I'm developing a rain sensor using an AD7745. I have the converter connected to a series of tracks on the PCB, measuring its the capacitance.
Since I want to get rid rapidly of any rain or humidity to detect reasonably fast when the rain stops, I have added some heaters below the tracks (they are on the same PCB, but not on the same area of the converter).
The problem is that I'm observing a strange dependence on temperature and capacitance read by the converter.
When the PCB is heating, the capacitance read is rising, but when the temperature rise slows down, the capacity begins to slowly decrease.
You can see it in this graph:
The blue line is the capacitance read (truncated to 16 bits, a long story...) and the purple line is the temperature read by the same converter (referenced to the secondary axis).
On the other hand, when the power of the heaters is removed, the capacitance drops fast and remains stable, but sensibly lower to the starting value:
As you can see, at the beginning of the cycle the value read was a bit more than 14000 and at the end a bit more than 12000.
I doubt that 20ºC will cause a physical change in the characteristics of the traces, so my guess is that the problem is on the converter.
Do you have any suggestion?