Automotive radiator cooling fan contoller

[MrAl]

Hello again,

You need not worry too much here. To put this project into proper perspective from a new design point of view, if you looked at all the control systems in use in the world today and not even any in space, you'd find that a fan controller circuit is at the very low end of the spectrum meaning it's one of the simplest control circuits you can build. It might seem like there is a lot to it, but consider trying to design something like a CNC machine control circuit.

The automobile layout would start with the control circuit and drive circuit. The control circuit would be mounted inside the passenger compartment while the higher current section of the drive circuit would be mounted under the hood. That's to keep the high current wiring short, while also allowing the control circuitry to stay inside where the environment is more favorable. Thus, the inclusion of IC parts is not a bad idea at all, and because of the simplifications and repeatability that they offer it would be a bad idea to sway away from using IC's at all. Add to that the fact that many of them can be found as mil spec parts, and you get a pretty reliable and repeatable circuit that you can depend on for years.

There's also the chance that you dont need the full power of the motor. You could do some tests to find out.
If you use PWM then you dont have to use the two windings independently either, you can rely solely on the PWM to do the speed control and/or soft start.

My suggestion would be to start with the simplest circuit, install it, get it working right, and then come back to a more sophisticated controller design. The only change would be to swap out the controller. This phase would give you a chance to get a feel of how it was going in general when mixed with the other workings of the automobile like your alternator. It's up to you, but this is probably the best bet.

PWM adds complexity via a triangle generator and comparator. This isnt that complicated though, a few more parts. PWM is necessary for either variable speed and/or slow start.

One thing about a window comparator. Strictly speaking, a window comparator puts out one logic level when the signal is between two set levels, and another logic level when the signal is outside these two levels. So that in itself does not make up a fan control circuit because we want slightly different operation. We want the output at one logic level when the input goes above some set point, and we dont want it to reset until the signal input goes below another level, where the second level is lower than the first. So we have to use a sort of modified window comparator.

For a quick illustration, see the two attachments. In the schematic, V2 is a test voltage used to test the control circuit. The output that would be used to turn the fan on and off comes from the output Q of the flip flop.
In the other drawing, the result of testing the circuit is shown where V2 is varied up and down and the output recorded. It's just a test of the control circuit.
See if this makes sense to you. Note that the sensor in this circuit would have an output voltage that DECREASES with increasing temperature, such as a properly biased silicon diode. This type of operation can be easily changed to an increase with increasing temperature though.

 

[KeepItSimpleStupid]

Fan Control
1. AC forces fan ON especially when still.
2. Fans in cars with electric fans sometimes go on after the car has been turned off. Lack of circulation and latent heat I suppose.

 

[alec_t]

I'm unclear why you want to use speed control at all. I thought the purpose of the fan was to get the coolant down to the correct working temperature as quickly as possible and maintain it there. That, to me, indicates the fan should work at max speed as soon as over-temperature is detected and then shut off completely as soon as the working temperature is reached.
And what is the supposed advantage of soft-start?

 

[MrAl]

Hi,

Speed control is really an alternate method of overall control, and would not be needed in a bang bang type setup where the fan just turns on and off at different temperatures. Speed control is a type of control that *constantly* controls the voltage/current to the motor, and could be called a 'linear' control method whereas the bang bang setup could be called a 'digital' method.

Speed control controls the motor speed during all times that the engine is operating, except during the first few minutes of startup. Once the engine heats up, the speed keeps the radiator at a certain temperature like 200 degrees F. If the temperature rises even a little bit, the speed of the motor is increased so that it does not rise any more. So for example if the temperature went up to 200.01 F the motor speed would increase slightly, but not enough to bring it back down to 200.00 F. It would however increase enough to keep it from rising to say 200.02 F which is good enough. So there is still a variation, but it is now very small. It's a linear feedback system and works the same way a voltage regulator works so the steady state error depends on the gain in the system.
The nice thing about this kind of control is that the sensor and most parts of the measurement system are always subject to close to the same levels (temperature, voltage, etc.) so there is little variation in measurement which could make the system very accurate.

The bang bang type of control is of course much simpler which does not require PWM unless we want soft start too.

Soft start is used to prevent banging the motor with the full stall torque current. When the motor is stopped, it takes more current to start it up then it does when running. To prevent it from getting the increased current level (stall current) a slow start mechanism would be used.
Just how much this helps the motor though might be subject to argument. Motors are pretty tolerant creatures but then again the brushes might be subject to breakdown with increased current. If the motor does not have brushes then all the better.
The problem is though that soft start really requires PWM, so i am not sure if it makes much sense to use soft start with the bang bang system because if we already plan to incorporate PWM then why not go all the way and do the full speed control system type control, which then would be simpler than the bang bang system (one additional op amp maybe).

 

[alec_t]

I'm aware of the two control methods and agree fully with what you say, MrAl, but the OP seems to want a system which combines features from both bang-bang and proportional control. That seems OTT for a car coolant system (but might be justified for, say, an industrial process requiring close matching to a particular temperature profile).
A compromise would be to make individual use of the two windings on the fan such that both are switched in at highest temp, one drops out at intermediate temp, and both are off at lowest temp: presumably why there are two windings .

 

[MrAl]

Hello alec,

So you think he wants to use both windings with a slightly more complex switching scheme?
That seems reasonable. But i will add again though that if PWM comes in at any point, then the more complex switching scheme is just more complexity for nothing.

So if we assume he wants a "full bang, half bang, no bang" type control (bang bang bang, ha ha) then i guess we can add another switching state to the bang bang technique. That would switch winding 1 on first, and if the temperature still rises then switch on winding 2.

What i dont think he specified though is whether or not these two windings are individual windings or if they share a common ground or common positive terminal. If they are completely isolated windings, then we can achieve slow start by connecting the two windings in SERIES. This would lower the start up current, and we'd also have to obtain some specs on that mode of operation too so we know if that could supply another level of speed. If that mode is available, then we just switch to one winding after a short period, then if the temperature keeps rising then switch to both windings in parallel. Would be pretty cool if we can do it that way. That way we dont need PWM either.

If the SERIES mode is not available, then we just switch on one winding and then the other if the temperature keeps rising. So we'd have three set point. I guess we could implement a forward control single conduction angle PWM, which would be simpler then a full blown PWM, and that would only be for start up.

Sound good?

 

[KeepItSimpleStupid]

I agree totally. Don't forget you can;t get the temp right on, unless you use a model based and/or PID control or some "derivative".

Off when staring the vehicle.
Off until t> T1
T > T1 then winding 1
T > T2 then winding 2

AC on then both windings on.

Something else? when the car is turned off. e.g. run full speed off a fixed sensor that requires no battery power. T> T3.

Is my vote.

 

[alec_t]

So you think he wants to use both windings with a slightly more complex switching scheme?

I think the OP was planning to parallel the windings and then have a combination of on/off switching plus PWM.

What i dont think he specified though is whether or not these two windings are individual windings or if they share a common ground or common positive terminal.

Actually in post #1 he says "I have a two speed fan motor with two windings that share a common ground." So that rules out a series connection and also means high-side switching would be needed for individual winding control.
Although maintaining a reasonably constant coolant temperature is an obvious target , in theory should temperature be varied according to engine rpm or other factors? My knowledge of engine physics is pretty poor.

 

[KeepItSimpleStupid]

I've never seen parallel connections. Usually it's a tapped winding. Parallel windings don't make any sense and a two phase brushless, I guess is possible. alec. Think Joe's pumps.

 

[3v0]

As I see it a PWM solution to keep the temperature constant is as good as it gets. Nothing else talked about here adds anything but complexity. I am willing to listen if I am wrong.

I am not a mechanical engineer but I expect engines need to be kept at a constant temperature to maintain optimal clearances between moving parts etc. So one temperature for all RPMs

 

[KeepItSimpleStupid]

I am not a mechanical engineer but I expect engines need to be kept at a constant temperature to maintain optimal clearances between moving parts etc. So one temperature for all RPM

Yep, the point is to get it to the operating temperature as quickly as possible and don't let it overheat.

 

[alec_t]

Usually it's a tapped winding. Parallel windings don't make any sense and a two phase brushless, I guess is possible.

We need confirmation of the winding arrangement.

 

[KeepItSimpleStupid]

Reminder, earlier in this post I showed a Velleman PWM controller kit that I have used in the past.

 

[MrAl]

I agree totally. Don't forget you can;t get the temp right on, unless you use a model based and/or PID control or some "derivative".

Off when staring the vehicle.
Off until t> T1
T > T1 then winding 1
T > T2 then winding 2

AC on then both windings on.

Something else? when the car is turned off. e.g. run full speed off a fixed sensor that requires no battery power. T> T3.

Is my vote.

Hi Chris, Alec, and 3v0,


Chris:

Yes i was thinking of a similar scheme:

Off when starting.
Off until T>T1.
When T>T1 first occurs, run windings in series for 2 seconds, then winding 1 alone.
When T>T2, then windings 1 and 2.
When T<T2, then winding 2 alone.
When T<T1, then all off.
The "2 seconds" is subject to adjustment after experimentation.
Running winding 2 alone instead of 1 again is just to share the run time between the
two windings as that might have a positive effect on longevity.

I like the AC on then both windings, that would probably be a good idea unless it can
work ok without worrying about that.

Not sure what you mean about that last part though, when the car is turned off.
You mean have another sensor like a mechanical switch?
Do we know that it is necessary to cool the radiator after the car is turned off?
There is no water pump circulation with the engine off unless it is electric too.

Also, for the PID controller, the derivative does not help with the steady state error it's mainly the integral that reduces the error to near zero. That's how most voltage regulators work and why they can be so accurate. The derivative control part is mainly to improve the dynamic response during times when things change. For a slow control system it most likely would not be necessary to use derivative feedback because the system can not be forced to change any faster than the fan can cause the cooling when it is run at full speed. In other words, the fan would be running at full speed already so derivative feedback would not help anything. The fan running at full speed means the drive system is saturated, which means it cant respond any faster.
But the integral part would be necessary of course, to get the error down to near zero, if we are using PWM. For the bang bang bang system though we dont even need that.

Alec:
Oh ok, you found a post stating that the two windings share a common ground. Well if that can not be changed via disconnecting the leads somehow then we are stuck with just the two windings in parallel or whatever. I am assuming that he tested the motor already by energizing the two windings and noting the effect on the running speed and making sure the motor does not overheat in that mode. It could be that only one winding at a time is supposed to be energized too, but i guess we need confirmation of this as you said. We'll have to wait for that i guess.
From what i know about the car's built in thermostat, the engine runs until it gets up to a certain temperature and then the thermostat opens. The only variation i know of is that they used to make thermostats that you could get for the winter which would keep the engine temperature slightly higher so the heater worked better. I dont know if they still do that however, they probably dont anymore. So i would assume a constant temperature.

3v0:
Yes I think PWM is the best bet too, but the other system (bang bang) is actually simpler because it requires less parts for the control circuit and the theory of operation is much simpler too.

 

[tyrebyter]

This car (1976 Triumph TR6 automoblie) does not have air conditioning.

This is what I'd like to have :

PWM - variable speed control on lo-speed because it is starting from rest. Highest torque at stall speed, maximum draw. hi-speed is either on fully or off.

Maintain (adjustable) 155 - 200 F set point with lo-speed. +- 0.1 degree F tolerance. Fan soft starts 1 degree cooler than set point.

Turn on hi-speed when temperature goes 2 degrees F above set point. Turn off hi-speed when temperature falls below hi-speed turn-on temperature.

Turn off lo-speed when temperature falls 1 degree F below set point.


This is what I have :

The temperature sending unit is mounted immediately below the thermostat. It screws in and measures the coolant temperature where the coolant is hottest and also very close to where the coolant exits the engine and goes to the radiator.

3-wire fan motor. 2 (+)tives and 1 (-)tive or ground. Connect either one of the (+)tives to 12 volts DC for lo-speed. Same lo-speed RPM/power/draw regardless which +(tive) is powered. Connect both (+)tives to 12 Volts DC for hi-speed.

I cannot determine absolutely, if this motor has brushes or not. I think it probably does.

This is what I'm learning :

It seems that transistors make very good DC switches since they can easily be made to work as a quick acting, variable resistor. They apaprently work better in the AC realm when it comes to varying the voltage they output.

I'm beginning to see the advantage of using an OpAmp or posibbly another ic since transistors do not seem to be easily configured to output a controlled but varying voltage.

This is what I'm working on :

I'm considering a radical redesign of my original posted circuit for the sake of simplicity. A single MOSFET reads the voltage from the gauge/sender circuit and switches the fan hi-speed on and off at a single set point, via a relay. I'll post it in a day or two.

 

[KeepItSimpleStupid]

Maintain (adjustable) 155 - 200 F set point with lo-speed. +- 0.1 degree F tolerance. Fan soft starts 1 degree cooler than set point.

That tolerance is unrealistic.

PID was used in a general sense: P, PI, or PID

You have to have, probably PI control to even come close.

Aside: PID means Proportional Integral Derivative
With proportional only, there is always an offset with measured and setpoint.
Integral control adds or subtracts the integral of the error term thus making the setpoint and measured value agree.
The derivative term tends to prevent overshoot.
A system is "tuned" to get the right constants.

I'm from the old days. Plenty of experience with tuning by hand with a strip chart recorder. later, controllers had built-in tuning. A setpoint disturbance is required to tune. I have done a few very sophisticated PID algorithms in software. You must remember to fix the reset windup problem in software. I did simultaneous voltage, current, power, and temperature control with output step maximums back in the mid 80's.
I also used the proportional band of one controller to make the setpoint of another RTD agree. Oven temp was +- 10 degrees of desired temp/.

See: https://en.wikipedia.org/wiki/PID_control


I guess you never read the warning inside the engine compartment that the "fan can start at any time". As part of the system you do still need a mechanical thermostat. The engine temp will increase when turned off. Pressure decreases causing the water to boil. Latent heat adds heat. No circulation (ersatz convection) is reduced because of lack of flow.

 

[tyrebyter]

3v0, Maintaining a constant (or nearly so) operating temperature is imperative in pretty much every car engine there is. It has to do with fuel mixture/combustion, oiling clearances and more.

MrAl, I like the idea of using both windings all the time utilizing PWM. It seems like it would be the an easier solution than having two control methods. I also like your idea to try a simplest method first, to observe what might need changing - (see my previous post).

AlecT, The coolant temperature needs to remain constant regardless roadspeed, RPM or throttle opening. Not a problem as long as the car is moving- plenty of airflow through the radiator and the thermostat does a good job. When the car is at a standstill, is when the electric fan will come into play.

KISS, For various reasons it's best that the fan motor is unpowered, never to turn on, when the engine is not running. Yes, the engine goes through what is called a heat-soak cycle and underhood temperatures climb for several minutes after switching off the ignition.

I'm beginning to think some photographs would help everyone see more clearly, what I want. In particular, a shot of the engine/thermostat housing/sender and the fan motor with three wires. Maybe a short video of the motor in operation. I'm also wondering if anyone ever starts rapping a gavel when creative minds start to deviate on here ... ;>)

 

[tyrebyter]

KISS, I'm not sure why one set point will not work. Meaning turn on at a temperature and then turn off at the same temperature.

The radiator fan will not reduce the coolant temperature until after the hot coolant has left the engine and entered the radiator. The coolant flowing past the sending unit will continue to get hotter until a period of time after the fan turns on. Enough time for the coolant to go through the radiator, get cooled, and then back up through the block and head to the sender again.

 

[KeepItSimpleStupid]

KISS, I'm not sure why one set point will not work. Meaning turn on at a temperature and then turn off at the same temperature.

If this were a relay turning on and off rapidly, we would call it chatter and it would result in unreliability.

A single setpoint to maintain "exactly" is really tough to do. Think about the following. The car may run from 0 to 60 miles per hour. the ambient temp is different, the wind is different. So, the fan speed has to be different under all of these conditions. Some form of PID can do this.

Take an simple toaster oven bi-metalic thermostat. It has built in hysteresis. On at some temp and off at another. When it turns off, the temperature is still climbing.

Look here: https://www.watlow.com/products/controllers/series-cv-temperature-controller.cfm and especially at the PID practical guide. Temperature controllers are expensive. You might even want to go that route for tight control.

They have all sorts of options for I/O. So, you might have to add some glue and make it operate on 12 V or do PID in a microcontroller.

 

[KeepItSimpleStupid]

Hypothetially:

Suppose you found a temperature controller with the following characteristics:
DC powered
0-5 in
0-20 mA out (usually isolated), 4-20 is also usually available
Reverse Acting
High limit alarm
Low limit alarm

Those options would not be unreasonable for an AC powered controller.

You linearize the radiator sensor with electronics to a 0-5 V signal and set the low and high values in the controller.
You then take the 0-20 mA and put it across a resistor to get the 0-5 V signal you need.
You set the setpoint.
Go through the tuning procedure.

Pretty much done.

For added giggles, you use the low temp alarm to force the fan off and the high temp alarm to force the fan on, but you should not have to do that. It would just take some time for the fan to get there. Even so, you should not need this.

Other temperature ranges could select the windings, although the difference between setpoint and ambient might be better or you could just leave it well enough alone.

Just a thought.