As part of developing a vehicle monitor (progress photo as below) for my Nissan Patrol turbo diesel 4wd, I'm wanting to digitise my coolant/water temp. Pre-existing on my td42 motor, I have 2 temp sensors, a 5v one for the ECU and a 12v one for the gauge. I believe the ECU one is primarily used in relation to the Glow Plug module. My hope has been to tap in to the 5v ECU sensor, without disrupting the factory controls, and so I don't have to plumb in an additional sensor.
Based on 3 resistance readings, the 5v sensor seems to be equivalent to the common GM sensor, and so I graphed those ohm vs degrees values and came up with a logarithmic line of best fit. According to my workshop manual schematic, the sensor is used in a simple voltage divider circuit, with the ECU supplying 5v, and the temp sensor as R2. By measuring R2, and the voltage across it when on, assuming 5v supply, I rearranged the voltage divider rule and first calculated R1 as 1095ohms, but then on following measurements, at different temperatures, I thought I found the supposed R1 to be as low as 760ohms (may have to re-check that measurement). I'll try to get further measurements and tabulate them. I also measured the open circuit voltage after R1 to be 4.78v, with ignition on, but engine not running.
I'm wondering if it's safe to read off a sensor which is simultaneously being used by the ECU, and whether my method for calculating the internal R1 is too inaccurate. Unfortunately the 5v supply is not externally accessible, I don't know if I should calculate with Vsupply as 4.78v. The other 12v sensor, i suppose i'd need to use a voltage divider, which would then affect the voltage the factory gauge read, I imagine. Plumbing in a third sensor is possible, but not ideal. Any help is greatly appreciated!
There are lots of different thermistor curves. If you really want an accurate temperature you need to measure the resistance at 3 temperatures (or more). You should be able to measure the resistance across the sensor when the vehicle is turned off. Then, run the engine for a little while and check again. Repeat for a third time at normal operating temperature. Amatherm has a good explanation, the wikipedia entry is pretty good also.
You can determine the pull-up resistance. Disconenct the sensor. Apply a known resistance across the pins and measure the voltage then back-calculate the pull-up resistance. Use a value that is near the sensor resistance to avoid setting a fault code.
You can read the voltage without affecting the ECU as long as you have a high impedance input. Anything that draws current from the pin will make the ECU think the resistance is lower and it will read a higher temperature. If the pull-up resistor is really only 1K then a few 10's of microamps won't make a difference. An ADC input will be pretty high impedance. Just don't have a resistance to ground from the sensor line.
Thanks SDturner that was a great help,
I found some supposed specs for the sensor which differ from the GM data, and a Steinhart Hart calculator so I'm implementing that. My trouble is getting some known temp references without pulling the sensor and losing coolant. I may have to just pull the sensor out or track down a spare and do some testing with my ds18B20 sensors that are also implemented.
I was concerned whether my ecu 5v was actually 4.78v, but then I realised that due to the form of the voltage divider rule, 5v and 1095 ohms should give me a very similar value to 4.78v and 1000ohms. Looks like I'm on track, thanks!
At the risk of setting a fault code you can disconnect the sensor and measure the voltage. It should be 5V, but the tolerance will probably be +/-3% or +/-5%. In my experience I see more initial tolerance and not that much drift over temperature or time, so once you measure the voltage you can use that in your calculations.
Do you have access to the temperature value either through the dashboard display or OBDII? Read the temperature from the car and the voltage at the sensor or turn the car off and as quick as possible measure the resistance of the sensor. Repeat a few times as the engine warms up.
If you only have a gauge or light on the dashboard, then you would need to pull the sensor to characterize it, or buy a replacement part (junkyard?) to characterize.
Thanks bud, unfortunately it runs on Nissan Consult rather than OBDII and I don't have a scanner. I was going to build/buy one but I realised coolant temp was really the only relevant spec, so building this is my substitute. Being a mechanical diesel, I think my ECU is 'dumb' and doesn't store codes
