of trying. I mean its an option, but seems to me dealing with yet another codec and process in the way, when all we want is i2s. https://www.ebay.com.au/itm/Audinate-Dante-Brooklyn-II-BNII-01-008-/202904683745

 

[Skeptischism]

Its a pity BBB didnt garner as much support as RPI. IMO it's a far better platform for audio

 

[somebodyelse]

1. Built on a low cost, readily available hardware platform like RPi. Ethernet and Wifi support of course.

• 2-Channel. Multiple commercial products available. Single/multiple boxes of varying complexity, can be low cost with high SINAD and high functionality.
• 8-Channel. Nothing readily available, no single box elegant solution available at any price. High SINAD Solutions are high cost and clunky, if even available (DEQX $$$; Okto Dac Pro is not available to purchase; overpriced ESS evaluation boards are being considered as potential solutions, etc.) This is where I see the ASR Streamer offering something new to the market, and where I vote for our efforts to focus.

If you think that's viable I'd love to see it. If it was pin compatible with FreeDSP hardware it might interest a wider range of people.

3. Ability to download parametric EQ or FIR filters for speaker and room EQ. Integration with other tools should be provided for automatic Room EQ.

• 2-Channel - CamillaDSP has ability for downloading pre-establshed profiles for headphones etc. Downloading REW or FIR developed elsewhere is possible. Developing them from the RPi itself is more awkward, as running what is a graphical tool from a potentially headless Rpi or similar is not straight forward. There isn't a "plug in mic and press calibrate" wizard
• 8-Channel - Same achievement/limitations as 2 channel.

From what I've seen it's possible but not yet simple, although it's quite fast moving so I may have missed something. Integrating AutoEQ would make life simpler, as would having a repository for the various proposed EQ presets from this site (amirm's settings for speakers and headphones, Maiky76's 2 settings for most speakers etc.) and making it easy for others to host them. Making the room EQ process easier with something like drc-fir might be worthwhile too.

 

[somebodyelse]

Its a pity BBB didnt garner as much support as RPI. IMO it's a far better platform for audio

I guess people don't look past the relatively poor headline cpu/memory specs to the useful IO options. Could its PRUs be used for spdif output, similar to the Pi Pico's PIO? I know next to nothing of them beyond their presence, and that MachineKit uses them for step/direction pulses to stepper drivers.

I had hoped Pine64's upcoming Quartz64 board might expose the RK3566's multiple spdif interfaces, but it looks like those pins are needed for other purposes.

Dante would at the minimum appear to need some glue logic to adapt the non standard i2s output. a thread on DIYA not long ago tried to use the output with some AKM dacs and gave up after a while of trying. I mean its an option, but seems to me dealing with yet another codec and process in the way, when all we want is i2s. https://www.ebay.com.au/itm/Audinate-Dante-Brooklyn-II-BNII-01-008-/202904683745

I hadn't even considered Dante from that side - I was thinking of sending from the SBC over the network to Dante-enabled DACs which seem to be either expensive or not perform too well.

 

[voodooless]

Making the room EQ process easier with something like drc-fir might be worthwhile too.

Fully agree with this!! Something free to rival with DIRAC would be nice: a fully working solution that gives you a UI to do simple automatic room correction (with contour options) via your web browser with multi-location measurements, would be a fantastic problem to solve. Features could include multiple sub integration, a possibility for a hybrid solution (IIR + FIR) like DIRAC (lower delay for video).

 

[phofman]

it does need clarification from developers for more than 4 channels. N

This is a VERY informative thread on the multichannel vs. RPi I2S https://www.raspberrypi.org/forums/viewtopic.php?f=107&t=193550 , details of the Audioinjector Octo card discussed too. IMO the most interesting part starts with https://www.raspberrypi.org/forums/viewtopic.php?f=107&t=193550&start=25#p1223301 .