Review of alternate Thermistor part options?

Howdy,

Like Steve DeGroof, I've destroyed a couple of my thermistors experimenting (mostly clogging my extruder up with HDPE). Getting ready to order more...


I found this thread from last year:
[forums.reprap.org]

I'm looking to get feedback on what thermistor parts have been tried and the results. Or threads I overlooked?

Thanks.

When I was playing with the temperature equations [forums.reprap.org] I found that, for high temperatures, there just wasn't enough resolution. As far as I can tell, using a 100k thermistor with a 1nF C3 should give a decent range for most temperatures.

DigiKey has a 100k one with a beta around 3800. I might have to order one just to see. The 1% ones are $10 but the 5% ones are only $.80. I'm not sure if the tolerance matters. If you take measurements and calculate the beta, you should be able to calibrate for any built-in inaccuracy, right? Plus, you can buy 12 of them for the same price as the 1%. All this assumes that a given thermistor that's, say, 3.55% off is consistently 3.55% off.

The reason for moving to a 100K thermistor is so that at the high temp end it does not have such a low resistance that it exceeds the drive capability of the PIC. It does not affect the linearity or lack of it. If it was simply to get a bigger time constant you could have just increased the cap.

To get good resolution over the whole range you need the parallel resistor and then a small change to the host s/w to take it into account. Without that you can't get good resolution at the high end without going off scale at the low end.

---

[www.hydraraptor.blogspot.com]

I got 5 of the 5% 100K thermistors from Digikey yesterday. A quick check shows that, while the beta and Rz varied from piece to piece, each one was consistent. They seem to work the same way as resistor tolerances in that, if you've got a %5 part and they also sell a 1% part, yours is going to be somewhere between 1% and 5% off the spec values.

If you don't mind doing a bunch of measuring and calculations, this isn't a bad option. I got all 5 for $4 and ordered them with a bunch of other stuff I needed, to save on shipping.

After redoing my calculations, it looks like the lowest C3 value you can get away with is somewhere around 200nF. If I worked it out right, you should get about 3-degree increments at 250C. I tried a 1μF cap and the readings didn't look right. I'm guessing the PIC can't handle that.

It would be better if the host s/w preferences just had two temperatures and two resistances rather than Rz and beta. Let the program do the calculation rather than the user!

---

[www.hydraraptor.blogspot.com]

ah, that is a great suggestion! we should run that past the team, nop!