ASR Open Source Streamer Project

[mdsimon2]

most ess implementations will be. any board that uses a single crystal and accepts 44.1 and 48khz will be async. usually they will be 50MHz or 100Mhz etc. vs dual 22.xx/24.xxMHz or 45.xx/49.xxMHz

I have a few of these for I2S testing, as @Skeptischism mentioned they have a single 50 MHz clock and run async. They are super cheap and work OK. The input header is really sad though (only one ground pin, not even a separate one for power). Although they have a MCLK pin it is not required as an input because it runs async. It also looks like they are out of stock although I bet you could find something very similar on ali express.

https://www.audiophonics.fr/en/dac-...-v21-i2s-to-analogue-24bit-192khz-p-8396.html

The DAC linked below is very similar to the one above but a bit more expensive and can act as a RPi HAT.

https://www.audiophonics.fr/en/dac-...c-es9023-raspberry-pi-a-b-20-i2s-p-11215.html

Audiophonics also still has the old miniDSP miniDAC8 in stock which has been discontinued by miniDSP. This would give you an 8 channel I2S DAC.

https://www.audiophonics.fr/en/dsp-...8-channel-24bit-192khz-akm-ak4440-p-9989.html

Michael

 

[roog]

Dear all,

I am really interested in this thread, the opening post being very encouraging, as the idea of building a DIY streaming device that can support DSP is high on my list of things that I should do.

However, not being very computer literate I was hoping that this thread might start to distil out a preferred solution which I might be able to implement.

Do we have a consensus on what this thing should be? Perhaps I missed it, nestled in amongst the many pages of 'finely tuned' geekery. :0)

 

[voodooless]

Do we have a consensus on what this thing should be?

If this thread shows one thing, it's the total lack of consensus on anything

 

[roog]

Oh, that's a pity!

Is there a front runner for the main item of hardware?

Come on you clever people. :0)

 

[voodooless]

Is there a front runner for the main item of hardware?

Originally @amirm had assigned project duties to @StefaanE, but we haven't heard from him in about a month. Could be that our failure to consent on anything scared him off I guess we can't blame him for it

 

[Skeptischism]

Yeah, I didnt know how to answer that ... ha!! indeed, there is not a lot of consensus here. In their defense, that is the case for the greater space we are exploring, with the hardware and software lacking standards, because they are being defined as we go in the embedded hardware and low power computing and DSP landscape. the companies involved are typically startups or hobbiests and the hardware and software architecture getting trickle-down from more profitable IP areas, often without a full team and budget for development. What they/we are trying to do, is on the edge and typically the realm of product development, rather than a band of enthusiasts. I cant claim to be knowledgeable enough in the software area to be a leader in that area, but know enough with how it interacts with the hardware to work with that person, should a set of design parameters be reached.

I will be making some hardware that should be suitable, but whether or not these guys have outlined exactly what that is by then, remains to be seen (joking). I dont have the budget of a large company to throw at development either, it is being funded out of pocket, so my development schedule has to match my budget and time constraints (has to work around my day job) that I have set for my own hardware.

as it happens i've just started a new job with a firm that does a lot of embedded IoT software and hardware development, so I may be able to leverage some resources there too. As mentioned previously, I guess i'll just keep popping my head in here. I started into the xmos and clocking module schematic last night and discovered XMOS chips are now out of stock just about everywhere and only a few chips left that are suitable for multichannel USB->i2s/adat/etc ... arrrgh. everything else is out of stock, due back in december!!! First AKM, then ESS parts were out of stck, now XMOS? seriously? all the manufacturers are hoarding!

 

[voodooless]

seriously? all the manufacturers are hoarding!

Yes, it’s basically all parts that are very difficult to get these days. Waiting times of 30 to 50 weeks tend to be very normal. Some cars might even have to stop production because they can’t get a $5 chip for it.

 

[Skeptischism]

yes, my mate linked me that story about the car production. it reminds me to use parts with second source where possible.

 

[Skeptischism]

also some atmatel MCUs ...

 

[roog]

I have read the stories concerning ICs being in short supply and I can see how this might impact the availability of assemblies.
Just to clarify my understanding, is it the purpose of this project to make use of 'generally available' assemblies and software to create an effective streamer and not to develop one from scratch?

 

[somebodyelse]

I have read the stories concerning ICs being in short supply and I can see how this might impact the availability of assemblies.
Just to clarify my understanding, is it the purpose of this project to make use of 'generally available' assemblies and software to create an effective streamer and not to develop one from scratch?

My take is that almost all the goals are already covered now that Moode has integrated CamillaDSP. The troublesome part is the active crossover as we haven't found many/any good low cost multichannel DACs that reliably play well with linux, hence the discussion about making one. If you already have a suitable USB DAC like the Okto then moode can do the crossover using its current GUI.

 

[Perra71]

Just an idea as I'm using the formula Z8 as a streamer.

Running Spotify from the Z8 and music from my Nas (Synology audio station) all going through my formula Z8 -> toslink -> cheapo DaC -> RCA -> preamp.
HDMI 2.0a CEC to tv
Only issues i have is i need a display to select Spotify/mp3s/wave files.

Maybe modding the z8 might be easier than going from scratch.
As you do have access to Google play store through the z8.

Just an thought.

https://www.formuler.tv/z8

 

[Skeptischism]

modding it to be what exactly? does it support multichannel via anything other than USB? if not, it has less features than moOde running on an rpi

 

[roog]

Thank you @somebodyelse, your explanation is very helpful.

 

[buz]

As you do have access to Google play store through the z8.

Android is hopeless as a foundation for this type of undertaking. Even Windows would work better.

 

[Skeptischism]

Android is hopeless as a foundation for this type of undertaking. Even Windows would work better.

utterly

 

[notabenem]

I am reviewing options and came across the FocusRite Scarlet 18i20 USB DAC (97DB SINAD, 102dB dynamic range) for 530EUR (Measurements here) . Has more than enough channels for this project and the performance is 'ok'. or there is the Clarett 8Pre with better performance (103dB SINAD, 118dB dynamic range) for ~770 EUR at which point we're getting into shooting range of the Okto Dac8. Both available on the market. There are also the Presonus Studio models, Motu ( in comparative price ranges and performance and I guess many others.

@Skeptischism What are the target parameters for your own project (technical and financial)?

 

[notabenem]

The motu UltraLite AVB can be had for 580 EUR with SINAD ~110dB and dynamic range ~120dB.

 

[Skeptischism]

It would depend on how many were being bought. I'm not a large company that has the ability to make 1000 pieces and then sell, so it would have to work out how many were being sold and price from there, as well as find a DC power supply i'm happy to bundle with it. As a result I will have to redesign the preregulator stage, which in my builds uses transformers, but for this, I would suggest a plug pack and DC-DC convertor to supply +/-VDC. then its easy for anyone to install in their case, without working with line voltage. I started a cut back layout for this project and will tally the BOM and work that out once i'm nearing completion, or at least have all the parts of the schematic finished. At that point I can work out pricing vs production run and see how it compares. if balanced output only, with no SE output suits, that cuts cost as well. the boards i'm working on for it is only having 8 x balanced out at this point. as for performance, using ES9028PRO, I would expect to meet datasheet spec, or thereabouts. I may take a small hit depending on what compromises are made.

So that gives 8 channel, up to 352/384kHz PCM (dual clocks, synchronous clocking)

~129dB DNR,
-120dB THD+N (@1kHz),
119.5 SINAD
Noise (RMS)0.5268 uV
ENOB19.56 bits

(above numbers for 8 channel, thus slightly worse than possible with 9028 in a stereo or dual mono design)

 

[AudioGuruJax]

I use Danville products and they are amazing best sound quality DSP processor I have heard so far way better than DEQX.
https://danvillesignal.com/products

Look at this:

https://danvillesignal.com/dspnexus-dsp-audio-processor