TC-08 and NTC Thermistors

Post any questions you may have about our current range of USB data loggers
Post Reply
peter.coster
Newbie
Posts: 0
Joined: Mon Oct 30, 2017 4:22 pm

TC-08 and NTC Thermistors

Post by peter.coster »

Hello All

I am new to the forum but not so new to the PICO TC-08. Years of using them in the traditional sense with thermocouples and more recently the 4-20mA adapter for flow rate and temperature readings which worked really well.

My next project is to log resistances of 10k NTC sensors and see how they react to temperature (different first order and beta characteristics) and i am not sure how to do this or if it can be done with the TC-08. I feel the unit is at its limit of functionality which is fine because its not meant for anything other than temperature sensing but if it could help me on this project i would be extremely happy.

Answers on a postcard as to whether its possible to measure 10k NTC thermistors or if i should buy another datalogger specific to resistance measurement.

Any help appreciated

Regards

Pete Coster

Gerry
PICO STAFF
PICO STAFF
Posts: 1145
Joined: Mon Aug 11, 2014 11:14 am

Re: TC-08 and NTC Thermistors

Post by Gerry »

Hi Peter,

First of all, the answer on a postcard is - if you are trying to do what I think you are, then it can be done with a better NTC thermistor alongside your 10K NTC thermistor, and the TC-08.

To explain why, you need to know about measurement goals, terminology, limits of measurement system capabilities, and more, so the complete answer (if you're interested) is as follows:

In considering your measurement goals it's worth bearing in mind that, high temperature accuracy (how close the converted temperature value being logged is to the true temperature being measured) is more difficult to achieve when compared to other types of measurement (the ITS-90 standard, upon which all current temperature measurements are based, is only accurate to 5 decimal places, and unlike some other standards, e.g. derived at atomic level, using unique methods, ITS-90 was derived using more precise variants, of generally used apparatus and methods, e.g. a gold standard PT100 sensor, the triple point of water, and other known preset points). However, high temperature precision (the smallest measurable difference between 2 temperature values) is easier to achieve, with your average Data Acquisition Device typically having a resolution (which is essentially the digital equivalent of precision) of at least 10 times it's accuracy. The resolution of the TC-08 is approximately 100 times it's accuracy. As you are interested in how the NTC sensor changes relative to a fixed resistance value, I would say that what you need is a measurement system capable of high precision, rather than high accuracy. However, for completeness, we will discuss both precision and accuracy.

A general rule of thumb in data acquisition is that, if you want to perform a calibration/measurement of a measurement system you should use a calibration/measurement system that is at least 3 times as accurate as the system you are measuring. A 10K NTC therimistor (which has a typical accuracy* of ±0.1°C at 20°C) is considerably more accurate and stable than a thermocouple (which has a typical accuracy* of ±2.7°C at 20°C) for measuring the same temperature range. So, a thermocouple would be beyond it's limit of accuracy* in trying to log thermistor temperature against its resistance.
*Note that when discussing how close the measurement is to the True value (qualitative assessment) we use the term 'Accuracy', but when referring to actual numerical values (quantitative assessment) we are actually defining how far away the measurement is to the True value, so the correct term is 'Accuracy Error', which over time has been (confusingly) abbreviated to 'Accuracy'.

So, although the thermistor is far too inaccurate and unstable, the TC-08 is a data logger (specialised for working with thermocouples, because of it's CJC, input range, etc, but still a data logger) with differential inputs, reasonable accuracy, and over 16-bits of resolutio, so it is capable of measuring voltage, resistance and current, to reasonable precision. You can measure the resistance of the 10K NTC thermocouple by using a constant current source to generate a voltage across the thermistor. The voltage should be matched at the maximum of your measurement range with the full-scale input for the input range you would be using on the TC-08 Terminal board (so you may need to divide down the voltage using a resistor divider). Alternatively you could use the 4-20mA input range on the Terminal board, and a voltage source, set to a voltage that provides 20mA through the thermistor at the maximum temperature of your measurement range.

Going back to the sensor, using another 10K NTC thermocouple to measure the temperature would not provide the accuracy needed, and using an inherently more accurate temperature sensor such as a PT100 sensor (which will give you 0.04°C when using 1/10 DIN) would require an entirely different data logger (our PT-104). So, your remaining option to log the temperature would be to use a high accuracy, high precision thermistor (e.g. something like this one: http://www.ussensor.com/products/thermi ... thermistor, but you may need to search for a higher accuracy). With the TC-08 you can measure temperature using a thermistor by by measuring across the sensor as a bottom leg of a potential divider, and using the manufacturers scaling data to create a scaling file in PicoLog.
However, the accuracy of the high precision, high accuracy thermistor you select (which, for example, can be something like ±0.2°C at 42°C) is going to be better than the accuracy of the TC-08 data logger (±0.584 °C at 42°C). The accuracy of the TC-08 is specified with an uncertainty (±) that is based upon factors such as the variation of component tolerances between individual units, and the variation of results between different data capture sessions. You can reduce the uncertainties due to tolerances by performing a calibration for accuracy (which performs measurements of temperature using one channel and one specific sensor to characterize the accuracy of the specific pair over a temperature range). This gives a much tighter uncertainty for the specific TC-08 model and specific channel + sensor, but only as long as they are always used together. The calibration house may then also be able to specify corrections required to reduce even further the extremes of temperature variation over the range. This can be done by making use of the extra resolution available to the data logger to make the corrections with a scaling file. At Pico Technology, we can do the calibration for accuracy for thermocouples, but you would need to approach an independent Calibration House if you want it done for a TC-08 + high precision, high accuracy thermistor.
You can reduce the systematic uncertainties due to measurement session variation by repeatedly taking the same measurements. By doing all this, the difference in accuracy between the TC-08 and a high precision, high accuracy thermistor should be easy enough to overcome.

Alternatively, the resolution of the TC-08 is considerably better than the accuracy of a high precision, high accuracy thermistor. So, if you are in fact just looking for high precision changes then you already have that capability in your TC-08, and probably don't need to concern yourself with accuracy, calibration, and multiple capture sessions.

As a final note, aiming for high accuracy and/or high precision temperature data logging, is not just a case matching specifications. There are some basic principles that need to be considered in order to achieve your measurement goals and the following article provides a great source of information on them: https://www.picotech.com/library/applic ... asurements.

Regards,

Gerry
Gerry
Technical Specialist

Post Reply