AT-09goes dark after two days, then won't connect

I happily installed the first part of my sprinkler controller. It can run either programs or take commands.

Since I'm still working with the sprinklers themselves, I'm just sending bluetooth messages to turn on for a couple of seconds for the moment (and doing that from a phone in my pocket beats the daylights out of hollering at someone to control the valves!).

After about two days, I went out and the AT-09 no longer appeared in the lightblue app on my phone. I checked with my iPad, and it didn't show there, either. It was still blinking it's LED in the ready state, though.

I reset the Arduino, and it still didn't appear, so I power-cycled the whole thing.

It then appeared, made connection once before aborting, and then would break connection while negotiating!

The MCU is a nano clone (3.3v) powered off of a an iPhone supply through the usb. The AT-09 draws power from the 3.3v pin of the MCU.

And now that I think of it . . .a day or two before deployment, I was still powering the AT-09 off an output pin (which has its own issues with parasitic draw). I saw it lose connection repeatedly after getting the command and the MCU turning on a single led from another output pin. [hmm . . . maybe this is why so many people misuse "loose" for "lose" . . . autocorrect is underlying "lose" in red . . . maybe it's renegade autocorrection . . ]

Anyway, the maximum draw is supposed to be 8.5ma, so well within the current capacity of the pin, and the single LED is drawing a similar amount, and there's nothing else drawing.

I switched it to run off the 3.3v pin instead of the output pin, and it seemed OK, and ran for the two days.

I've swapped in another AT-09 which previously stayed up for a couple of weeks in my hothouse.

I could use any ideas here . . . losing bluetooth isn't critical, as long as the MCU keeps running the program (it will run briefly every hour for the new seed to keep the soil moist).

I don't like not knowing why things like this aren't working, though. A lawn can die in a couple of days of no water if t happens in august. The bluetooth isn't about keeping it running, but so that it can report in when it runs.

hawk

Is any of your stuff out in the weather where dew or other moisture can get onto it?

Paul

no, not this round--it was in my garage, with the window closed.

And the one that ran in the hothouse did so in extremely high humidity (high enough that water was condensing and dripping!)

It's possible that if you're running something off a pin that it can suddenly require a large amount of current. This could give the MCU problems and make it go into an unstable state. Have you confirmed the chip is okay and the system is working properly after a reset? If not, you might need to reflash the bootloader and try to recover the system. In general, I'd recommend powering devices using a FET switch controlled by a pin, rather than directly off a pin.