She was being a fickle lady and wasn’t much help which is why I ended up here (though my google skills might have been off off-kilter again…).
That said, I managed to have great overnight tests. In my code for the RN2903 I set pwridx to 5, which based on Microchip’s documentation, that is 20dBm. This is a bit loud, but my real tests are outside, so I’ll want that oomf. Last night I decided to set pwridx to 10, which should be 10dBm and I only had 13 of the errors over an 8 hour period.
I had wondered if maybe the signal was too loud and I was ending up with some slop in adjacent channels making it look like I transmitted twice, but I thought it unlikely because the reported channels of the errors weren’t adjacent (at least consistently) to the actually transmitted channel. I am re-thinking that based on last night’s test, but am not 100% sold yet.
The interesting thing to me is that the RSSI and LoRaSNR values aren’t terribly different between the two tests. They both seems within the range of possibility (of course both are reasonably high values), I just would have expected the values when pwridx was set to 5 would have shown more of a saturation. Who knows…