Resistance temp sender equation/formula

MMiskoe
Posts: 17
Joined: Sat Nov 05, 2011 2:42 am

Resistance temp sender equation/formula

Postby MMiskoe » Mon Sep 09, 2013 2:43 am

I'm using an Autometer single wire temp sender, 5V reference from the DL-1, 1000ohm resistor and getting back very clean data, voltage ranging from 2.7 (room temp) to .5 or while hot. I've got an OBDII reader that provides deg F of the same location (the two sensors are in the same coolant passage), the DASH provides the voltage.

From teh variable manager under "create formula" portion inputted the voltages displayed in one column, actual temp in the other column, it created a formula. However, it is incorrect. Here's the values used, and the formula created:

0. * x^3 + 0.00404 * x^2 - 0.692 * x + 40.7

Deg - Volts
150 1
160 .87
170 .81
180 .71
185 .67

When the formula is used to correct the displayed value, it is no where near where it should be. Is there any other way to create a formula like this? Or is there something else within the program that I am doing wrong?

I have more data on the lower scale, starting at about 60deg F, where I need the formula though is where the voltage is down in the .45 range.

Support

Postby Support » Mon Sep 09, 2013 11:02 am

It looks as if there is a problem with your equation, since if you put 1v in to the equation you will get nowhere near the result you need. Are you logging a raw voltage on the DL1 and converting to a temperature in the Analysis software and in the DASH2? If so the equation would be something more like:

-108*x + 257

Which for a value of 1v would give you 149 and for 0.67 would give you 184.6.

Martin

MMiskoe
Posts: 17
Joined: Sat Nov 05, 2011 2:42 am

Postby MMiskoe » Tue Sep 10, 2013 12:40 am

Are you logging a raw voltage on the DL1 and converting to a temperature in the Analysis software and in the DASH2?


Yes, the DASH displays the raw voltage and I've written it down by hand concurrently with an OBDII reader that can display temp.
If so the equation would be something more like:

-108*x + 257


Two things about this - one this equation works OK in the range I posted here, but it at the lower end it falls off badly (off by 100 deg F) so I am wondering how accurate it is further up the temp scale.

Second, why does the Analysis program not return this for an equation? I tried entering either all of the data or only portions of it and it was never close at all. So what did I do wrong?

Support

Postby Support » Tue Sep 10, 2013 10:49 am

please send me a screenshot of what you are doing in Analysis and I should be able to see from there. I only generated the equation based on the information you gave me, if you have other data points you want to be included in the equation then send them to me and I'll take a look. You can email me at mhill at race-technology dot com.

Martin

924RACR
Posts: 113
Joined: Tue Jul 13, 2010 5:34 pm
Location: Royal Oak, MI
Contact:

Postby 924RACR » Tue Sep 10, 2013 6:07 pm

Not sure about the Autometer temp sender, but the VDO senders I'm using are exponential/logarithmic in their output (per the mfgr). Same wiring setup, I suspect, with 1000Ohm resistor.

When I tried to use the Analysis SW to create a mapping, the accuracy was very poor, since it was only trying to do a polynomial fit.

However, you are allowed to manually input any equation you like for the conversion.

I pulled the raw data into Excel and created my own log function to re-map the temp output, and this has resulted in a disturbingly accurate calibration! ;) It's now within 1-2 deg F of temp as compared to an IR pyrometer.

For one example (200C VDO temp sender), the equation was:
-270*LOG(0.2*x+0.15)+55

Again, that of course is to convert to deg F, not Celsius, from input voltage X.

So if you can get enough measured data points, you might achieve a similar improvement in accuracy if you use the above format and just fiddle with the constants to curve-match as best you can.

Seems like Race Technology might wish to consider a future option/upgrade to use log or exponential functions for the curve fit?


Return to “Sensor-related questions”

Who is online

Users browsing this forum: No registered users and 10 guests