I have a PA-II sensor. I realize that it generates heat and that the reported temperatures are reduced by 8° F to compensate. I see two problems with that:
A. I suspect that compensation is only appropriate for some given “average” temperature, and that extremes temperatures, say 0° and 100°, it is much less correct. (I am still trying to deduce how much and in which direction I think the compensation should be adjusted at either extreme.)
B. After any power failure the reported temperature is still compensated downwards even though the unit is cold when power is restored. My sensor reported 12° at the time of power loss and 5° three hours later when power was restored. The actual ambient temperature did not materially fluctuate during that period.
To resolve these issues you might consider upgrading your units to include a remote temperature probe that can be mounted outside and away from the sensor.