Automotive radiator cooling fan contoller

[tyrebyter]

Note also the flow rate in an engine is probably much greater that the example. Perhaps you can find this information out too and that would help us to more accurately model the system.

I can see coolant flowing past the radiator cap opening but at a very lazy pace when the engine is idling at operating temeperature - thermoststat is fully open. We are discussing a 750 RPM idle speed and about 3.5 gallons of coolant. With the fan running, I can tell when the thermostat closes because the coolant becomes stationary. This is about all the emperical evidence I have, regarding coolant flow rate in this car.

Coolant flow rate is only part of the system, though. The fan only affects the coolant in the radiator. The engine block and cylinder head will re-heat the coolant, before it reaches the sending unit. I believe this will add to the mechanical delay in turning off the fan.

If I find the duty cycle of the fan is based on too short a time i.e. the fan turns on-off-on in much less than a minute with no electronic delay/hysterisis, then I'll have to admit you are right.

 

[alec_t]

if we allow a small temperature deviation between turn on and turn off (build in some electrical hysteresis) then we see a much longer turn on, turn off period. How much longer exactly would be the subject of an experiment

If it's any help, here's a simulation. You can play with the top and bottom temperature values to see how they affect the on/off interval. Thermal inertia is represented by R1/R2/C1 time constants.

 

[KeepItSimpleStupid]

One can always add time to the hysteresis thing. A pressure switch that I have offers 2.5, 5, 100 and 500 mS. However, in this case, I doubt that's needed.

 

[tyrebyter]

Not sure what "brake power is": https://www.physicsforums.com/showthread.php?t=566125

Coolant Flow rate/Brake power graphs.

Brake Horsepower or bhp is a way of describing engine's power output. Using a "rolling road" or "chassis" dynamometer, the drive wheels cause a large drum to rotate. The large drum has an equally large brake attached to it. The amount of braking force required to keep the drum RPM from exceeding a set point, fighting against the engine, is brake horsepower.

Not to be confused with "Break Mean effective Pressure" or bmep as it relates to what happens in the combustion chamber. A direct drive or engine "dyno" can accurately measure bmep which a "chassis dyno" cannot do since power has to be transmitted throughout the drivetrain before the drive wheels rotate.

Come on guys, am I really getting automotive cooling system coaching, from an electronics forum ? If that is the case, then maybe one of you would be so kind to tell me "why did all cars have fans that ran continuously" up until the mid '60's ? ;>)

 

[tyrebyter]

I'll need some time to study the *.asc file you posted, alec_t.

I think I get it now It's ok to keep an Op-Amp or comparator full on or full off. How about ... is the converse true ? Is it kosher to keep an OpAmp in the linear range at extremely low frequency - less than 0.1 hz ? Basically steady-state with slowly varying dc voltages at the input and output.

Reason I'm asking is that I want the output of X4 (V-lo speed-out) in the following schematic, to provide a variable control to drive a PWM circuit :

 

[KeepItSimpleStupid]

"why did all cars have fans that ran continuously" up until the mid '60's ? ;>)

Nobody cared about pollution and fuel efficiency and at that point, who would want vacuum tubes controlling their car. Heck, you had a mechanical oscillator at 200 Hz, called the vibrator in tube radios. If you want a faster warm-up turn off the fan. My 82 whatever had a mechanical temperature sensitive clutch. We might be carrying around an 80 lb computer like used in the Saturn IV. https://en.wikipedia.org/wiki/Saturn_Launch_Vehicle_Digital_Computer

Side discussions are OK. We like to learn too. So, we ask questions. Why do you need this is a good question". If you can model the system, or anticipate issues, the design goes better. Sometimes a second or third pair of eyes help.

 

[alec_t]

Is it kosher to keep an OpAmp in the linear range at extremely low frequency

Yup, no problem when in that circuit.

 

[tyrebyter]

KISS, it was a trick question.

Side discussion:

Unless I'm ignorant of a car that had otherwise, ALL cars had mechanical fans up until the mid '60's. Therefore they never shut off or in other words ran all the time !

The late '60's muscle-car era ushered in flex-fans and then fan clutches to remove parasitic drag at high engine RPMs but htese did not use electric motors to drive the fan. They certainly were not considering fuel economy and exhaust emissions when building beyond larger than 400 cubic inch engines with 15:1 compression ratio and twin- four barrel carb setups offered on "stock" cars for sale to the general public.

I believe it was the Little European front-wheel drive cars that started the electric fan boom. Most notably it was the front wheel drive Fiats and Volkswagens. It was difficult to get enough air-flow through the radiator with a belt driven fan, if the engine was tranversely mounted. The radiator had to be beside the engine, next to one of the front wheels as in the first of the front wheel drive cars that were water-cooled. Most notably the British MG-1100s, Austin 1800s, and classic Mini-Coopers. Hidden from the airflow that comes in through the front grill opening, the radiator did not do a great job of keeping the engine cool unless the car also had an extremly large capacity fan, with up to 15 blades in some examples. Also elaborate fan shrouds were commonplace to make sure all airflow from the belt driven fan, was directed through the radiator and exhausted out through the front wheel, inner fender.

Citroen had an air-cooled front-wheel-drive car, the 2CV but it had no radiator just like some of the very earliest Honda automobiles that were front-wheel-drive.

Putting an electric fan onto a front-mounted radiator cured a lot of problems with cooling systems in these little transverse mounted engine, front wheel drive, econo-boxes. This happened along with the advent of the USA leading the way to controlling exhaust emissions. In europe, the cars did not have to burn as cleanly and for years they got much faster versions of the same cars that sold here, but still used electric fans. Without the constraint of exhaust emissions, European car manufacturers were free to develop much more power with basically nothing more than tuning changes. What we got here at home, were watered down, lower horsepower versions of some fairly quick little cars.

 

[tyrebyter]

Alec_t, I cannot get the ltspice software to work with your latest schematic. To be truthful, I have not been able to get any circuits to work with that software. My downfall is apparently that I don't know enough about spice modeling to make components from text files or parameters. Also, I don't fully understand the voltage sources well enough to to select something compatible in my "other" simulation software.

I have not been able to find a model for the CD4013B ic. Other flip/flops I've tried do not work either in the software I have been using to make the schematics I've been posting.

At this point, I'm working to get the rest of the pieces in place, namely the PWM control section for lo-speed operation with soft start.

 

[alec_t]

Alec_t, I cannot get the ltspice software to work with your latest schematic

Ah, if you're missing the CD4013 model you could join the Yahoo LTspice User Group and download free models of umpteen things, including the CD4013 and many ICs in the CD4xxx series.
Visit
LTspiceFiles - Yahoo! Groups

BTW, any model file you add to LTS should go in the ...../lib/sub folder, and the related symbol file should go in the ...../lib/sym folder (or an appropriate child of that folder).

 

[tyrebyter]

alec_t, Somehow it came to me to try the LM324NS opamp and it is in my model library. Using my simulation software, I swapped it into the TempSwitch schematic you first posted and it works ! I get a VERY steep curve through the 100 - 150 ohm region which is exactly what I think I need for good control over the turn on point. It seems that up to certain limits VR1 sets where the opamp starts shutting off the output and VR2 controls the steepness of the curve. It seems VR1 is somewhat touchy and makes major changes with little chnge of it's wiper position. It will take me some more time studying/playing this circuit to understand what the 4.7V Zener does but I feel it is probably a big part of this little picture.

I joined the Yahoo LTspice users group several days ago and yesterday, found the CD4000 library and found the CD4013 model but it was not what I was expecting. I guess I was looking for a neat little package or symbol/graphic depicting something that looks like the symbol in your latest schematic. Instead what I saw was text. Parameters, I think. I don't know what to do with that.

The simulator software I'm using is "SIMetrix/SIMPLIS Intro" from Simetrix Technologies. It has a tutorial that walked me through the basic steps to making schematics with it.

I saw some tutorials listed using LTspice but when I try to find models on Yahoo, I keep running into not knowing what to do with what I have found. I'm not sure I'd be able to make sense out of LTspice but I guess I better try haven't I ? Before everyone here runs out of patience with me and before my tial period ends with SIMetrix.

 

[MrAl]

Hi,

I think brake power is the power of the engine itself as measured with the right equipment that subtracts other secondary losses. So that would be the raw engine horsepower i guess.

So it seems that maybe 1 liter per second would be a starting point. But because of the variables i am starting to think that we just have to rely on experimental data on this one. That is, build a base system just to use to make measurements (or keep) and take some readings. The two set point system seems the simplest to start with.

tyrebyter:
I think there should be models in LT Spice that you can use. You dont have to have the exact parts to do a few runs to see how this works. They should have some comparators and op amps in there.
I'll try to throw something together myself.

 

[alec_t]

It seems VR1 is somewhat touchy and makes major changes with little chnge of it's wiper position.

Yes. The schematic is just a proof-of-concept one. Once you've decided on the resistance range you actually need you can use a smaller value of VR1 but add a series resistor either side of it. That will make for finer adjustment.

 

[tyrebyter]

alec_t, According to one source, the resistance range of the temperature sending unti, for normal operating temperature, gauge reading at "N", in the centerpoint of the needle's sweep, is around 100 - 150 ohms. I'll know better once I take the sending unit out and take some warmup/cooldown measurements in water, on a stovetop. That will happen before I begin accumulating electronic components for this curcuit.

For now, I'm still trying to proof the circuit which hopefully will run a two winding motor with two seperate control schemes.

 

[alec_t]

I wasn't referring to the sender resistance; rather the useful resistance range of VR1 as the pot is adjusted. In my simulation I represented the sender as simply a voltage source varying from 8V to 5V (using your figures).
100-150 Ohms seems quite a wide tolerance for what should be interchangeable/replaceable senders. It means that your temp-setting pot will need enough range to allow for the temp range you're interested in plus a ~50% margin for sender tolerance. That 50% tolerance also means that your temp gauge reading could be way off target (unless there is some adjuster mechanism?). I guess most people only want a rough indication of temp anyway, the important thing being if the reading suddenly rises or falls.

 

[tyrebyter]

Alec_t, I think I may have said it wrongly. Indeed there is quite a lot of discrepancy from one sender to the next. The quality of the parts that suppliers sell for these cars has always been a problem. I've learned to cope.

I've also plugged your circuit into the one I'm s-l-o-w-l-y getting a handle on. I'm planning next to put together (on my computer monitor at least) a PWM circuit comprised of the other two quarters of the LM324 ic. I'll aim for the 40khz that KISS came up with as an inaudible frequency for the motor.

I have collected some examples of PWM ciruits that use two diodes and a resistor/capacitor circuit to drive two opamps running as comparators.

I now have control over what temperature the lo-speed will function and what temperature the hi-speed will switch on the motor at 100% RPM. Between the two, I believe there will be a pseudo/partial electronic hysterisis possible (just in case you guys are more correct than I am in assuming it's necessity) with adjustment capability.

X1 will be replaced with a fixed 10 Volt regulator. R2 is the temperature sending unit. VR3 controls lo-speed operation. VR2 controls hi-speed turn-on temperature.

Here is where I'm at now, before adding the PWM circuit :

 

[alec_t]

Neither opamp in your circuit has any hysteresis. I think that could result in instability/oscillation in practice.
The gain of X3 is very low, by virtue of the 500k feedback. Is that intentional?
What are R3/R4/R9 for?

 

[MrAl]

Hi,

Why is R3, R4, R8, and R9 in the circuit? R3 and R4 dont do anything for example.

Why is the 100k pot connected differently than the 500k pot?

Why use a 500k pot? Resistors in this application should really be kept at 100k or below.

Trying to get an idea what you are thinking here.

 

[tyrebyter]

alec_t,

I will be putting in what I have learned are called "bypass capacitors". One for each opamp. If memory serves me correctly (my head is swimming in lots of new tidbits of information), they will be on the inputs to ground and each about 0.01mfd. I'm still not sure how/what they do exactly but I understand it is specifically to prevent oscillation. For now, I'm attempting to keep this circuit as uncomplicated as I can until I have everything working in simulation. Apparently the software does not have a problem with oscillation at this point.

I made R8 500k deliberately to result in low gain, trying to use this portion of the 324 as a voltage invertor, not as an opamp comparator circuit.

MrAl,

Trying to answer both yours and Alec_t's questions in one reply. Per your suggestion (voice of experience, I assume) I changed the 500k pot to 100k and the simulation graph looks the same. I connected the pot to X3 differently because X3 is being used as a voltage inverter instead of a comparator. It will be delivering a varying voltage (I think/hope) to the PWM driver. On lo-speed I want the fan to vary it's speed according to temperature, albeit within a very narrow temperature range. I made the pot 500k in keeping with my goal of reducing current throughout the circuit to a minimum. I have yet to do any math to see what the actual power dissipation will be through each component. I've been just merrily swapping in and out components and adjusting values to get the plotted curves I'm looking for. I hope this sheds more light on how little I know about what I am doing with electronic circuit design. Kinda groping along by feel.

As far as the resistors R3, R4 and R9 are concerned ... 9 was in the opamp comparator circuit I had. 3 and 4 were both in the transistor switch curcuit I had. When I plugged both curcuits together and into the rest of it, these resistors remains as they were. I guees I didn't see the forest for the trees, until you both pointed it out. I can easily see now that I can combine a couple of the resistors, or more but i think they need to be there in some form. I'll spend some more time doing that later. For now, I tried removing R8 just this minute and it reults in X3 putting out a flat voltage just under 4.5V. Apparently it has to be there.

It might not be apparent but I'm totally in a fog regarding current flow. Yes, I understand ohms law so I can calculate the currents but I am attempting (maybe unknowingly wrong to do so) to control voltages here. Using voltage dividers seems to be my only solution in places. Witness the R3/R4 combination. Am I aiming at the wrong target ? Should I be more concerned with how much current each component is delivering to the next ?

 

[MrAl]

Hello again,

R8 is not only not needed it could make the LM324 act as a linear device, which you dont want here. Therefore, you need to eliminate R8 and get that part of the circuit working without that resistor.
The LM324 should be perfectly capable of putting out the required voltage without that resistor. If it doesnt work without that resistor, then something else is wrong.

Same goes for R3 and R4. Remove them and get the circuit working without them. If you cant get it working without them, then something else more serious is wrong.

Im not sure how these three resistors got into the circuit either, what made you put them there, but they should be removed.

It was also pointed out that the two comparators probably need some hysteresis. This comes from putting one resistor in series with the non inverting terminal of the op amp, then another much larger resistor from the output to the non inverting terminal. When the output goes high for example it is because the non inverting terminal went higher than the inverting terminal, so the small feedback from the output back to the non inverting terminal raises the non inverting terminal voltage a little bit more so that the comparator can not immediately turn back off as the voltage dwells near the crossover point. The positive feedback eliminates the dwell and forces it to stay in one state rather than flip back and forth.
The values would be rather high for the feedback resistor and rather low for the series resistor. For example, 1k for the series resistor and 100k for the feedback. If that turn out to not be enough, then 2k for the series resistor. If that's still not enough, then 5k, then 10k, etc.
The percent hysteresis is roughly calculated from:
Rseries/Rfeedback
So with a 1k series and 100k feedback the hysteresis is roughly 1000/100000=1/100
which is 1 percent. With 10k and 100k, it is 10000/100000=10/100
which is 10 percent, however 10 percent means that the trip point low is 10 percent lower than the trip point high so that is probably too much for this application. It really depends on the output of the sensor though so a good place to start is at 1 percent.
This is one of the more foggy areas of the design and may require some experimentation.
Another point is to put a capacitor across the sensor, at the input to the PC board for the controller.

One of the problematic things that comes up in automobile circuits is trying to eliminate noise that comes from various areas. One problem area is the ignition. The coil, drive circuit, and ignition wires cause lots of radiated noise. Dealing with that is not always easy. Troubleshooting may require a scope.

It dawned on me that you might be trying to learn electronics along with building this project. I think that is very good. I am giving you this advice with the hope that it helps get the project going as fast as possible, but you might not be in that much of a hurry if you are also trying to learn more about electronic circuits. So take your time if you wish, and i hope you can get this going ok without too much trouble. Keep in mind however that this is not the best project for a first project if that's what it is for you. Part of the complexity comes just from the fact that it is for a car, to be operated in the vehicle itself. So another idea would be to test it in the home first under more controlled conditions to make sure it works as designed, then transfer it to the car and see what happens. All that is required is a power supply and maybe a pot to simulate the sensor, and of course a load like the motor if not the motor itself.