GM Coolant sensor

[tom cloud]

>
>     That is what they are for, but wouldn,t it be great to get rid of an 
>     op-amp.  Hence try and reduce the amount of circuitry required.  
>     Something like a simple LCD temp read out (very useful for diff, 
>     gearbox and water in race cars), temp sensor, a small resistor divider 
>     and a panel meter, and its guarranteed to be very reliable, once you 
>     start throwing in op-amps, etc.  you need very steady power supply 
>     rails to avoid inaccurate readings.  So is there a positive temp 
>     coefficient sensor around ? (could always make one if I have to, but 
>     its easier if there's a car that's already using one).
>     
>     
>     Dan        dzorde at aesprodata.com.au 
>
>
>______________________________ Reply Separator
_________________________________
>Subject: Re: GM Coolant sensor
>Author:  diy_efi at coulomb.eng.ohio-state.edu at INTERNET
>Date:    9/18/96 11:19 PM
>
>
>>     Does anyone know if there's a positive curve temp sensor around, ie. 
>>     increasing voltage for increasing temp.  Would make life a lot easier 
>>     for analogue circuitry.
>>     
>>     Dan        dzorde at aesprodata.com.au
>     
>I thought that's what inverting amps were for. 
>Tom

Dan,

The first monolothic op-amp was the uA-709.  It was not frequency
compensated, required offset adjustments, could latch up and was sensitive
to its power supply.  That was about 1967 or so.  Only a few years later
the uA-741 (yes, the venerable 741) came out.  It solved all of those
problems.  The ability of an op-amp to not be affected by its power
supply is called (guess what?) Supply Voltage Rejection Ratio.  For the
741 (remember, it's an OLD op-amp, the new ones are much better) the
SVRR is 77 dB minimum to over 96 dB.  That means that a change of one volt
in the supply will create a change of no more than 1/(10^(77/20)) <= 140
micro-volts.  If your sensor has a range of 0 to 5 volts, then this
represents an error of .00014/5 = .0028%.  A digital meter typically
is only .5% accurate.  An analog meter only 5%.

If you use any electronics, you're gonna have to 'scale' voltages.
If your sensor output is fairly lo-Z an' your meter is hi-Z, you
can maybe use a resistor voltage divider (assuming you're going to
a lower voltage).  If your meter requires any offset voltage (like
maybe your sensor has an output from 1 to 6 volts and your meter works
on 0 to 5 volts) you'll need a transistor circuit or op-amp.  In other
words, you live in an analog world (hell, digital circuits are just
saturated analog amplifiers!!) and you're gonna have to deal wid' it.

If you are inexperienced with operational amplifiers, go to Radio Shack
and buy their $3 book on op-amps.  Actually, they have a several good
beginning books that are incredibly inexpensive.  There is a series by
Forrest Mimms that I recommend highly.  The little book I mentioned
that costs about $3.00 can be bought with a small kit of parts for
probably $10.00 so that one can build some simple circuits.  All
you'd need is a couple of 9-volt transistor batteries and a voltmeter.
Once again, Radio Shack sells a pocket DVM for about $25.

[Note: I do not recommend the purchase of parts from Radio Shack --
in fact long experience dictates strongly against it.  I recommend the
above books to my students, however.  The DVM mentioned is not very
good.  Better to buy one for about $10 more from some place that
carries quality products -- like maybe Digi-Key, Mouser, Newark, etc.]

Now, to answer your real question re: pos. coeff. temp. sensors.  Thermistors
are neg. coeff. but T.I. used to make a product called a 'sensistor' that
was pos. coeff.  (last time I saw one was probably 1972).  Don't have
any idea if they still make it.  There is a company called Omega that
carries all kinds of stuff like that (and produce very good catalogs
and reference sources).  Their number is 800-826-6342.  (But, you'll
still probably have to condition the output -- i.e. use an op-amp!)

Tom