>The resistance of a thermistor at any temperature can be calculated using
>something known as the Steinhart-Hart constants, as long as the material of
>the thermistor is known.  I've been playing with a PID temperature
>controller based on a 16C63 for the last month and I knocked up and Excel
>spreadsheet with the formulae on it and a load of constants for different
>materials.  If you found the resistance at say 0 and 100 C you may be able
>to figure out the material and calculate resistance for any arbitrary
>temperature.  Drop me a mail if you'd like a copy of the spreadsheet.


Yes, if you can send me the spread sheet you may bsaving me a lot of time,
and I will greatly appreciate it.

>
>Probably the best way would be to borrow/hire a thermocouple based
>temperature meter and bond a thermocouple very close, or onto the original
>thermistor based sensor.  Start the engine and as it warms take spot values
>every few degrees.  This would give you a curve tailored for your
particular
>sensor, as I strongly suspect that standard automotive temperature sensors
>are not very accurate.


I tried something similar with a digital thermometer, but found out that the
thermal mass makes a huge difference on the response time, and I would have
to let the temperature settle for a long time before making any readings.

>
>Having said that, you say it has a computer attached?  I'm assuming this is
>for fuel injection?  Quite a few fuel injected cars have two sensors, one
>standard cheapo one to drive the nasty little temperature gauge and one
>rather more accurate for the ECU.  Seems wierd they don't just use one but
>there you go.  The cheapo standard ones generaly have one terminal and rely
>on being grounded through the engine block, the ECU sensors have two
>terminals and are isolated from the block.


I am using the ECM sensor, it has two isolated terminals that are connected
directly to the computer.

>
>If you take the voltage from the standard sensor, beware of the voltage
>regulation system used by the manufacturer.  Certainly in older cars a
>hideous switching regulator based on a bi-metalic strip and a heater was
>used, giving a switching frequency of a few hertz and causing all kinds of
>headaches.  A modern fuel injected car probably has something a little more
>sophisticated.


I suppose this one is acceptable. I«ve run a few tests and so far it looks
ok.

>
>Of course to avoid all these problems, just use a standalone thermistor or
>thermocouple.  I've just had a thought, somewhere I've seen little
>temperature modules with LCD display and remote thermistor sensor that also
>output temperature data as a serial data stream.  They were pretty cheap as
>well.  Over here in the UK I think Maplin may well sell them.  Worth a
>thought as all the linearisation and interfacing has been done for you.


I going to search for them. And if you remeber where to find them, let me
know.

Thanks

Calvin

>
>Have fun
>
>Mike Rigby-Jones
>mrjones@nortel.co.uk
>