said by SmokChsr:I'll admit that I did not test to this level, my application was not that critical. I just put them on the bench and compared to ambient. I imagine if I was doing something really critical I would want to compare them in 2 different solutions to also verify tracking.
My primary concern was not accuracy, but to ensure I didnt have an LM34 that read high up in the panel, versus one that read lower in the storage tank, as that would cause the pump to circulate heated water into the panel, when in fact the panel was at a temperature lower than storage.
I actually pulled the 1 meg resistors out of the design as they caused too much hysteresis, the semi-matched potted LM34s worked out to have a 5 degrees offset which is perfect for this application.
Its crude and simple but will last forever.