Cheap ARM MCUs for RBCD audio

[abraxalito]

While I was searching around to see whether other debuggers supported multi-drop, I came across this one : https://black-magic.org/

RP2040 is listed as one device that's supported.

 

[mkc]

As said, even another RPI PICO together with openocd can be used as a debugger.

https://www.digikey.be/en/maker/pro...with-vs-code/470abc7efb07432b82c95f6f67f184c0

 

[miro1360]

@abraxalito I can confirm, that RP2040 works as USB/I2S bridge (16-Bits 48kHz or 44.1kHz or 32kHz) with waveshare firmware. Pitty that guys from waveshare don't share the source code for firmware (they share only compiled file).
https://www.waveshare.com/wiki/Pico-Audio
I didn't find a working USB-I2S C++ or .py code for raspberry pico

 

[abraxalito]

That's very interesting @miro1360 . I wonder how they implemented the clock? I might give it a try on a spare Pico to see what the jitter's like.

 

[PJotr25]

I found it for RP2040 https://github.com/alexstanoev/FoxDAC maybe someone will need it ...

 

[miro1360]

@PJotr25 Very useful solution. The output from pico is SPDIF (not I2S) and using WM8805 is converted to the I2S. WM8805 significantly reduces jitter

 

[miro1360]

That's very interesting @miro1360 . I wonder how they implemented the clock? I might give it a try on a spare Pico to see what the jitter's like.

The clock is calculated from internal PLL, not ideal ... but PJotr25 posted nice solution. Altough WM8805 complicates the connection, but at the same time reduces jitter and brings additional optical inputs

 

[abraxalito]

Well if the solution calls for a WM8805 anyway, CM102 does the job well enough.

 

[miro1360]

CM102 is nice, do you know how much jitter it has? Maybe its I2S can be good enough for R-2R

 

[abraxalito]

Ha I remember I was going to check on its jitter, but never got around to that. Does it matter much when being fed into WM8805?

 

[miro1360]

Its jitter matter only when I2S directly from the CM102 is used (CM102 directly with DAC).
WM8805 has 50pS jitter

 

[abraxalito]

As far as I remember, CM102 doesn't have I2S output, only S/PDIF. Which is why I thought of it when you mentioned the Pico only outputting S/PDIF.

 

[miro1360]

@abraxalito You are correct, I got the CM102 wrong, CM108 has I2S output

 

[abraxalito]

Yes CM108 has it for sure, but its package isn't so easy to solder.

 

[abraxalito]

CM102 is nice, do you know how much jitter it has?

I've found out something about CM102 since we had this discussion. The drivers for it only operate in 16bit mode and my Win7 installation puts out some kind of dither which turns to low-level noise which can be distracting. Changing to an interface with a driver which accepts a 24bit input solves the low-level noise problem. Its a pity as this rather detracts from CM102's usefulness on Windows. I don't know if other OSs put out annoying dither - anyone have any experience?

 

[miro1360]

@abraxalito Interesting information Is it related to the I2S function?
I have no experience with CM108 I2S (do you mean CM102 audio or CM108 I2S?), only with PCM2706 where is no such problem.
There is also CP2615-A02-GM chip which can create asynchronous endpoint ... when the mouser stock it I will route a small PCB and test it ... the only issue is the less standard I2S sample length (48-Bit instead of 64-Bit word), so the gule logic for R-2R DACs must be modified

 

[abraxalito]

Hi miro - I don't think this issue has anything to do with I2S. I believe (though haven't verified) that it will affect CM108 as well as CM102. There might be a way to get around it using some other driver - perhaps ASIO, or some other configuration within Windows. I will need to explore. I think I still have a PCM2706 board somewhere so I can check whether its affected too, my guess would be yes.

I looked at CP2615 a few years ago and rejected it - it doesn't allow 44.1kHz sample rate, only 48kHz. Unless there is an updated version? My DS is from 2019, rev1.2.

 

[bohrok2610]

CP2615 is capable of 44.1kHz but there are other issues. It uses internal 12MHz clock to generate the LRCK so the actual frequency is 44.1176kHz. It only supports implicit asynchronous feedback which is not available in Windows UAC driver. Regarding jitter I would not expect much due to the internal clock.

 

[abraxalito]

Here's an interesting project which I just found out about, open source planar headphones driven by digital EQ from RP2040 : https://www.cnx-software.com/2023/0...aspberry-pi-rp2040-mcu-ti-pcm3060-24-bit-dac/. Its the first application of RP2040 to audio DSP that I've seen. The hardware is pretty mainstream (TI codec, paralleled opamp HPA), I'm wondering what if there was a digital input and the output was current driven?

There is another RP2040 audio DSP application I've since learned about here - https://www.hackster.io/elektroThing/ds-pi-rp2040-audio-dsp-board-471bdc

 

[abraxalito]

Aliexpress currently has a few interesting cheap ($3-ish) M4 evaluation boards - I've mentioned them on my blog already but I figure they're relevant here too.

First up is the Arterytek AT32F403, a 240MHz M4 - https://www.aliexpress.com/item/1005004842376803.html. Not the most frugal in terms of current draw but delivering a lot of MIPs/$.

STM32G431, a 170MHz M4 I've mentioned before has become more affordable (depending on what discount Ali decides to offer you, I'm getting 75% for some reason) - https://www.aliexpress.com/item/1005005303669884.html

The best M4 in terms of power efficiency currently looks to be the HDSC HC32F460 at 200MHz and under 20mA : https://www.aliexpress.com/item/1005003651371848.html. Its English datasheet I've attached, the reference manual (Rev1.5) is too large so here's a link to the manufacturer's page : https://www.xhsc.com.cn/Productlist/info.aspx?itemid=1850. You need click on the second tab of four near the top to find the list of docs for download.