I am using the following code in a photon: volatile unsigned long h = 0; volatile unsigned long t = 0; volatile unsigned long t_last = 0; volatile unsigned long hist[32]; void ISR (void) { unsigned long now = micros(); hist[h] = now - t_last; t_last = now; h = (h + 1) & 31; } voi…

I’m thinking it’s because micros() does not wrap at an unsigned long value like millis() does. It wraps at 59,652,323 microseconds, so you need to explicitly take care of wrapping with micros. https://docs.particle.io/reference/firmware/photon/#micros-

Note that the wrapping only applies to 0.4.3 and earlier. Later versions, micros() will increment through the full 32-bit value space. noInterrupts() only disables registered interrupts for pins - system interrupts are still enabled. If you want to disable all interrupts, use an ATOMIC_BLOCK()

I discovered that original issue and posted it here. I’m aware that recent versions don’t have that issue. Thanks. The interrupt is off of a GPIO pin (D2), so that really should do the job.

I added ATOMIC_BLOCK() { } around all the sections where I had noInterrupts(); … interrupts(); I still see the same behavior. Anyone have any ideas? Also, something interesting, in hex the value is one of three: 0xffffffc2a, 0xfffffc27 or oxfffffc31. I find that odd.

@mgssnr , can you post your setup() code?

I think this one deals with the same [image] Sporadic value from micros() Troubleshooting Hey everyone, Just starting with a Photon. I am trying to measure pulse times via a low triggered interrupt. I think it’s a pretty simple handler. I started getting occasional rando…

Thanks for the response. The relevant bits of the setup() code would be that I’m setting the ISR to trigger on the falling edge of D2. I’ll post it this evening.

Yes, I saw this, and it looks similar. I didn’t see that the other post had disabled interrupts, and I thought perhaps it was somehow different. For what it’s worth, I think the implementation of micros() is the problem. It would be better to have a timer tick away at the proper rate and micros()…

Yup, this was how I thought micros() worked by just using system ticks to calculate a decadic fraction of a second. And I hoped @mdma would come back to the other thread …

In 0.6.0 millis() and micros() will have a thread-safe block to prevent inconsistent computations, and millis() will tolerate having the SysTick interrupt stalled.

Bizarre behavior between ISR and loop()

Troubleshooting

ScruffR April 19, 2016, 6:16am 7

I think this one deals with the same

Topic		Replies	Views
Sporadic value from micros() Troubleshooting	16	3528	April 19, 2016
Millis() goes backwards inside ISR after woken by pin Troubleshooting	6	1027	December 16, 2018
Micros() rolling over after 59.652323 seconds, not 71 minutes [SOLVED] Troubleshooting	47	14397	August 18, 2015
Millis() and micros() lose time when interrupts are disabled Troubleshooting	17	6516	October 2, 2015
delayMicroseconds() safe in interrupt? Troubleshooting	17	5407	April 12, 2015

Bizarre behavior between ISR and loop()

Related topics