These type of sensors often have a warm-up time, often measured in minutes. The datasheet will tell you. If you don't let it warm up you will get poor readings.

@NeverDie Yes, you are correct - it is based on ATMega1284P. So not relevant My point is that I'd rather stick to the datasheet rather than overclock, but that's just me. And yes I realise that Moteino is overclocking, but the size, voltage and flash are really appealing to me.

I decided to fix the broken trace I so naively destroyed while breaking the power LED out with a screwdriver. The image from @tripy really helped to figure out which trace I actually broke. It helped a lot to desolder the voltage regulator as well, leaving me with pads I can solder to. I left the regulator out after the repair since I am running the thing from 5V USB. With the regulator out of the way the correct pad was easily identified. Maybe someone else finds this useful, definitly saved my kitchen lighting project today

@hek Thanks for the comments. Sounds good. I'll report back when I have some progress.

hello, i know it is late, but others could see it too. i had the same problem running small (LOW POWER) modules on power banks, because the battery or power bank has a built in power meter, while it is charging and consuming power, there will be power on the USB port, but the arduino's consumption is very tiny regarding charging a phone. you need to disable this feature somehow.

Yes i made several reset, exclusions, inclusions, vera reset and clear EPROM but nothing worked... I made a new sensor and it works perfectly Strange... I will try again when i will have more time. Thanks for the help

@jonnyfishman said: would it be better to trigger the callback manually to pick up a response? No that won't help (as long as you call gw.process() ).

Of cource it's powered down - but is his done by a sw-command via SPI (and also the pwr-up) or by transistor ? I'm surprised about the quality of the rNF - the power down seems very good?