Find a solution for Audio Library and FastLED library to work together

[richclarkson]

As a quick test I added a very simple LED sequence to the audio code I've been working on. It seems that as soon as any information is sent to the LEDs it messes with the Audio, the LEDs still work fine but the audio readings become very erratic and almost strobe like. Possibly something to do with timing or interrupts. From the forums there are plenty of things to try code wise and a few things we can try hardware wise including something like this if it comes to it. First thing i'll do is see if I can recreate the issue with a paired down version of the FFT code to make it easier to see what is going on, i'll comment when i've got it up in the FFT branch.

[richclarkson]

@sfaletti I got a minimal version of the file here so we can test this Audio + FastLED issue, let me know if there is anything in there that doesn't make sense to you. Running on Arduino IDE. Might also be worth submitting to the PJRC forum if I can't make any progress on it today.

[richclarkson]

Going to keep a checklist here of things I have tried that have not worked. So far it seems something in FastLED.show(); is the cause. I will continue to update this list.

• [x] add delay - slightly helps depending on the amount of the delay
• [x] add delay + run serial plotter - same as above but slightly better
• [x] change LED data rate suggested here - no change
• [x] change clock speed of the teensy - no change
• [x] have the LEDs update only once every 100 cycles - no noticable change
• [x] comment out FastLED.show(); - audio works great! but LEDs don't light up
• [x] add #define FASTLED_FORCE_SOFTWARE_SPI 1 as discussed here - audio works great! LEDs work okay but there is some weird redness coming through from the top of the strip when the audio spikes, this is reduced when I add a delay but still evident.

[sfaletti]

Is it worth trying another library? If the delays are working it definitely sounds like a race condition.

[richclarkson]

Yes that's definitely on the list to try. Looking at the OctoWS2811 LED Library, the Adafruit Dotstar Library and the NeoPixelBus library.

[richclarkson]

NeoPixelBus working great on the minimal version of the RMS file. No weird spikes or redness coming through! Definitely something to do with the FastLED library in the FastLED.show(); function. I'm happy to switch over the the NeoPixelBus library but not sure how to bring the Fire2012() code across to work with NeoPixelBus for the Fire Lamp mode (which seems to be a feature people love). @sfaletti Do you think it would be practical (or even possible) to use NeoPixelBus to drive the LEDs during audio reactive stuff then use FastLED to do the non music reactive stuff (lamp modes)?

[richclarkson]

If we do have to move Fire2012 across looks like someone has already looked into it and seems not too bad. Some other elements of FastLED that would need to be brought across: Rainbow function and global brightness setting.

[sfaletti]

How hard would those really be to roll on our own? If we have hsv calculations they shouldn't be too hard to implement.

[richclarkson]

I'm sure that Rainbow will be fine, and global brightness shouldn't be too tricky. Fire2012 is the only one that I'm not sure about. I suppose it's just a case of figuring out what is going on in its functions: qsub8, random8, qadd8, and HeatColor.

[sfaletti]

It might be possible in that case to grab the LED values from FastLED without writing them to the hardware and causing the timer issues. We can then pass them to a working library. I'm saying without knowing if we can get access to the library at that level. It's also not ideal.

Is it possible to change the pins and timers used by FastLED?

[richclarkson]

I'm sure I tried changing the pins but will try again and let you know. However when it comes to the timers I'm not sure which timers to change in FastLED or where to change them.

[richclarkson]

Jeepers I could kick myself. Changing to Pin 2 & 3 and using FastLED worked. Audio coming in great and LEDs are behaving wonderfully.

[richclarkson]

Added a FastLED and RMS_minimal combined file [here] called Animations_with_RMS(https://github.com/richclarkson/Sabre/tree/LED_animations/old_src/Animations_wtih_RMS). It compiles and runs fine in arduino IDE. It only runs a paired down version of the RMS so the Ripple Music Mode has not been tested yet as that needs all 8 FFT channels.

[sfaletti]

Glad that's working. I'm getting the control inversion / dependency injection working now. Hopefully I can knock it out tonight and then tomorrow or Wed we can get the two systems integrated.

Can you test each channel individually? Maybe running the same sound loop over each channel to get a sense of the split might be a decent test. Or try something like this or this.

[richclarkson]

Yes I'll do that tomorrow, I had all 8 channels well with serialprint but would be good to get it into ripple mode and make sure that's working.

[richclarkson]

Yes I can confirm Ripple mode working with 8 channels. I used this file.