Differential Temperature Measurement

[NH3Valves]

I am looking to measure a differential temperature of about 10 degrees F delta at around 0 F ambient with a reasonable amount of accuracy in a lab environment. I need to do this because I am using a single loop controller incapable of accepting multiple inputs and computing a difference. It seems to me that using 2, 2-wire RTD's in parallel (one as cold, one as ref) wired to a 3-wire RTD to 4-20mA converter would work assuming the RTDs are matched in length and type so the reference resistance in each circuit is the same. Would this work? Is there better ways of measuring a delta T?

 

[Lewish]

Jim Williams has done some remarkable work with temperature measuring circuits. Check the app. notes on the Linear Tech website.

Also, Bob Pease has a number of articles about it. Check the National website for his articles and also any available app. notes.

Both men have achieved accuracies of better than 0.1ºC with some rather common components.

 

[IRstuff]

The only temp sensors that can do that on their own would be thermocouples. It would be possible to flip the reference thermocouple and put it in series with the cold thermocouple and the reference should wind up being subtracted from the cold.

BUT, you'll need to watch out for ALL the other contacts for thermal voltages. TTFN

 

[cbarn24050]

Hi 2 pt100 in series, with 2 100 ohm resistors in a bridge circuit.

 

[Iden]

Another possibility is a Radio Frequency Identification (RFID) Tag with a temperature sensor. Identec Solutions Inc of Kelowna, British Columbia, designs and markets such a device. The interval of measurement can be set to >1 read/second, or hours between temperature log. Use a tag for each item, and a PCMCIA scanning device on a handheld or laptop. The scanning device can gather information and identify many tags in a second. Contact; mflahr@identecsolutions.com

 

[hacksaw]

just order matched pt-100's. it only cost a few bucks to have the factory check the cal. with NIST traceable standards. forget "jury-rigging" the rtd transducer. there are special transducer setup for differential temperatures, otherwise use two single input designs.

once you are past the sensor part, you need to deal with the installed accuracy of your meas. associated with sensor placement and conduction errors. you've not indicated what your response time requirments nor your accuracy requirments.