Thermistors

[SteveL]

Labview 8.0 connected to a USB-6008

I am using 'BC Components' NTC Thermistor. I used the express VI to create a Vex input and entered the A,B,& C values from the datasheet. The thermistor is the 10K version and I placed a 10K resistor in series with it. I wired as though it were a 3 wire device, as is shown in the connection diagram, and applied 2.5 volts (generated by the same USB-6008) to it. (I also tried a seperate power supply)

I get completly the wrong temperature readings out of it! I called the NI help line in the UK and the engineer confirmed that I had done everything correctly, but couldn't understand why it didn't work.

Any thoughts?

Steve Lawson

[peteski]

Labview 8.0 connected to a USB-6008

I am using 'BC Components' NTC Thermistor. I used the express VI to create a Vex input and entered the A,B,& C values from the datasheet. The thermistor is the 10K version and I placed a 10K resistor in series with it. I wired as though it were a 3 wire device, as is shown in the connection diagram, and applied 2.5 volts (generated by the same USB-6008) to it. (I also tried a seperate power supply)

I get completly the wrong temperature readings out of it! I called the NI help line in the UK and the engineer confirmed that I had done everything correctly, but couldn't understand why it didn't work.

Any thoughts?

Steve Lawson

Steve,

I've recently been rather busy or else I might have noticed this and replied earlier. I ran into a similar problem using 10K Thermistors with a USB-6008, and found in looking at the "specifications" that the input impedance of the USB-6008 is only about 144K. This is only an order of magnitude away from the impedance of the thermistor, which introduces non-linearities. I tried to model using the 144k "quoted" impedance, but discovered that I did not have enough information on the USB-6008 output circuit to model it adequately, and the NI tech rep I was contacting about this wasn't able to help very much. In the end, I decided to use a bunch of "known" resistors (measured via ohmmeter to verify values) and create my own calibration curve by measuring the voltage reading of the thermistor detection circuit and relating it to the known resistance.

Interestingly enough, what I came up with seemed to create an almost linearized voltage to temperature function! It seemed that over the range of values I calibrated for, the non-linearity of the voltage vs resistance relation cancelled out the non-linearity of the Steinhart-Hart thermistor relation, at least within the allowable error for the application (which was something like a couple of degrees C). I can't guarantee this would be the case throughout the entire operating range of the thermistor, though.

I would strongly suggest that you do something similar. Resistors are relatively cheap, and you should know what range you expect to be operating in, so with about a half an afternoon of set-up and work you could have your own calibration curve. I'd be a little leery of sending my results for your direct use, because I can't guarantee that the variation between the USB-6008 that I used and the one you are using wouldn't be enough to throw off the calibration significantly. The USB-6008 is a rather "cheap" device, so I'm not as confident that the production of them undergoes the same "scrutiny" as, say, NI's PCI/PXI MIO cards.

Hope this helps!

-Pete Liiva