Simpler analog input board

[AndreyB]

We already have a working universal analog input board - the only issue is that it is too universal and this makes it bigger, as in too many square inches.

With a less universal board we would gain in simplicity - for example, for thermistors we would only need one pull-up resistor and that would be it. Question: can we use 3 volts pull-up for coolant and intake air thermistors instead of usual 5 volts?

[abecedarian]

My suggestion is to use 5v to the sensors then divide the returned signal 0.5x through an op-amp. You're not losing much doing so- likely less than the error of the ADC.

An 8 bit ADC has 256 'values' and with a 2v5 reference and using an op-amp as I mentioned could resolve to 0.01 volts. A 12 or 16 bit ADC could resolve better than 0.001v and 0.0001v, respectively, not considering LSB errors of course.

[abecedarian]

I think you can ignore the post I just made since I think I mis-interpreted your question.

Can you send 3v3 to a temp sensor and read the resistance and determine the temperature?
Yes, you can.
But since the temp sensor is spec'd at 5v, the resistance read will be different if sent 3v3.
So you will need calibration points based on 3v3, or take a chance doing some math.

[AndreyB]

My suggestion is to use 5v to the sensors then divide the returned signal 0.5x through an op-amp.

This is the way this works already - but that's plenty of components (like 4 or five). Why not feed sensors with 3v and have only ONE component per sensor (pull-up resistor?)

The sensor is just a thermistor - it changes it's resistance with temperature. It does not know anything about voltage.

[abecedarian]

My suggestion is to use 5v to the sensors then divide the returned signal 0.5x through an op-amp.

This is the way this works already - but that's plenty of components (like 4 or five). Why not feed sensors with 3v and have only ONE component per sensor (pull-up resistor?)

The sensor is just a thermistor - it changes it's resistance with temperature. It does not know anything about voltage.

Agreed, the sensor knows nothing about voltage. But, the ECU must know the correlation between voltage and temperature.
Or at least the resistance, and thus voltage, the sensor exhibits at various temperatures.

[AndreyB]

Agreed, the sensor knows nothing about voltage. But, the ECU must know the correlation between voltage and temperature.
Or at least the resistance the sensor exhibits at various temperatures.

The ECU already knows that it should take voltage, convert it to resistance and calculate temperature based on the resistance. The code is more or less in thermistors.c

[abecedarian]

My point was that most sensors operate at 5v0 so pull-up to 5v0 and measure the voltage drop across the sensor works.
If you change the voltage to 3v3, it will draw more current, and without proper calibration may not have its readings interpreted correctly.
The response of the sensor may change if the voltage changes.

[kb1gtt]

I had posted this some time ago, it might be handy for a comparison. http://jaredharvey.com/Files/projects/Engine_control/snow_blower/Murata_temp_curve.xls To change to 3.3V change cell C5 to 3.3 instead of 4.5. Take note that many thermistors do not like to go over 2mW, so when column G goes over 2.00, that's starting to cause some problems with the sensor. Using a 2.5k resistor instead of 2.2k, and dropping to 3.3V helps keep that column under 2mW.