Control Engineering

Peter Welander

Peter:

 

I enjoyed reading your April 19, 2007 article: Thermocouple vs. RTD vs. thermistor.

 

I have seen a similar article comparing thermocouples, RTD's and thermistors by many different authors two or three times a year for many years now in various trade journals.  I call it "The Temperature Sensor Article".   It usually is quite general in nature, and because of this generality it sometimes contains incorrect or misleading information.

 

Regarding thermocouples, it is well established that the voltage signal is not generated by the junction.  The voltage signal is a function of the difference in temperature between the two ends of the wires.  And in fact the voltage signal is produced sequentially along the thermocouple wires in the regions where there is a temperature gradient.  This is important to understand in some applications.

 

To say that RTD's are more accurate and more stable than thermocouples is generally true, but not always.  Here at GEC Instruments we have developed proprietary technology to enable highly accurate temperature measurements with thermocouples.  We recently shipped a 56 channel thermocouple system to a major aerospace company where they needed high accuracy in point measurements within a thin wall heat exchanger and with 0.062" diameter probes.  They could not get this with an RTD system because the smallest probe diameter was too large and it was not a point measurement.

In some instances, thermocouples can be more stable that RTD's.  We know of an application where type S thermocouples are being used to make highly accurate temperature measurements at very high temperatures in steam turbines.  These thermocouples replaced RTD systems because the RTD's were drifting too much due to vibration.  The type S thermocouples are much more stable in this application. 

 

With regards to RTD's you say: "At some temperatures, the reference voltage can actually heat the sensor and throw it off."  This is true for all resistance temperature sensors, including thermistors.  There is self heating of the sensor due to power generation from current flowing through the resistor.  The key to accurate measurements is to use a very low current to minimize this effect.

 

Now on to thermistors.  All thermistors are not created equal.  That is the danger of a general statement about stability in an article such as this.  Until I began working with some special stable thermistors in 1988, I believed that thermistors were not stable -- because I had read over and over in various reincarnations of "The Temperature Sensor Article" that thermistors are not stable.  Here at GEC Instruments we use special stable glass bead thermistors that can be more stable than any RTD sensor or thermocouple.  These special stable thermistors are not subject to drift from shock or vibration, as are RTD sensors.  There are companies that sell very expensive thermistor probes with an accuracy of 0.001 °C and a guaranteed stability of 0.002 °C per year.

 

I am writing this to express my belief that generalities in articles such as this can sometimes do more harm that good.  They can lead readers to believe that thermocouple systems are not accurate when in fact some thermocouple systems can be very accurate.  And many years ago I read in "The Temperature Sensor Article" that thermistors are not stable.  This made me shy away from thermistors when I needed accurate measurements.  Later I found out that some thermistors are very stable.  You are perpetuating a myth to state that thermistors have the lowest stability in the group.  Again, all thermistors are not created equal.  And all thermocouples are not created equal.  And all RTD's are not created equal.  And the accuracy often has more to do with the calibration and with the readout device than with the sensor itself.  

 

Jerry Gaffney
Chief Engineer, GEC Instruments

Thanks Jerry. Your points are well made. I can't claim to be the most original thinker in the world. My only response is that my initial advice in the article was that a user should attempt to choose first by process requirements and second by technology:

"If you know what these [process requirements] need to be for your application, they can be the basis for your decision and you won't have to worry about measurement technology. When you find a device that does what you want and has a transmitter to convert the measurement into 4-20 mA or digital output, there is probably little reason for you to concern yourself with whether it's a thermocouple or RTD."