ECU / EMS water temperature sensor
Turbo Mini Forums

Two questions really...

Can anyone confirm the thread size of the Rover sender. I hope it is M12 X 1.5 (ie, one of the versions of metric fine, rather than standard ISO metric coarse).

Does anyone know where to get the Temp/Resistance curves for Lucas sensors ??? I can find everything I want to know about Bosch (and am tempted to use their's instead) but Lucas' web site gives me nothing....

Rod.

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

You wont ever find published scaling for the Lucas sensor, it just doesnt exist

I did the scaling on a Rover SPi sensor my self, I'll see if i can find the data

Published data is not always representative of the actual sensor as the cold junctions are usualy encased in resin and brass, scaling is just for the cold junction only

Thanks Sprocket, if you can find the data you measured for the Lucas one, that would be most helpful.

I take your point about published data vs actual and I'll measure my own one once I get it (it looks like Megasquirt only requires three readings to build its "curve" so that's easy enough) but I'm trying to decide whether to buy the Lucas one or the Bosch one and the Bosch data is very different to the General Motors sensor(s) for which Megasquirt is initially programmed so I'm wondering whether the Lucas one is closer.

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

the rover sensor is a funny thread,

I did podt whjat it was but can't remember where,

I think its 1.25 mm pitch

I found with VEMS that most use Bosch sensors and there are several files already available for those. The lucas one compared to pretty much any of the bosch ones is a much shallower curve, so, if we overlay the data on a graph and have them crossing over in the middle, the lucas one reads high at low temp and low at high temp, thats if you were to use one of the bosch files. We are talking around 5 degrees in total

Its not really a problem, its just that we are humans and we tend to see the error and try and fix it, when in all honesty its not really required. As long as the ECU sees a changing reading that is consistent with temperature, it doesnt matter what it is, you can tune round it.

My advice is, if you are starting a fresh instalation, use the bosch sensors, its easier in the long run.

I really dont know why they dont write software for Megatune so you have say ten points were you can enter the actual resistances against the temperature. Surely that has to be easier that all this easytherm bollox. Thats at least how some of the major PLC manufacturers do it on unknown sensors.

Thanks Sprocket, I think I'll use Bosch, if only because all the information is readily available from their website, including thread size, and (mini13), if the lucas one is a less common metric fine, I don't want the hassle of trying to find a tap (at least M12 x 1.5 is relatively easy to source even though it's not standard M12)

And yes, I am starting from scratch.

I do find it strange that Megasquirt only wants three points, but at least with MS-II (which I have built) you can actually enter the data into Megatune direct rather than having to use Easytherm.

Cheers,

Rod.

Edited by Rod S on 29th May, 2008.

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

hang on,

I think I'm talking shit.

i'm getting connfused with the rover IAT sensor, which i used,

yep talking crap,

as you say the bosch water sensor is M12 x 1.5,

my rover air temp was (I think) M8 x 1.25 (an odd tpi IIRC)

No problem,

After much research I think that most of the EMS water sensors are M12 x 1.5 (one version of metric fine) and most EMS air sensors are M14 x 1.5 (again an odd metric fine). That is the open element air sensors seem to be M14 x 1.5 and most turbo applications seem to use the open element type, presumably for faster response.

At this rate, I'll be spending more on non-standard metric taps than the sensors themselves !!!!

And I've changed my mind again since this morning, I'll actually use Marelli sensors as (a) I have a spare set of Cosworth ones, and (b) I have the factory manual and it gives enough resistance/temperature values in the diagnostics section to be able to tell Megasquirt what little it needs to know.

And I'll measure the actual ones at three points anyway to confirm.

Rod.

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

Just wanted to add in that I spent several hours today plotting the temperature vs resistance of the brown Elmwood ES120-4 used on the mpi motors and If anyone wants to see the data I have attached it.

There are two sets of data, one where I started it in ice and warmed up and the other is where it was warm and I cooled it down. All done by hand and recorded and plotted on graphs.

I made a note about the resistors needed to modify the MSII board to function properly with the sensor. Doing so will keep you from having to update megatune and I too think it is silly that they only have 3 points, how can you extract a good curve for an exponentially increasing/decaying resistor vs temp. I did a further analysis to see how close they could approximate these graphs and it was surprisingly close (100ohm difference)
assuming they used an exponential equation to generate the info, if not then its going to be off.
8-)

EDIT: removed attachment for revised one in second post

Edited by dhaines on 18th Dec, 2008.

"sorry, Jedi mind tricks don't work on retards" Mediocre Films 2008

Actually, they use the Steinhart-Hart equation to establish the table of temperature vs resistance from the 3 points. You can read more about it here: http://not2fast.com/megasquirt/doc/thermistor.html.

Jean

http://www.jbperf.com/

Good call Jean. I redid my analysis to include the Steinhart-Hart equation and plotted the measured vs exponential vs steinhart on one graph. I also noted the resistance at 81 deg F which was about 100ohm between the exponential and steinhart, however the exponential was the same as the measured at this temperature point. I dont know which one I like more, the stein or exp. Mabye I can try and get some hotter and colder data.

Edited by dhaines on 18th Dec, 2008.

"sorry, Jedi mind tricks don't work on retards" Mediocre Films 2008

Interesting work. Struggling with degrees F though.

However, it is not worth getting too hung up about the accuracy of the sensor as far as the ECU is concerned.

You will need to trim the warm up curve to your sensor. Usually the ECU will ignore the coolant temp at anything above 80 degrees C. So as long as it is reasonably consistant between -10 C and 80 C that will be fine.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

I messed up the graphs in the previous attachment. This one should be much clearer. I also plotted the log and exp against stein. The stein does a good job at getting the sensor data correct as it gets colder but on the hot side its mediocre.

I know that it is not worth worrying about a few ohms difference but I wanted to know how close the approximations would be to the measured data.

Sorry for the Deg F, I guess I could have changed it but we still use it over here :$

I understand why the ecu ignores anything above 80 C or 176F. The steinhart equation goes crazy just above that point. You can see this in the graphs where it literally goes straight up and also in the data chart.

Attachments:

60.51kb

Edited by dhaines on 18th Dec, 2008.

"sorry, Jedi mind tricks don't work on retards" Mediocre Films 2008

The other thing to note is that the MPi sensor is still a NTC 2.5kOhm sensor, so the standard build Megasquirt resistor will be just fine.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

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

Im over here in the US. Colorado-which is why I am going with the baro correction.

I thought that the resistors on the board had to match the coolant sensor at 81F??

from the megamanual: "To avoid changing the tables in the MegaSquirt® code with EasyTherm, the bias resistors at R4 and R7 should have a value equal to that of the thermistor sensor you will be using when it is at 81° Fahrenheit, 27º Celsius. However, the preferred method is to use EasyTherm whenever possible. "

"sorry, Jedi mind tricks don't work on retards" Mediocre Films 2008

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

Edited by Paul S on 18th Dec, 2008.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."