I have just removed my Electromotive coolant sensor (GM part) which was located in the coolant pipe exiting the rear of the head. It is quite large, the pipe of relatively small ID and therefore restrictive in my view. Anyway I've handed responsibility of supplying the TECgt with coolant temp information back to the OEM coolant sensor. I'll need to shift my warm-up curve a little and I'll also need to think about the temperatures at which my radiator fan switches on and off. To help me do this it would be useful to know the temp vs. resistance curves of the two sensors. So does anyone know the response curves of the two sensors or where I could obtain them? FWIW, after a few days going nowhere, at 19C (ambient air temp) the stock coolant sensor was reading 27C on the TECs lookup table. If my memory serves the difference reduces with increasing temp. Edit: Electromotive sensor curve is in the TECgt manual (Dohhh.). Still looking for the stock (Mk2) sensors curve though.

The curve should be almost the same. In fact hydra uses the GM/Delco calibration for the mazda sensor on the NB/MK2.

also, it's in the FSM with a graph.

Thanks. My FSM only provides two data points to test the sensor:
20C = 2.21 - 2.69kohm
80C = 0.287 - 0.349kohm

Could you post the graph (if it's not too much trouble)? My FSMs 2 data points are a little sparse. I was sorta hoping that someone (as sad as me) may have tested the sensor in a saucepan to test temp vs resistance...

Hi (again )

If the sensor is the same as in my Protegé, and from your two data points it seems to be, my Mazda FSM says:

-20°C = 14.6-17.8kΩ
20°C = 2.2-2.7kΩ (matches your FSM)
40°C = 1.0-1.3kΩ
60°C = 0.50-0.65kΩ
80°C = 0.29-0.35kΩ (matches your FSM)

--Ferdi

what year miata are we talking about?

My car is a bit of a Frankenstein, but the head is a '99 VICs with stock sensor

99 stock sensor should be:

deg C Ohms
0 6000
40 1100
80 320

Spent 3 hours today very slowly heating up a pot of water. My digital thermometer (±1°C accuracy) was spot on 100°C when the water boiled.
Interesting enough, I couldn't get the water hot enough to go over 100°C.
This was measured on the 2 pins for the ecu, not the one pin for the dash gauge.




I can also confirm that my deepfreezer is -21°C and my refrigerator 4.5°C

Fabulous stuff Frank. Looks like the error increases as temp falls, good that the two sensors are similar at higher temps so going closed loop etc is not affected. May need to adjust warm up curve and coolant ignition advance curves at lower temps though.

You nailed the 100C data point, are you similarly confident in the 0C data point as well?

The first tests below room temperature were done with the sensor wrapped in a deepfrozen hot/cold pack, but that proved to be not accurate enough, so I'm redoing it with various fluids in de deepfreezer. You'd be amazed at how fast a box of icecream heats up when you take it out of the deepfreeze. In a matter of seconds, the resistance of the sensor changes (don't tell the wife I stuck it into the icecream).

I changed the results above according to today's testresults. These are very accurate. At 0°C (water going into frozen state - half water half ice), I have 5850 ohm. The digital thermometer read 0.3°C at that point. It stayed at that temp for over an hour. Interesting to see what happens to water when it freezes and how long it stays at 0°C during the freezing process.
Will do the part below 0°C tomorrow

you can't use a saline ice bath? those are 100% wife compatible.

Yup, I have one of those in the freezer right now. I'll have a look in the morning (3.40 AM overhere right now).

Just did the part below 0°C.
Next is the GM sensor.

Hey Frank - had a chance to look at that GM sensor yet?

Awesome work, I tried this once (I thought with Joe P, but I don't recall now) and got nowhere. :-)

This is a very common thing, must have been too long since you've been in highschool and taken chemistry. The hotter molecules leave the pot immediately, hence it's effectively impossible (at normal pressures) to get water over 100C. That's why it's a good reference, get it hot as you can, it'll be at 100. Same for freezing, you can't get it below freezing and be a liquid... unless you put antifreeze (or salt or...) in it. Any additional energy comes from forming ice - which is why it takes so long to melt. (Also, the ice can be colder than 0, water cannot)

I forget the numbers, but roughly in going from 0 degree ice to steam, 1/3 of the energy is the temperature difference, 1/3 is melting energy, and 1/3 is boiling energy. Perhaps it's 1/2, 1/4 and 1/4, but the "latent heat of melting/vaporization" is significant!

Actually, it looks to me to be most inaccurate at the higher temps. An error of 100 ohms at boiling is 14 degrees! An error of 100 ohms at ~30 degrees temp is 1 degree. You need to have single ohm accuracy at the high end of the scale to figure out where you are, but at colder temps even a rough idea is fine.

__________________
You ain't got no problems, Mt.net'ers. I'm on the ************. Go back in there, chill them ******* out and wait for The Wolf, who should be coming directly.

Ha, I took me forever, using the smallest flame possible, keeping the thermometer suspended in the pot, making sure not to touch the bottom (hotter) and stirring it all the time with a teaspoon. I don't recall the details, but it was very interesting.

Haha, yes, I left highschool some 25 years ago, so you could say it's been a while. Although I'm an engineer, I had forgotten all about that and it only came back when I found I couldn't get the temperature over 100°C. Water is very interesting at 0°C and 100°C.

slowmx5 was referring to my first graph which wasn't smooth around the lower temperatures. That was because at first I used a cold/hot pack that I wrapped around the sensor. Not good at all. The later tests were done with a bottle of frozen water in which I drilled a hole, tapped with M12x1.5, added water to make sure there was no air and screwed the sensor in it. I also drilled a small hole in it for my thermometer. Everything went back in the deepfreeze for a day or so. I also used bottles of salt water etc. Took me forever to do.

I will do the GM sensor as well one of these days.

Dear god, too much work! :-) BTW, you can heat water hotter than boiling or colder than freezing by adding dissolved particles - sugar, salt, or... surprise: Antifreeze! When I did this I was able to get much above 100c. I didn't go for cold, but you can.

Honestly, I doubt my motor ever sees 10*c temps. :-)

You can use high pressure, too (just like your cooling system does - a mix of pressure and antifreeze to prevent boiling). One of my measurements (I lost come coolant but it was just for a test) was to stick a very small thermocouple under the hose clamp and into the temp sensor housing outlet. That's how I calibrated my water temp dash guage.

__________________
You ain't got no problems, Mt.net'ers. I'm on the ************. Go back in there, chill them ******* out and wait for The Wolf, who should be coming directly.

Mine does, that's the reason I wanted it to be as accurate as possible in the higher region (1°C data points).
I typically cruise at 95-96°C. Spirited driving sees 102°C and testing my topspeed instantly gives 104°C on a hot day.
Did a coolant reroute yesterday but haven't tested it yet.

slowmx5: do you have a re-route?

2 points on a resistance vs. temp curve are enough to generate the entire curve, because it follows an exponential equation with 2 parameters.

I'm too lazy to find it, but google "NTC curves". Philips IIRC had an appnote with equations.

Add 1 more point if you are doing voltage vs. temperature and you don't know the pullup resistor value.