I2S dma errors & pcb suggestions

[Yannick Richter]

Regarding the software i am not able to get it working and spent some time debugging without any results yet:
After one iteration of the mainloop the chip hangs and does not read another sample buffer after the first buffer is processed.
In the serial terminal on every call to the i2s_adc_enable or i2s_adc_disable function it will print a dma error: "I2S: dma error, interrupt status: 0x00002000"

The led matrix is set as a 2x64x64, hub75 enabled, i2s SAMPLEBLOCK = 1024 as default with 2 dma buffers and a sample rate of 44.1kHz.
Increasing the amount of dma buffers to 4 for example will process 3 buffers before it hangs.
The ADC data itself looks valid.

Regarding the PCB design i have a few suggestions of things i noticed while assembling:

If it is intended for the led matrix to be driven by the board i would replace the LDO with a switch mode converter or leave out the power input side completely and expect a regulated 5V input to just pass through.
The expected currents are way too high for a linear regulator and the diode drop will be significant as well resulting in very poor efficiency and high heat at those currents expected by a larger panel. I am completely bypassing the power input section with a fixed 5V PSU.

The analog circuitry should probably be supplied by 3.3V either via a separate LDO or the ESP32 onboard one so it safely clips to 3.3V reducing the need for resistors as there are also a lot of unique resistors and capacitors that may be possible to combine reducing the complexity.

Maybe i misinterpreted some design decisions but thats probably how i would have designed it.

[Mark Donners]

Hi Yannick, Thanks for you feedback, I replied below From: Yannick Richter ***@***.***> Sent: Sunday, 6 March 2022 19:02 To: donnersm/FFT_ESP32_Analyzer ***@***.***> Cc: Subscribed ***@***.***> Subject: [donnersm/FFT_ESP32_Analyzer] I2S dma errors & pcb suggestions (Issue #2) Regarding the software i am not able to get it working and spent some time debugging without any results yet: After one iteration of the mainloop the chip hangs and does not read another sample buffer after the first buffer is processed. In the serial terminal on every call to the i2s_adc_enable or i2s_adc_disable function it will print a dma error: "I2S: dma error, interrupt status: 0x00002000" * If you didn’t change the software drastically, I should work as is…I have a 64*64 2 panel matrix up and running. * I see two options…zip and send me your sketch folder with all files and I will compile and download it to my hardware and see what happens. Last week someone had troubles with the compiler and I’d like to rule that out. * Upload a precompiled version for 2x 64*64 panel, you have to use chrome browser: https://donnersm.github.io/Webflash-FFT/flash.html * The led matrix is set as a 2x64x64, hub75 enabled, i2s SAMPLEBLOCK = 1024 as default with 2 dma buffers and a sample rate of 44.1kHz. Increasing the amount of dma buffers to 4 for example will process 3 buffers before it hangs. The ADC data itself looks valid. After compiling, you can see the lib versions used….are those the same as in my manual? Regarding the PCB design i have a few suggestions of things i noticed while assembling: If it is intended for the led matrix to be driven by the board i would replace the LDO with a switch mode converter or leave out the power input side completely and expect a regulated 5V input to just pass through. The expected currents are way too high for a linear regulator and the diode drop will be significant as well resulting in very poor efficiency and high heat at those currents expected by a larger panel. I am completely bypassing the power input section with a fixed 5V PSU. As it says in the manual, use only for small panel or small amount of ledstrips… And don’t use full brightness……the current delivered by the board is not for larger setup. As I also stated in the manual, I recommend using and external power supply for larger setups. I limited my board with a 4A fuse but I never go near the 4A because I dimmed the brightness of the panels. The analog circuitry should probably be supplied by 3.3V either via a separate LDO or the ESP32 onboard one so it safely clips to 3.3V reducing the need for resistors as there are also a lot of unique resistors and capacitors that may be possible to combine reducing the complexity. There is more to it then that…I also need to compensate for impedance for the line in and I need to amplify the mic. ( short R10 if you mic response it to low) And It has been designed to have a good frequency response within the complete audio band. IS the the best Way? I don’t know….Ask 10 designers to dray a amplifier and you’ll get 10 different designs…So I don’t see it as good or bad but just different. But don’t get me wrong your feedback is highly appreciated. Don’t hesitate to contact my directly if you need help getting the code to run. Using whatsapp we can quickly communicate. If you come this far in building…you don’t want to give up on software 😉 My phone for whatspp: [phone] Regards, Mark Donners Maybe i misinterpreted some design decisions but thats probably how i would have designed it. — Reply to this email directly, view it on GitHub <#2> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/AOOHCEWIFB66HYSHFEI3ZNTU6TXJDANCNFSM5QBNLOSQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub> . You are receiving this because you are subscribed to this thread. <https://github.com/notifications/beacon/AOOHCEULYT22JJV3QFEWRQLU6TXJDA5CNFSM5QBNLOS2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4RJOQGVQ.gif> Message ID: ***@***.*** ***@***.***> >

[Yannick Richter]

Thanks for reminding me to check the library versions, should have tried that before.
Downgrading the hub75 library from 2.0.7 to 2.0.5 actually fixed it so there did something change in the newer versions it seems.

Still need to do some debugging of sampling and FFT later but at least i get adc samples and a smooth fire animation.

I would like to switch to add an alternate mode with wifi support which does not have to run at the same time as the ADC if there are conflicts. Mostly for remotely displaying messages/graphics or possibly even video via a simple protocol.
Would you be interested in such features as well?

Regarding the power supply using 5V for the amplifier section might be helpful to get the best performance. But still i would leave out the LDO for the led supply.

Anyways nice project :) I see that you already put in a lot of work and it has a lot more potential as well.