connect NTC thermistor to input, low pass filter?

I am using a thermistor in series with a known resistor to make a voltage divider, center point tap to the Arduino I/O pin for an analog read.

This is to work in a pretty harsh electrical environment (automotive), and will have a couple meters of cable between the thermistor and Arduino.

I don't like the idea of having a long "antenna" like this directly connected to a microcontroller I/O pin.

Would it make sense to put a low pass filter at the pin, by introducing a series resistor into the pin, and maybe 100nF cap to ground at the pin?

I've already experimentally made with a nice table of thermistor temperature to voltage values. Is this low pass filter going to skew analog read values? (It's a slow sampling speed.)

by introducing a series resistor into the pin, and maybe 100nF cap to ground at the pin?

yes good idea.

(It's a slow sampling speed.)

The sample speed remains unchanged. However the speed of change from the thermistor will be slowed. However, thermistors do not respond very fast anyway due to the thermal mass of the sensor. The amount of delay you will get will be approximately RC, that is multiply the capacitance (in Farads) by the resistance (in ohms) to give you the time delay (in seconds). This is insignificant compared with how fast a thermistor can respond. You will be lucky to get thermistors responding faster than 50 Hz.

I guess we could talk about the type of cable used for the run to Thermistors... shielded, of some sort. To help fend off inductive pickup.

I was thinking maybe 1K in series, with 100nF for the low pass.

Enough to limit an unexpected inrush, say, in the horrible event that somebody connected +12 battery to my input under the hood to try and “test it”.

…now I’m spooked into wanting to use a 5.1v zener shunt to ground on the pin as well. I guess that would pretty much make it bullet proof.

Hm… zeners have capacitance. I wonder if I can do away with the capacitor if I have a zener hanging there instead. Thoughts on that?

Oh, and what I meant by slow sample rate is my application sampling using analogRead... not how fast the AVR ADC internally samples when it's collecting a reading. I know hanging a low pass filter will make the Thermistor change a little more lethargic at the pin, but that's OK they don't change fast as you say.

My application would be happy sampling 1Hz.

I might even use a rolling average as the working value. So I'd always be computing the midpoint between the last working value and the new sampled value, to arrive at a new working value. Not fast stuff.

have a look at:- It shows you can use catcher diodes as well.

Yes you can make the series resistor higher but 1K is going to interfere with the data.

The capacitance of a diode is too small for it to be effective as a filter.

what I meant by slow sample rate is my application sampling using analogRead.

Still not sure how a filter affects that.

Nice page Grumpy.

So you suggest staying below 200 ohms for the series input resistor, cool. I’ll bump the design down from 1K to 100 ohm, and stick with 100nF. And a zener.

An issue I’ve got with two clamp diodes is that over voltage is clamped to the Vcc rail… and that goes into the AVR which just seems wrong. I know I could clamp Vcc over the Vreg to Vin as well, so if the spike is higher than Vin, then excess gets pushed to the rail of the car battery… but that still means MCU Vcc is seeing a shot as high as Vin.

Whereas a zener shunt wastes the voltage to ground, yes it depends on the curve as you say… Maybe the combo of zener plus depending on internal clamps on the pin is a good combo. The internal clamp catches the fast transient while the zener catches up.