I think I have something to contribute. Using your Y = mX + b equation I derived voltages for a temperature range from 6C to 50C. First, you must have a series circuit with a constant voltage, that is how your thermistor gets linearized, the current in the circuit changes with temperature and generates a (almost) linear voltage. In addition, I believe, that if your circuit has a 3900 Ohm resistor is series with the thermistor and the supply voltage is 5.0 volts THEN you are using the same thermistor as I have described previously. My thermistors are 10,000 Ohms at 25 Celsius. The 25 Celsius resistance is how NTC thermistors are identified, if you get my drift. Neat stuff....
On Edit: In the 6C to 50C data range, if the two end points are used to determine "Y = mX + b", I estimate the maximum error to be 2.5C (25.5) at midpoint (28C) using the raw voltage as opposed to the linearization calculation.
* This post was
edited 10/17/09 07:05pm by Harvard *
I use a laptop with an Autotap software & interface that plugs into OBD II connector. Displays tens of parameters in numbers, on gauges or in graphs. It also has a provision for logging chosen parameters.
In this chart the straight line is the R = E / I if I (current) were a constant current source.
One of the two curves is the resistance from
R = EXP( (1.0 - A * T)/ (B * T)
The other curve is the method used by the OP. Obviously the use of the formula is not worth the effort.
* This post was
edited 10/18/09 09:26am by Harvard *
You got a keen interest in sensors! I'll supply my data for Ambient Air. I measured temperature with a Harbor Freight multimeter that has a K type thermocouple. I suspect the accuracy isn't better than +/- 1 deg.C. Here's a copy of the excel calculations.
Column 1, 2, 3: Measured voltage, measured temp (C), temp conversion to F.
Column 4: Calculated temp (C) using m = -21.38, b = 98.2
Column 6: Calculated temp (C) using m = -21.053, b = 99.05
Column 7&8: Calculated temp (F), m = -39, b = 212; % error
Column 9&10: Calculated temp (F), m = -40.5, b = 217.96; % error
The Sprinter service manual provides a table to determine transmission temperature. The fluid temperature is a critical parameter in determining fluid level. It doesn't appear to be a ntc resistor. This puppy is damn linear! I use m = 254.7, b = -244 to calculate transmission temp.
That's a nice set-up also. If you want to get extended parameters (like transmission temp) it will cost $450. But it sure is easier asking the OBD connector what the temperature is than trying to do it yourself. It's like re-inventing the wheel. I've been retired for 10 years now. Told DW I needed to exercise my brain with an engineering project. So I guess I'm out there banging my head, re-inventing the wheel. There are a lot easier ways to get this data though!
I viewed the Autotap video clip on o2 sensor. It's pretty cool.
Sal
liborko wrote: I use a laptop with an Autotap software & interface that plugs into OBD II connector. Displays tens of parameters in numbers, on gauges or in graphs. It also has a provision for logging chosen parameters.
liborko wrote: I use a laptop with an Autotap software & interface that plugs into OBD II connector. Displays tens of parameters in numbers, on gauges or in graphs. It also has a provision for logging chosen parameters.
I too do this for retirement entertainment value. I have written a single C++ application that logs all OBD II EUs and a GPS interface via COM ports plus a 10 channel 10 bit A/D chip via the LPT port. The only thing I have learned from all the collected data is that the GPS output speed is in knots.
Offline
