LM35 - reading garbage.

jremington:
That would depend on the context of the manufacturer's recommendations, which in this case are intended for sensors operating in a noisy (e.g. industrial) environment.

No! It depends on what is actually contained in the datasheet included by the manufacturer, which they place there to inform their customers, cover their legal liabilities and protect their reputation to keep the stockholders happy.

The total discussion in the datasheet I cited re: 'noisy environments' was this on page 19:

In noisy environments, TI recommends adding a 0.1 μF from V+ to GND to bypass the power
supply voltage. Larger capacitances maybe required and are dependent on the power-supply noise.

That is TI's recommendation regarding environmental noise. Of course, that presupposes proper attention is given to ALL layout considerations including interconnectivity, which IS DISPLAYED on page 16 showing the Star connections, twisted pair power leads and the shielding where needed dependent upon configuration, to which I noted initially!

If one consults their Analog Engineer's Handbook regarding buffering low level instrumentation signals, one can see that signal conditioning is an efficacious practice as I mentioned originally, also.

I doubt I'll be bothering to TRY to help others on this site again given the "welcome" I've received from some of the more senior members.

Have a good day!