Just to share a surprising outcome. My unit has been running, indoor, continuously for six and a half years. A couple of years back readings began to diverge. They post now consistently more than 20% apart.
Plantower rates PMS5003 max consistency error at ±10%, with a minimum of ±10 µg/m3, so with two sensors a 10 minutes average bias of 20% is the maximum I am accepting before remedial action.
Also bias is consistent i.e. A is always posting that higher than B. Thus I am not sure bias qualifies within the definition of max consistency error above. As it is not random.
Anyway finally decided to take the plunge and following the excellent videos available disassembled the unit, an early iteration judging from the assembly materials, and the sensors themselves. I then gave them a nice wipe with a soft cotton cloth. Not very dusty but some at the fans, sensor conduit and in front of the exhaust holes.
Proud of myself re-assembled everything, re-connected the unit and…surprise. Bias is still there. Unchanged as far as I can tell.
I am not sure about the cause. Fans may be spinning at different speeds, skewing readings, or one laser optic may be dirty, may be subtle since haven’t noticed, or anything else.
I still trust readings, bias is not very large yet, but I thought may be interesting to share.