low engine coolant alarm

[BrendanP]

I'm bread boarding a low engine coolant alarm as a demonstration to one of the fitters where I work of picaxe in operation.

The sender (an off the shelf part) is essentially a metallic probe that protrudes into the header tank of the vehicles radiator. The probe is mounted in a nylon fitting that screws into a bush that has previously been silver soldered into the header tank. The header tank itself is grounded to the vehicles chassis while the probe is electrically insulated by the nylon fitting that it is mounted in.

When the radiator is filled the water/coolant act as a conductor between the probe and ground. When the coolant level drops and the probe is exposed the circuit is open.

What would be the best way to monitor the state of the circuit? The simplest way would be to simply treat the probe as a switch that is either open or closed. But then I'm thinking that different coolant/water mixtures might have different levels of resistance, a high resistance might be seen by the picaxe as a open circuit leading to false triggering.

How would I do it with adc so I could add some 'smarts' to the thing with software. Im thinking that one way would be to charge up a electrolytic cap and then read back the charge in the cap with adc. If the circuit was open (because the coolant was low) then the cap wouldn't charge was well. It might charge a bit because of a sheen of water on the surface of the probe.

 

[Dippy]

I suppose 'Stamp' users would use an RC method - not quite as you have described but similar principle.

Why not consider your thing as part of a pot div? And ADC the mid point.

I don't know the resistance/impedances involved, so it may require a simple op-amp circuit before ADCing.
It may require a bit of noise/transient filtering. But not Rocket Science.

Is the sensor in a position where contact may be lost due to vibration/splashing?
If so a little extra thought required but solution easy in code.

Will it change much with temperature?
Maybe you can make it trainable?
Sounds a nice project.

 

[MPep]

Hi there,

Would this help? Try this

MPep

 

[BrendanP]

Thanks guys.

I don't anticipate that the probe will ever be exposed to air except in a valid alarm condition. This is because the vehicle (a 50 ton kato crane) has a header tank fitted above the radiator. The header tank is about 250 mm X 250mm by 300 wide. The probe is fitted at the bottom of it so that no matter how hard the vehicle corners, bounces around etc the probe is always covered. The idea is that if even if the header tank gets low and the system alarms the radiator itself is still full, thus no cooked/seized engine.

Temp will vary from 0 ish in the winter on start up to bloody hot in the summer, some where below boiling I imagine, 90C maybe?

I have just lashed together a rats nest on a bread board that Stan would disaprove of. Any way I've got a 10K R in series with the sensor (potential divider set up with the sensor acting like a variable R depending on if its conducting to ground via the coolant) and measuring the connection between the two with adc.

I silver soldered the sensor into a empty new paint tin today at work for testing.

Its working ok on the bench. Im going to go boil the some water and see if hot water makes a difference.

can anyone see glaring fault in this approach?

PS. Ive just added boiling water to paint tin, set behaves the same as with cold water


Test code below.

high 1 'test piezo
pause 1000
low 1

main:
readadc 2,b1
if b1>150 then goto alarm
pause 100
goto main

alarm:

for b0=1 to 8 'use b0 to count up to 8 in this for....next.... loop
high 1 'sound piezo
pause 30
low 1'turn off piezo
pause 30
next b0
pause 200
for b0=1 to 8 'ibid
high 1 'sound piezo
pause 30
low 1'turn off piezo
pause 30
next b0


goto main

 

[Dippy]

That sounds a fair first test. I never comment on code.

Question 1:
Pre-amble: I know nothing about the insides of the coolant system, except they start off hot and clean and end up hot and grubby.

Will you have to investigate the inevitable drift as the probe and metal bits inside get covered in gunk/oxides and other crap?


I only ask this as the 'Real World' can be dirty and the only cars I've had with coolant level warning have relied on a float switch in the header tank.

I'm just thinking of the peformance weeks/years down the line.


Question 2:
Is this just water or water+coolant/anti-freeze?
How much does the ADC vary when coolant is missing?

 

[mst]

usually coolant also has some dielectric properties in an attempt to stop electrolysis

 

[boriz]

Use an AC probe signal. IE: The voltage on the probe oscillates equally above and below the voltage on the tank so that the electrons travel in both directions. Otherwise oxides and impurities in the fluid will collect on the positive terminal (probe) over time, reducing its electrical contact with the fluid. The rate of engunkification will depend on the magnitude of the current and the quantity of impurities. Remember, pure water does not conduct.

Only measure the conductivity in short bursts, a couple of seconds apart. This will also significantly reduce engunkification. (By a factor approximately equal to the duty cycle percentage. You should be able to get away with maybe 0.01% or less)

Use high quality stainless steel for the probe.

 

[moxhamj]

Very good advice from boriz on 'egunkification'. BTW I like that. A new word that sort of encapsulates electrolysis and electrodeposition and electrolyte acid/alkali reactions all in one word!

AC and infreqent sampling and very low current sampling? boriz might want to expand on that. I'm not sure how one would do AC ADC sampling for instance, but it ought to be possible.

 

[D n T]

Countering engungification

If the gunk in the cooling system settles on the probe, as it does everywhere in an engines cooling system, will it alter the ADC reading? If it does you might consider monitoring this as a secondary function at the start of your code. If for example you have enough build up to alter the initial reading you might have the piezo beep twice or something. You would have to add a probe ADC test section in your piezo section. If the percentage of ethylene glycol alters the ADC of your probe you could monitor that, another handy angle, never tried it, if you do, post the results.

 

[Dippy]

Thats what I alluded to in Post # 4 D&T.

I think this needs some deep thought...
I'm not sure I'd be happy relying on a simple electrical conductive system (whether AC or DC) when used like this and over a period of time.

 

[hippy]

I'm not sure how one would do AC ADC sampling

One could possibly connect a potential divider across +V/0V, centre tap to an ADC input. Then from sensor tap via sensor resistor to an output pin. You can then read ADC with the pin high or low and that will reverse current direction.

 

[kevrus]

Ive come across stainless steel probes that are used as level sensors in condense pumping sets to control the pumps/high level alarms with quite dirty condense at high temps..some needed cleaning approx every 18 months, (although some didn't??)

As a matter of interest...or not...a similiar principle is used to measure water conductivity which corresponds to TDS (Totally Dissolved Solids) hence controlling water dosing equipment on industrial boilers

 

[MPep]

In the post I linked to, AC is used, from PWM, then the 'receiving' probe has a rectifier which then goes to ADC input. Check the circuit and code out, they should give enough of an idea to carry on.

I am not certain about the longterm viability of an 'active' probe system.

"engunkification": - what a fantastic word!!!

 

[BrendanP]

Well boys, nothing is ever as simple as it first appears to be it seems. I hadn't even thought about the "gunkification" issues.

All the vehicles in the fleet have anti corrosion/glycol in the cooling system. The mechanics/fitters have a tool they measure the concentration of the mix with and then add more as needed when they do the schedule services.

When we do a prestart on the crane we are using that day we have to actually remove the radiator cap and look to see that there is coolant in the radiator. They always look to be very clean. You don't see that greasy crud coating the inside of the cap or tank like you do on older cars. You have (meant to) to fill in a log book on the machine each day confirming you have done the inspections.

Despite this some people simple tick the boxes and don't look. This happened recently, the guy ran the machine down the road with no coolant (radiator hose failed) and cooked the engine, a $30K rebuild and the crane not working for a month.

The latter model machines have the alarms factory fitted. I mainly lashed this thing together as a demo to the fitters at work of what can be done with mcu's and also to show roughly how the factory units work. I've got a back lit LCD displaying a message as well now. Due to the advice I've received here I'll also be able to touch on some other issues to consider in use of these devices.

I have examined the probe that is used in the sender I bought from a radiator repair place and it is stainless. Its actually just a stainless machine screw about 50mm long and 6mm in diameter with a phillips head on it.

I had already thought about the adc reading changes with the glycol/anti corrosion soloution as opposed to just plain water. I'll take the set up into work monday and do some tests. I silver soldered the probe into a small unused clean paint tin that I fill with water and pour out to test.

I will also change the adc sampling rate to every 10 seconds or so.

 

[Dippy]

That would be a good experiment.
Also, different Glycol/Whatever : water ratio too.
Also, hard and soft water I would have thought.
Your lads don't use deionised water do they?


I hope we don't end up confusing things here...

Some people have mentioned AC or toggling pins to construct AC. A 'must' for reliable long-tem sensing I would have thought.

But Brendan is talking about a single probe (I inferred from his first post) and, essentially measuring the current through the coolant down to ground via all the grounded metal bits - somewhat tricky using a simple pin-toggling method for 'AC'. But not impossible with extra electronics. However, I do prefer the 2-prong approach (haha).

I would have thought that, regardless of the visual cleanliness of the coolant that some coating would happen on all surfaces(?). As to whether it would be signicant compared with Fluid/No-Fluid ADC value I don't know.

Perhaps look at delta R as well as Abs R ?
Or keep an eye out (in code) for time-related drift.

I take it you've had a good old Google to see if others have tried long-term measurements like this?
I'm sure many have had a similar idea over the PIC era and I'm sure that using an electronic method isn't impossible.

But, good luck. Its an interesting project and I'm sure you'll get something working.

 

[D n T]

Apologies if I state the obvious

Have you tought of connecting a temp sensor to you unit, I can send you a dallas on as per the PICAXE website if you want one or two or three. Because I have heard of situations where the coolant alarm has failed and the temp gauge is in air so it doesn't register the coolant temp. This is the kind of thing that can happen as you drive down the road. If you blow a hose that is high up in your system, like the top hose, your reservoir tank might still have coolant in it when the pistons have grown just enough from the heat to lock themselves securely in the bores and seize the engine, ching, 30K rebuild.
The fitters would appreciate the temp gauge and digital reading.

Off the track a little bit, you might want to consider another device as well, an O2 meter, using a 5 wire O2 sensor, because you need the heater circuit, because these help when tuning to get the right fuel air mix.

 

[Dippy]

I had exactly that DnT when I had a hose failure in my car. i.e. the engine overheat sysmbol came.
So I can see what you mean.

But the temp sensor is in the engine block where heat is being generated.
Would it work in (or just below) the header tank?
I dunno.

Whatever is done it is going to have to be tough with very strong physical connection and seal.
I just hope Brendan doesn't think a cork and Knitting needle will suffice
(Kidding by the way.)
Seriously though, the only proper experiment is in-situ in some vehicle - with all the rigours that involves. (OH NO, not the powering-a-PICAXE-from-a-vehicle-supply-panic again!)

 

[D n T]

temp sensor sensing

Dippy
When a temp sensor is operating in normal conditions it is immersed in fluid . the sesor is mounted in a thin metal jacket which is in contact with the fluid. This fluid heats up the jacket thus sensor through contact.
If there is no fluid, the sensor and the jacket are heated from the engine block, there is usually a porcelain insulator between the sensor and the threaded body, so the block temperature has to be higher than the coolant temp to register the same temperature.

You can use a surface temp sensor and mout it on the back of the block ( to point farthest from the radiator and thus the hottest, but you will be reading a a bit different ( higher) to the coolant temp because the block transfering the heat to the coolant.

I reckon I have seen and sold the type of probe Brendan is using, it will definately seal.

 

[Dippy]

Yeah, I realise all that DnT, hence why I mentioned it and mentioned heat.

But Brendan, I believe, wished a sensor in or just below the header tank (?).
Hence why I'm not sure that will work. No immediate heat source.

And surely you want to know if the fluid is low rather than waiting for it to get 'lost' as far as the block? Surely you'd want to know ASAP?
Well, I would anyway. In fact I'd want to know as soon as it was low in the header.

 

[mst]

You can use a surface temp sensor and mout it on the back of the block ( to point farthest from the radiator and thus the hottest, but you will be reading a a bit different ( higher) to the coolant temp because the block transfering the heat to the coolant.

Block surface temp realy doesn't have much to do with temp of the water. An important thing to remember is that coolant temp is just what it says it is, the temp of the coolant. Oil temp is a much better indication of what's going on with the motor as it's in direct contact with the parts your concerned about and also part of the reason why things seaze up (oil too hot then it goes out of spec), if were worried about the pistons seazing in the bores or spinning bearings from overheating we should be looking at the cooling/oiling systems temp which are directly resposible .

 

[Dippy]

I'm not sure we're talking about the same thing here.

DnT's point was that if you remove/lose coolant from he block then the block temp goes up.
Obv. as it is heat generator. NSS!

Or maybe I'm not following you mst?

Anyway, I really don't think any of this is much to do with Brendan's sensing system, but I'll leave you to argue the point

If you want to warn someone after coolant has been lost from block then I'll stick with someone else's design

 

[mst]

Sorry, I'll try make more sense

Block surface temp has no relavence to coolant temp. How fast you are driving (block to air cooling) and how hot the day is will change the block temp even if the coolant temp is rock solid.

Coolant and external block temp won't rise untill you loose alot of coolant.
Imagine you have 2 coke bottles that are 600ml each. They are connected at the top via a hose and at the bottom there is a pump pulling water from one and pushing it into the bottom of the other. Now only fill the bottles with 700ml of water. When the pump is switched on the pump will almost empty one of the bottles as it fills the other at which point the overflow from the full one feeds the empty one just enough to keep the pump going.

Same happens inside your motor. The result is total coolant temp starts to rise as the radiator no longer being full of coolant is no longer efficient. All parts of the block recieve the same 'input' temp water so you will not see a diference front to back that is relavent to coolant temp.


Edit: If your worried about seazing motors perhaps you could add a oil temp sensor to the project also!

 

[D n T]

hot engine

MST, I agree totally