Automotive sensor conditioning guidance

Scroll down to see circuit.

Code and circuit works great on the bench. Ready for install, but it seems that to ensure a stable install in a a noisy environment, I may be well suited to protect the signals with either components in the circuit, or wiring, or code. I have searched through posts here and some are relevant but some advice either feels like a guess or does not have an explanation.

QUESTIONS (see below for current circuit and details that may help with answers):

  • What would should I add to the circuit for filtering or protection?
  • Would shielded cable or twisted wires help or be necessary? 24g wires okay?
  • 0.10% resistors necessary or 1% is good?
  • Why pullup vs pulldown for reference resistor? Should I change?
  • Should I have a delay between my 5 data samples for a reading?

I have not calibrated for absolute accuracy, but it reads as expected and the resolution sweeps great at 0.1°.

DETAILS:

  • For the 10k thermistors, I use steinhart equation with specified constants, resolution coded to 0.1°c.
  • For the 2 Delphi coolant sensors, the spec sheet only has a lookup table, so I fit steinhart best I could and is within 0.1°c through the working range.
  • Time constant of fanChange->tempChange creates a PID cycle that requires temperature readings about every 5 seconds, but my display refreshes at 2Hz so that is when I get readings.
  • Resolution is more important to me than absolute accuracy (but I'd like to be within 1°c absolute). I just want to make sure transitions are smooth without jitter for each PID cycle at 0.1°c readings.
  • Critical working temp ranges have thermistor resistance range from 1,250 ohm->130 ohm and total non-critical range starting at 50k ohm downward.
  • Code reads the resistance using 10k pullups and converts to 0-1023.
  • Code averages 5 immediate looped readings to get a value.
  • I have started wiring with 24g leads from Arduino under dash to thermistors under hood, about 6 foot run and potentially passing some noisy stuff but I'm isolating best I can.
  • 4 thermistors & 1 pwm fan signal will all be run through engine compartment where things can get noisy.

THANK YOU for any help. I'm a hobbyist so any detail you can provide with your suggestion would be helpful so I don't screw up your good advice! Exact spec of suggested component, where to wire, and polarity if necessary. Any help received will certainly be passed on as anyone might ask me for help.

  • Scott

A couple things I have picked up by clicking around.

  • A low pass filter can and probably should be put on each of the 4 thermistors. 0.1uF capacitor through a 100k resistor. Cutoff frequency = 15.9Hz and time constant of step response 10 milliseconds. Fine. I'm not sure what other gotchas for these filters?

  • A look at tolerances revealed to be cautious about tolerance of pullup resistors. I'll put in 1% or better. The stuff I have laying around is 2%. I calculate that tolerance of temperature reading to improve from +/- 1.2°f to +/- 0.6°f at my expected working values.

I saw an interesting article about wiring a transistor as a switched gain to turn on another resistor reference in parallel at a certain temperature so the range of sensor resistance readings is split and optimize more of the range to better approximate the reference resistance for the divider and lower tolerances. Clever, but my working range is pretty tight where I really care about it.

Here's a look at the sensors portion of the circuit and the addition of the low pass filter. I'll wait a couple days to solder up in case anyone has input, but I think I'm comfortable with this.

To many questions in one post..
Pull up is usually safer than pull down. No 5volt running to the sensors that can short to ground.

Shielded or twisted is always good. Adding a PWM signal wire to the sensor wires might not be so good.

Pull up resistor tolerance is irrelevant. Absolute value (measure it) can be entered in the Steinhart-Hart formula.
A stable value is important. Use common/cheap 1% metalfilm resistors.

"Resolution is more important to me than absolute accuracy"
Then you shouldn't use 10k pull up for the radiator thermistors.
Highest resolution is when pull up resistor value is about the same as thermistor value at the temp of interrest.
A ~2k resistor (1k8, 2k2) might be better for a ~80C radiator.
Leo..

Thx Wawa.

Great point on tolerance of resistors.

The RC filter updated in 2nd post diagram look okay?

RC filter (if needed) should be ok.

For the pull up resistor value...
Google "10k thermistor chart".
Leo..