Thermistor Bridge Linearisation

There is no such word as "copywrite"; it's "copyright".

And it would either be "copyrighted" or "copyright protected", in the sentence above.

I glanced at your link before I realized the link was your whole textbook. You really shouldn't waste other peoples bandwidth and steal others words that way. For your sake I hope that you did not make that link available to the public.

When any transducer has a non-linear response there are generally two approaches to making a precise reading. What you are trying to achieve is in-process measurement compensation. A non-linear measurement circuit attempts to compensate for the non-linearity of the transducer. This is rarely an easy circuit to design, even when only one non-linear parameter will be compensated. More often a post-process compensation is done after making a very precise linear measurement of the transducer signal. Then a non-linear calculation of the reading by either a look up table or quick calculation of the digitized signal.

The most important attributes are missing from your question to suggest which approach one should attempt. We do not know the thermistor, anticipated temperature range, acceptable lag time, nor the degree of precision needed from this measurement. With a thermistor there can even be possible self heating and/or wire lead heat sink cooling problems.

This book was not posted to the internet by my self, a simple Google search would indicate that someone else has put this book in to the public domain.

Was the "link" removed, or did a dog steal the sausage? I could not find it after an exhaustive search...

The post containing the link has been removed.

Without breaching copyright, please give details of the method with which the difficulty is being experienced.

If you understand basic electronics, you do not need the text.

(1) I have been using a bridge circuit to read out a thermistor in the range of 0-100 C in my home shop, as relates to a project I am collaborating with a colleague on.

(2) Bridge circuit will really help when data-logging, especially at the higher temperatures when the thermistor presents a lower resistance. Much cleaner response than a simple voltage divider, and more sensitive to changes in temperature. One does not balance the "Wheatstone bridge" in this case, but simply reads the error signal between the the left side (Va)(with the active element in situ), and the right side (Vb). It is elementary to set up the equation for the error signal: (Vb-Va). Express this in terms of the Vss (supply voltage). Solve for R(thermistor).

(3) All thermistors come with a data table, or the OEM will provide the parametric equation (usually a third order polynomial.) Having the determined value at a point in time during your experiment, simply log the error signal, calculate the R using a spreadsheet, then calculate temperature in the next column over.

I left out one thing, as I ran out of edit time:

For example, I am using a 10KΩ thermistor, and use it in the left hand leg from Vss (at the top), matching its reference value at 25 °C with a run of the mill 10KΩ resistor placed on the right hand leg. Then I use half these values on the down legs left and right connected to ground, thus 5KΩ each. In order to get good accuracy, it is important to know the exact values of the other resistors, and hopefully these do not change much in the outside conditions where the device is located.

I searched for linearization circuits, but not much luck when I was cobbling this together. I also tried to incorporate a potentiometer to adjust values, but I find this is practically of no use the way I am logging results and calculating in Excel spreadsheet.

Various techniques have been proposed, but the best ones rely on a network of multiple thermistors, where response is all but completely linear, presuming the thermistors are non-identical. The maths become somewhat hairy for the theory, but it opens up some clarity of purpose in selection of the other circuit component resistance values. Of key interest is the selection of the shunt resistors to be used only in a particular range around a target temperature (for each individual thermistor).

Since I am data-logging the whole thing, using something like the Steinhart-Hart model to convert R to T, I do not need to have multiples. If doing this completely in an analog manner, you need multiples for best accuracy.

One final consideration is the self-heating aspect: you must limit the power dissipation in the thermistor to a value that will not appreciably raise its temperature.

Have you considered purchasing a pre-packaged linear output device?

http://www.ti.com/lit/ds/symlink/lm35.pdf

The above link for one such device offered by Texas Instruments (not a commercial, or a recommendation, just an example of what is out there).

It's certainly not forbidden to mention the book the OP was referrring to:

Principles of measurement systems

by John P. Bentley

ISBN: 0-13-043028-5

Principles of Measurement Systems: Amazon.co.uk: Bentley, John: 9780130430281: Books

LINEARIZATION OF WHEATSTONE-BRIDGE

https://pdfserv.maximintegrated.com/en/an/AN6144.pdf