LM35 "PIC" of the bunch

[PerranOak]

I am having a go at designing a temperature logger using and LM35 and a PIC ... 16F probably.
I have to get the analogue part right first or the digital part will never work; I realise that.
So, the LM35 goes up to 0.35V at 35degC so I amplified it with an op amp so that I would get a better FSD and so more significant figures. I used a very simple non-inverting circuit.
Problem is that the gain on the op amp seems to vary wildly with the input from the LM35.
I set it at 25degC and then cooled it to 20degC and the op amp output suggested a temp that was a whole degree different.
This can only mean that the gain changed!
Is this normal?

[David Hess]

Is the supply voltage to the LM35 +4 volts or higher?  I am not sure what the LM35 does with a supply voltage of +3.0 or +3.5 volts.

You might need an RC filter at the non-inverting input to the operational amplifier.  A feedback capacitor between the output and inverting input would not hurt either.

[PerranOak]

Cheers David.

Yes, the supply is at 4.9V and the LM35 measures accurately. It's only the output from the op amp that varies.

I'll  try your suggestions, thank you.

PS
Ah. I've just noted what the problem is. I had tried to offset the LM35 because the output goes negative below 0degC and this is no good for the uC. However, I did this by placing a resistor between ground (i.e. battery negative) and the ground pin of the LM35. It did offset the output but interferes with the op amp gain!

PPS
Durr. I've just realised why. I am amplifying the resistor offset too! Sorry for sharing the thought process!