Blog #8: Characterising Thermistors – Practically Running Multiple Thermistors

RE: Blog #8: Characterising Thermistors – Practically Running Multiple Thermistors

DAB — Thu, 18 Aug 2022 20:06:11 GMT

Very interesting blog, the results should be interesting as well.

RE: Blog #8: Characterising Thermistors – Practically Running Multiple Thermistors

Andrew J — Thu, 18 Aug 2022 15:33:19 GMT

It will make for some interesting results. I was going to mention the possibility of putting an OpAmp on the multiplexor output but I see Michael has already done that!

I'd be very interested in the following (not from this experiment): using a thermistor to measure case temperature of a TO-220. In particular, how to best attach to the IC (without using the ring connector version - in any event, case temperature not tab temperature) so that a good contact is made and maintained over thermal cycles and that has either no impact or constant impact on the temperature reading, or at least and understood impact; and as an adjunct, at what point self-heating becomes irrelevant at higher temperatures. A while ago I used a thermistor to measure a bridge rectifier temperature but attaching it was the devil's own job. I used Kapton tape in the end but it wasn't ideal. In a new project, I want to use a thermistor as an alternate opinion to a heatsink temp monitor IC (shout out to [mention:0d133614bf01463a9e17ef49e25b2c0b:e9ed411860ed4f2ba0265705b8793d05] for this as well.)

RE: Blog #8: Characterising Thermistors – Practically Running Multiple Thermistors

robogary — Thu, 18 Aug 2022 12:41:27 GMT

Gough Lui - Spot on with the multiplexing of thermistors.

For applications using a whole bunch of thermistors on bigger machines - a dozen, to several dozen, to hundreds or even thousands, the thermistors are multiplexed, but in a different way. The excitation is multiplexed as well as the thermistor value reading.

The reason is two-fold: thermistors are intended to be scanned infrequently i.e. once a second at most, oft times even slower. There are exceptions tho. There is wasted energy keeping the thermistor powered when not making a reading.

The second reason is each thermistor circuit draws some load, even if small, the loads add up and increase the power supply cost ( on bigger machines). Multiplexing them saves some of that cost.

The bad thing about doing that is, if troubleshooting, and put a DVM on the ckt, the troubleshooter cant get a valid reading steady state.

RE: Blog #8: Characterising Thermistors – Practically Running Multiple Thermistors

michaelkellett — Thu, 18 Aug 2022 12:25:19 GMT

I look forward to seeing some test results.

There are some nice modern solid state switches and multiplexors being made but some are currently hard to obtain.

I rather like TIs TMUX1219: 2:1 mpx, 3R on resistance, 15nA typical leakage about 60p @ 100 off, in stock at Mouser

One I've used with thermistors is the ADG805, 4:1 mpx, 0.5R on resistance, 0.1nA leakage, about £2.5 @ 100 off, in stock at Mouser (NB abs max supply voltage = 4.6V)

This is a rather extreme way of doing it when you don't much care what the bits cost !

The quad op amp is a Microchip MCP6V64 Zero drift chopper amp. £3.28 at Mouser (1 off) but none until Novemebr 2022.

VA in the circuit is the ADC reference voltage. The point of the complicated amplifier is to get a signal centered on the ADC mid point with some gain, while not introducing offsets drift. It's used a part of an experimental medical system to do precise temperature control. The thermistor feed resistors are chosen to be equla to the thermistor resistance at 37C, the working temperature range is 32 - 42C.

Most of the resistors are 0.1% 10ppm/C parts.

The MCP6V64s are quite nice, I'ves used them with strain gauges with some success as well.