🚀 go-pugleaf

"Antony N. Lord" <antony_lordX@hotmail.com> wrote in message news:<antony_lordX-383E5A.22443202012003@mec2.bigpond.net.au>...
> >   Actually, the LM35 and variants are more accurate than most commercial
> > thermometers; one has to shell out a fairly good chunk of change to get
> > a calibrated thermometer as good as or better than these ICs.
> >   No calculations, lines of best fit, or any other trash like that.
> 
> I suppose that is true - but surely there has to be calculations / 
> adjustments made due to voltage loss etc. in the cable run between the 
> sensor and the interface.
> 
> More importantly - any clue as to bringing the maximum output voltage 
> (at the upper measuring temperature point) closer to the 4095mV level.
> 
> It seems pointless not to be using the "full scale" of the input board.
> 

To get the output closer to FS of 4095mV, one often uses an amplifier.
These can be had for a few mere dollars. Try Linear Tech for suitable
devices and app notes. You will need some offset control too.

Now also remember that once you add noise, offset, gain, cable loss
etc. from the system, you will need to recalibrate the sensor. The
accuracy can probably be kept by measuring the mV output from the
direct from the LM35 and then measuring the corresponding output from
your amplifer (or at the end of the cable run). This will need to be
done at at your lowest temp, your highest temp, and preferably
somewhere in between. Then using some basic calculus, you should be
able to derive an equation. IT may also pay to check for linearity, by
testing more than three points.