Thermistor

[JKellyNPFD]

I need a linear circuit for a thermistor that has at least +/- 0.5 Deg C accuracy. Anyone tackle this before?

 

[nbucska]

For what temp. range? How many? any other requirements?


----------------------------
Please read FAQ240-1032
My WEB: <

http://geocities.com/nbucska/>

 

[JKellyNPFD]

Range of -30 to 50 C is more than enough. I'm going to have a total of 6 probes but i can just mirror the circuitry for the rest. I found linear thermistors sold by omega and honeywell but they are expensive and from the schematics they only look like resistor networks. I found some circuits online which are nothing more than a voltage divider but i'm unable to verify there temperature accuracy. Or how well they linearize the curve. Also on some circuits i see that the solution calls for two thermistors along with a resistor network, any ideas what the benefits are to the second?

 

[Skogsgurra]

http://www.capgo.com/Resources/Temperature/Thermistor/Thermistor.html

is one paper that describes linearization with one resistor. It is possible to make that combination even more linear (and less sensitive) with one additional resistor. I used to have an Excel sheet for that on my computer, but that seems to have been a few computers ago.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[nbucska]

Resistors are temperature dependent, too and there are
thermistors with different -- even positive or negative --
temp. coeff. The networks you have found combine these
effects to improve the linearity.

On a single 6 thermistor temp. sensor you may reach .5C accuracy by calibration. The simplest would be using a microcomputer ( OP.AMP with noise filter, A/D converter and calibration table in the uC)

If you think for mass production, I would use some different e.g. resistive sensor . A welldesigned
diode-sensor may only have to be calibrated at two
points ( 0 and 100 C) to satisfy your need -- less
than 5 minutes per assembly.





----------------------------
Please read FAQ240-1032
My WEB: <

http://geocities.com/nbucska/>

 

[cbarn24050]

You might consider PT100 probes instead. Much easier to get accurate and not expensive.

 

[itsmoked]

If you're using a micro why care about linear? There's a company with 0.01C systems - they just use the micro to convert the non linear readings.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[nbucska]

Keith:
I suggested the micro, the OP is not willing to read my FAQ
or disclose any more info.



----------------------------
Please read FAQ240-1032
My WEB: <

http://geocities.com/nbucska/>

 

[Madcow]

Or you could use a micro.

HeHe!!

 

[nbucska]

Jkelly:
Sorry for not answering your question, so
*** YES ***



----------------------------
Please read FAQ240-1032
My WEB: <

http://geocities.com/nbucska/>

 

[itsmoked]

Madcow; Let me add to that; 'processor.



Hopefully the OP will get back to us.

Here's two links (I had time to look up just now, again, as I was reading them just last night!), that discuss sub 10th degree stuff.

http://www.google.com/url?sa=t&ct=res&cd=1&url=http://

http://www.ilxlightwave.com/appnote...HGDeiSNgFWfkRNumg&sig2=XY3wS39dnTpoL86cOiDvNA

http://www.google.com/url?sa=t&ct=res&cd=2&url=http://

http://www.ilxlightwave.com/appnote...gsD__0zblX3oh2TWA&sig2=TTVDEqJnWXBWKjGVDXlx7Q

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[IRstuff]

Resolution and accuracy are two different matters. If you get a thermistor or PRT, you'll need something with about 0.1ºC accuracy just to do the calibration:

http://www.omega.com/Temperature/pdf/CL550_CL551_CL552.pdf

Out of the box accuracy for simple sensors is not much better than 1ºC:

http://www.national.com/pf/LM/LM92.html

TTFN

FAQ731-376

 

[JKellyNPFD]

Thanks guys. I appreciate all your posts. I would like to use the micro except i'm using a PIC and i'm not that fluent in programming them (still learning) which is why i was looking for a hardware solution. I will probably just get the resistance tables from the manufacturer and try and program that into the controller. It seems to be the most accurate and most inexpensive method.

Thanks again.

 

[ScottyUK]

The simplicity and accuracy of National's LM35 precision temperature sensor IC might be tempting.

http://www.national.com/pf/LM/LM35.html

----------------------------------

Sometimes I only open my mouth to swap feet...

 

[IRstuff]

Calibration to 0.5ºC ideally requires at most 0.125ºC accuracy on the calibrator, which runs about $4000, so the cost savings may be mythical.

TTFN

FAQ731-376

 

[nc2bexpress]

JKelly (and anyone):
I've found myself in the same situation of trying to find a simple hardware solution. Unfortunately I haven't found a solution yet.
I'm not trying to hi-jack your thread, nor do I want to start a new one since your issue is very similar. I figure an inquiry on how far along you have come might be more appropriate.

Hope all is well!