DVRT Temperature compensation?

[Finnick]

Has anyone else out there ever worked with a DVRT as a displacement sensor at high temperature? I am currently doing so, and have electronics that operate up to 175°C but there is drift, mostly due to the resistance change of the windings of the sensor. Does anyone have any ideas/techniques that they know of that would help in this situation?

Thanks!

 

[VEBill]

One technique (in general) is to use a bridge circuit.

 

[Finnick]

Yes, that is what I am doing but I am using resistors to complete the bridge, could this possibly be the source of my problems?

 

[VEBill]

The symmetrically-positioned resistor (the one that balances the bridge) would need to be some arrangement of resistor(s) (or the like) that shares the same temperature coefficient as the sensor. An extreme example would be to use another identical sensor fixed at mid-range. And this part of the bridge would need to be exposed to the same thermal environment. And if the temperature changes rapidly, then the thermal time constants also need to be matched. The other resistors in the typical 4-element bridge circuit can just be good stable resistors.

The bridge approach is just one possible suggestion that springs to mind. It's an old school approach (not that I'm THAT old).

If the sensor is three terminal, then perhaps a ratio-based approach would be all that's required.

A modern uC approach would be to measure the temperature and use a look-up table to compensate.

 

[VEBill]

Subminiature DVRT: "Core position is detected by measuring the coils' differential reluctance... This differential detection method provides a very sensitive measure of core position while cancelling out temperature effects."

http://www.microstrain.com/sdvrt.aspx

 

[Finnick]

I have contemplated the microcontroller approach, but I would rather have the analog signal be as free from drift as possible, I know I have tested the results with just the sensor under heat and seen a resulting drift in the output. I am driving the bridge with AC, but have one side pulled to +12VDC rail then a pullup, then the bridge and then the drive signal. Could this approach be incorrect?

 

[VEBill]

I've not worked with DVRTs, but the information suggests that the ratio approach ("differential detection method") is what's required.

And if the sensor operates with an AC stimulus, then I don't immediately see why the 12VDC would be needed.

Have you consulted with the sensor OEM to see if they provide reference designs for the necessary electronics?

 

[Finnick]

Another question just came to mind, how would I go about finding a matching temperature coefficient to my sensor?

 

[Finnick]

Thanks for the ideas, VE1BLL, I have contacted the manufacturer for circuit ideas and I even purchased one of their amplifiers but they use a monolithic chip for part of the circuit (which doesn't work to high temperature) and they also have a patent on some of the circuitry so they are a bit tight-lipped about it.

 

[VEBill]

"...they also have a patent on some of the circuitry so they are a bit tight-lipped about it."

That's not how patents work. Patents (and even patent applications recently) are published for all the world to see. And they're supposed to provide complete information (although they're normally written in gibberish to very intentionally obfuscate).

http://www.USPTO.gov

 

[VEBill]

"...how would I go about finding a matching temperature coefficient to my sensor?"

The approach is probably based on a three-lead device and using a ratiometric measurement.