One of the trickiest sensors to monitor is an RTD, aka resistive temperature device, imho. They're relatively simple where their resistance changes over a temperature range. These types of sensors are used to monitor coolant/water temps, oil temp, and even air temp. They also have a slow response time, which can easily be 3-5 seconds but usually no quicker than 0.5 s. It's difficult to read them with a regular ADC if their apart of another circuit too. Why? We're loading the circuit and disrupting it's calibration when we tap it. Since its a resistive device, we are basically creating a voltage divider of which their output changes drastically with load. This was one of the first technical hurdles that launched me into developing D-DAQ.

Turns out I needed a high input impedance method to get an accurate reading. This brought me into the world of op amps, and boy, it is a big world... Anyhow, over the course of a year or so, I became more familiar with terms critical to this application such as input bias current and offset voltage. I had watched those videos a few times just to understand what was going on. These 2 specs were very important because the ADC is 10-bit for 3.3 V, which gives ~3.2 mV divisions. Next up, I wanted to minimize error via gain and also use high gain for low readings and less gain for higher ones so I decided to setup a 2 stage amplifier. That, or a logarithmic amplifiers to counter the exponential response of the voltage increase.

How to implement such was tricky and the interwebs provided a few solutions, one of which was a logarithmic amplifier, but few applicable as they would vary over temperature thus causing calibration/characterization problems. I need a precise circuit. How about PWM? Well, even after 2 or 3 RC filters, the output was too difficult to make consistent over various duty cycles. This bumped me over to FETs. I've known about them, but wasn't keen on their operational specifics. A few weeks later after playing around on circuitlab.com, I came up with a response I felt worked really well.

40 K ohm to 80 ohm. That was is the initial RTD range that I needed to monitor. The circuit the RTD is on is a 5V source with a wire resistance <0.5 ohm. I used a couple regressions to plot extended ranges for the RTD outside the stated spec. As it stands 40K and 20K can give the readout for the same temperature. There will need to be calibration for measuring these sensors on various cars, but from running the regressions, only 2 temperatures are needed. The further apart the better though.

I spent about 9 month on and off trying to figure out a solution. After all, it's basically just math in a physical form. Since it could be done on paper, it could be done physically. Don't give up as there is always a way.