Calibrating and Testing Coolant Sensors
Turbo Mini Forums

I have not posted in awhile, but I still check in from time to time.

I have a Specialist Components kit on an MPi block in our 64 Traveller. It has been running Ok after a few suggestions from here. The ignition hunting on idle seems to have subsided, but I cannot say why. It does happen on occasion.

With all the Minis running pretty well and my son's 67 El Camino on the road, it is time to fine tune the SC kit. The problem I seem to have is the car when at speed runs too cool. The temperature enrichment tapers off at 80, but when moving the car will run at 70. I actually have a 90.5 thermostat. It seems like the thermostat is sticking, but I have tested it and it seems to function, but maybe a little slow on closing.

What I would like to do is test the Rover MPi temp sensor I am using and verify the table that is provided in SXtune. It seems like it should be relatively simple to do this, but I am not very good at electrics or electronics.

I have this that I think came from the Rover service manual:
Engine coolant temperature (ECT) sensor
The engine coolant temperature (ECT) sensor is located on the coolant outlet elbow. It is a temperature dependent resistor (thermistor), the voltage output of which varies in inverse proportion to temperature, in that the output increases as temperature decreases or the reverse. The change in resistance is monitored by the ECM on a KG wire and as a result the ECM can adjust the length of injector opening time required. The ECM supplies the coolant temperature sensor with an earth path on a KB wire.

So I thought it would be easy to test out of the engine with a 5 volt power supply and a voltmeter.
Can anyone suggest how I would make the connections to do this? I seem to read a constant 5 volts no matter how I connect the sensor.

I can test the resistance fine, but I would like to check the voltage output to verify the temperature table

With the sensor only, you will always get the same voltage as what you're powering it with. To get a varying voltage, you need a bias resistor (a resistor between 5V and the sensor). But to get the voltage from a temperature table (I assume you have a table of temperature vs voltage at ECU connector) you need to know what the resistance value is for the bias resistor used inside the ECU.

There should be a resistance vs temperature table somewhere which would simplify your measurement. A resistance measurement is the usual method for calibrating a thermistor.

http://www.jbperf.com/

Thank you Jean I knew I was missing something. I was hoping to do some tests to check the table provided. The same sensor was used in MGF engines and the tables I have seen for those are different than the version provided in SXTune.

This is the graph of the provided table


This graph is copied from the SX Tune interface.

Edited by tmsmini on 13th Oct, 2018.

This MGF site has a full table with voltage and resistance values.
https://www.mgfcar.de/sensor/index.htm

Some MGF actually used two temp sensors one for the gauge and one for the ECU

First there is something wrong with that graph (or the table it was generated from) pictured above.

As Jean says, you can only see a voltage vs temperature curve once the bias resistor is connected and the more normal way to approach this is to look at the resistance vs temperature curve (or table) of the sensor itself.
The sensor itself will have an NTC thermistor inside it, (Negative Temperature Co-efficient) which obeys a very precise equation called the Steinhart–Hart formula (google it) which produces a nice smooth curve so the voltage achieved after adding the bias resistor should also be a nice smooth curve. Go to DigiKey or Mouser and look at the datasheets of any thermistor (the bit inside your temperature sensor before all the metal case etc is added) and look how smooth the curves in the datasheets are. That curve in your graph is not smooth at all.

Secondly, the MGF one won't have matching voltages simply because it will have a different bias resistor. However the resistance values look valid and the resistance values themselves will always be what they are whatever bias resistor is used by different ECUs so long as you measure them without the sensor connected to anything - which is the easiest way of seeing if it's giving the right output.

A long time ago, one of the members on here (hasn't been active for ages) did a comparison of four Rover sensors. I can't find the thread but I do have a copy of the results (just in case I was ever to use a Rover sensor) and, as he published them at the time, I'm sure he won't mind me repeating them.

Attached as a pdf copy of his spreadsheet (TM won't allow the spreadsheet itself to be attached) and also attached a jpg copy of the sensors I use (Weber/Marelli from a Cosworth) so you can see they are the same profile/curve, just different values.
The spreadsheet (pdf) is very similar to the MGF table, provided you just look at the resistance.

So if you want to check your sensor I suggest you test it alone, just measuring its resistance completely disconnected from the ECU, in a jug of iced water, then a pan of boiling water and a few temperatures inbetween as the pan cools down.

However, all that assumes is that the SC ECU is actually calibrated for the Rover temp/resistance curve by default. The Megasquirt, by default, is calibrated for a common GM sensor but is easily user calibrated for other sensors - just like I had to do for my Weber/Marelli sensor - just by entering three points from the table because the Steinhart-Hart formula only needs three points, the maths of the curve is that precise.
Does the SC ECU not have the ability to put an alternative calibration in and, if so, maybe the wrong values have somehow got entered.

Attachments:

46.27kb

37.57kb

Schrödinger's cat - so which one am I ???

Thank you for your comments. I will do some resistance testing today.
The SX Tune interface does allow for adjustments of the coolant sensor table. And it does display voltage in the interface as well. I was just looking to verify some values independent of the SX Tune software and the Typhoon ECU.

The only surefire way to do this...is calibrate it yourself.

Have the sensor plugged into the ecu, reading temperatures or voltages etc on the ecu...then making direct comparisons against a good known accurate temp gauge...be it a thermometer or other known calibrated means of reading temperature.

ie stick sensor in some water in a pan along with the thermometer and heat gently, and plot your own readings.

any other way involves guesses or assumptions.

Edited by stevieturbo on 12th Oct, 2018.

9.85 @ 145mph
202mph standing mile
speed didn't kill me, but taxation probably will

http://www.jbperf.com/

While watching my pot boil, I did some searching and found this one too from a few years ago:
http://www.turbominis.co.uk/forums/index.php?p=vt&tid=222846
He has some tables attached.
I actually corresponded with Darren a few times and met up with him at a Mini race in Minnesota a few years ago.

The sensors provided now are not Elmwood though, no markings on them except for a date code on some.
A couple of different type with the larger plastic body on some.

Edited by tmsmini on 13th Oct, 2018.

Just realized Darren did all his work with the fahrenheit scale.
I wonder if I can copy and paste into excel and apply a conversion formula.

Anyway my manual testing only verified that all my sensors works OK. Sensor 1 and 2 are Elmwood and 3 and 4 GTR206. the one labeled GTR206 is a brand new sensor.

Edited by tmsmini on 13th Oct, 2018.

Not the best test setup, but it worked.


Could only get down to about 3 C though.

Different ecu's will use different internal pullup resistors. Screw formulas, test properly, as installed connected to the ecu.

9.85 @ 145mph
202mph standing mile
speed didn't kill me, but taxation probably will

Schrödinger's cat - so which one am I ???

9.85 @ 145mph
202mph standing mile
speed didn't kill me, but taxation probably will

The term pullup might be common but it is definitely not correct in that case. A pullup is used for pulling up a signal to a specific level (a logic 5V or 3.3V for example) which is not what you want here because you want a varying voltage. It may look similar on a diagram but it is different from a bias resistor.

And that might sound pedantic and argumentative but that's a simple rebutting of your statement that this is common and correct.

And finding the bias resistor value is not that complicated. You could measure the input without a sensor to make sure that the voltage is 5V then use a potentiometer in place of the sensor to find the resistance where you the input voltage is 2.5V (or half the open voltage if that's not 5V). The potentiometer resistance at that point is the value of the bias resistor.

Edited by jbelanger on 14th Oct, 2018.

http://www.jbperf.com/

Schrödinger's cat - so which one am I ???

It doesnt need a formula if you test/calibrate as I described.

Keep it simple always works, and as installed calibration can only be correct, with no doubt.

9.85 @ 145mph
202mph standing mile
speed didn't kill me, but taxation probably will

Edited by jbelanger on 17th Oct, 2018.

http://www.jbperf.com/