In what ways does NTC thermistor quality affect measurements?

Question

I've been looking at standard (i.e. 10 kohm, -50C to 150C) NTC thermistors and am wondering what the tradeoffs are for quality vs cost for applications where +-1C is acceptable accuracy.

For instance, it's well known that cheaper carbon resistors are less accurate and have more temperature drift. However, assuming that thermistor quality manifests itself in this same way then the variance would seem to be a non-issue once each thermistor is individually calibrated with a LUT.

Or do cheap thermistors drift meaningfully more with age than expensive ones?

Answer 1

Most passive components with different tolerances, like resistors, thermistors, capacitors, etc, are all made on the same assembly line, using the exact same process. Then they are assigned to tolerance bins based on their measured, at room temperature, value.

Lets say you have a nominal 10 Kohm resistor. Those with values within 10 ohms of 10 K are labeled 0.1% parts. Those with values outside that range but within 100 ohms of 10 K are labeled 1% parts, and the remainder go into the 5% or 10% bucket. The components with the values closer to the nominal command higher prices.

Since all these parts come off the same manufacturing line using the same process, I would expect that parametric variations such as aging or temperature caused changes would be pretty close to the same.

You mentioned "quality" in your question. That's somewhat different than the initial tolerance. Parts destined for hi-rel applications like aerospace or space generally have a lot more paperwork, inspections, maybe lot testing, etc involved to guarantee their performance in those rigorous environments and long life times, which may be 20 years, 25 years, or more from design through deployment and end of life. A lot different than a typical consumer product.

Answer 2

The relationship between price and performance is not necessarily straightforward. For example, around 30 years ago a joint venture between two Japanese companies produced very inexpensive thermistors with high (a fraction of a degree C) accuracy, good reliability (around room temperature) and matching electronics to read them using direct single cell power with no regulation and no expensive voltage reference. That was expanded into narrower ranges again with home clinical thermometers. The price/performance ratio was off the charts, but necessary to address an enormous market.

So why would be use an expensive YSI thermistor rather than an almost free consumer product? Hint "quality" (as in conformance to specifications, products from any reputable manufacturer are high quality) is not the reason..

1. Wider temperature range with good performance. Hundreds of degree C rather than tens.

2. Better seal such as glass which is resistant to chemicals and other things in the environment.

3. Inertia- some grad student in 1993 used XX model and wrote a paper on it, so the new experiments will also use that part.

In general, thermistors are quite reliable in a benign close-to-room temperature environment. The accuracy can be mediocre or quite good. Sensitivity is high, as is nonlinearity, to the point where dynamic range is an issue without good circuit design in wide temperature range applications. It's often hard to thermally couple them closely to the mass being measured, and response time/self heating specs tend to use optimistic measurement conditions (in oil or even water flowing at 1m/s for example).

I did mention "reputable manufacturer"(s). I'm not sure that parts made by 'unknown' companies for the massive battery pack market would be as good a bet as those from well-known multinational firms.