Frozen cyan light

I found this topic dealing with the same problem but I am not using delay() inside the Intervaltimer callback. My code runs for several hours and then locks up with a frozen cyan light and the core becomes unresponsive - need to push the reset button to get it back. Are there maybe any other rules with Intervaltimer to be aware of?

@MarkusL, can you share your code so we can advise?

@peekay123: here is the code (zip archive of Spark Dev folder). I freezes at about an hour runtime - the serial communication over USB also stops working.

I tested disabling the timer but the core still freezes - it just takes longer now.

Can you send us the amount of Flash & SRAM your application is using? Did you reduce the spark variable string lengths by any? If so how much did you reduce it by?

If you were able to free up memory, then I can look at your code a bit deeper, if the CC3000 gets into a weird state then your loop may be blocked for up to 10 minutes before it starts to recover on its own.

@mtnscott: I did reduce array lengths but not all of them. I am now using 2x 622, 1x 200 and 1x 64.
How do I measure Flash & SRAM usage?

So this is strange: when I run my program on the Spark without being connected to anything it runs fine. If I plug it into my hardware it freezes after some time.

How are you powering the Spark in your hardware? What is your communications I2C? SPI? with the rest of your hardware?

I am powering via 12V to 6V DC-DC converter to Vin. I am using OneWire to read temperature via the ‘spark-dallas-temperature’ lib. I am not doing I2C or SPI.
The rest is just analog/digital I/O. I am using 2 RISING interrupts (pinMode = INPUT_PULLDOWN) but there is currently no signal coming in because it’s not yet hooked up to 120VAC

Do you have an oscilloscope? Can you look at the 6V you are putting on VIn under full load of the Spark Core?

Have you enabled the interrupts? If so I would check to make sure that the pins are not left floating.

I have no oscilloscope but I have run a program that simply dumps all analog input to serial and that runs with no trouble at all.
Only thing that I can try now is backtracking function-by-function.