Strange Situation

[CrankyCoder]

im using wireless, nrf24l01.

Wondering if possibly there is a weird power fluctuation and it's causing the arduino to reset. or if there is something in the light on/off that is messing with the PIR.

an interesting note, and it just occurred to me. Both the bathroom and kitchen lights, the reason they aren't automated is they are both florescent lights... i wonder if that has anything to do with it.

[dbemowsk]

@crankycoder I assume you have capacitors on your incoming power on your sensors. If not, adding some may help power fluctuations if that is the problem. It is said that fluorescent lights can cause some interference in the 2.4GHz band. If these are simply screw in fluorescent bulbs, maybe change them for standard incandescent bulbs for a period to see if that makes a difference.

[CrankyCoder]

@dbemowsk no caps, definitely worth a shot, i figured since i was powering it off a usb port off another device it would already be cleaner power. but have been considering moving location anyway and changing out to some other power. I was looking at this post (https://forum.mysensors.org/topic/9550/build-a-reliable-power-supply-chain/9) and considering maybe trying something like that.

[dbemowsk]

@crankycoder It's not so much about the power being clean. The caps will give little bursts of power when the attached sensors fire or whatever the sensor does. It prevents the arduino from browning out. I usually put a 10 uf or larger electrolytic and a 0.1 uf ceramic cap on the incoming power regardless of the source. In case you didn't know, always use caps with a voltage rating higher than your supply voltage. If you are pushing 12 volts to the raw pin of your arduino, I would use either 16v or 25v caps.

[zboblamont]

@dbemowsk @CrankyCoder Having re-read the thread, starting/firing of the fluorescent is the likely culprit, ballast initiation emitting RF reacting with some component combo of the Node and accessories, but identifying which and solving it could prove a nightmare compared to removing the emitting interference.
Fluorescents were famed for running efficiently, but often overlooked is their considerable power demand at startup which made them more power hungry for on/off locations.

For CFC bulbs, substitute for incandescent (as suggested), you can swap to LED later.
For tubes LED versions are available but require some rewiring in the case to bypass the ballast.
I switched out 90% of cfc and tubes to LED for energy savings, and have reduced bills to verify it was worth it. ;)

[CrankyCoder]

@dbemowsk oh yeah, for the radio i have a capacitor. Just not for the main power coming in to the board. I have bags of capacitors so this should be an easy test to see if it helps :)

[CrankyCoder]

@zboblamont good to know! I have 2 florescents in my garage that i swapped for LED. These didn't need to bypass the ballast. Might be worth a test to swap those out for the ones in the bathroom (where i see the most interference)

Guess i know what im going to try when i get home!

[zboblamont]

@crankycoder
No idea how the ballast behaves in that circuit configuration, the type used here were single ended power requiring a rewire including bypass of the ballast itself.
Worth a try, the suspect spike should at least reduce in intensity.
The common practice of not earthing the fittings might also reduce the shielding effect of the ballast case, even if it used to be mandatory for bathrooms.

[CrankyCoder]

So a little new info. Last night i had the lights turn off with the door closed, and no one turned on/off the kitchen/bathroom lights. But what did happen, I turned off a little space heater in my room. When i turned it off, the lights off. Finding this very interesting, i grabbed my phone (in the dark) and check the HA status. The motion detection was triggered... ok... so maybe it's not the PIR/repeater that's the issue...

It's the LIGHT MODULE!! There is a good chance that the power fluctuations aren't messing with the repeater itself, but actually resetting the arduino with my lamp relays. I think this weekend i need to pull that out, check the code to see if i am saving state to eeprom and also throw some capacitors on the incoming power to see if that helps.

[zboblamont]

@crankycoder A cause you didn't notice previously?
Mains borne issues should raise questions why other appliances did not cause effects unless in close proximity. Perhaps they did but you didn't observe it unless there? The mains interference just doesn't make sense...
At least you seem to have narrowed down the suspect device, progress...

[CrankyCoder]

The space heater was pretty close in proximity and never noticed it cause... well i live in florida, i only use it like 3 times a year lol.

But, i also know this house (built in the 70s), had someone do the electrical that i have to question many a decision. Some outlets not grounded. There is an outlet in my kitchen that only turns on when the switch for the garbage disposal is running.... i haven't found exactly what device would only be needed when the disposal is running... but it's a mission i am on...

So i think throwing some caps on the input side of things is still worth a shot.

[zboblamont]

@crankycoder Ok on the heater, but what I was getting at was if it is mains borne interference, appliances elsewhere might also trigger the phenomenon but you hadn't noticed as you were not in THAT room....
Sounds like you have some rogue amateur wiring there, for kitchens and bathrooms a ground wire is mandatory for safety I always understood, on a metal ballast case it provides RF shielding.
Perhaps you could shield the board for your Light Module if the caps don't sort it out?