Automotive radiator cooling fan contoller

[tyrebyter]

I want to be able to pick the turn-on temperature and turn-off temperature of an electric radiator cooling fan motor. I also want to use the stock temperature gauge sending unit while rataining normal functioning of the gauge. A bonus would be soft start for the motor and variable speed according to temperature - up to and including full speed as if switched on by a relay. I have a two speed fan motor with two windings that share a common ground. I have a regulated 10 volt supply to the (current controlled bi-metal heated) gauge and have determined gauge resistance and sending-unit-per-temperature resistances.

My head is swimming in PWM circuits, discreet verses, microcontroller verses, 555 timer ic and more. I'd rather be able to fix this if needed so prefer discreet components, tranistors, digital flip-flop gates etc. I would rather not use a microcontroller and software based parameter control functions - if that is the correct nomencalture.

The gauge and sending unit circuit is in effect, a voltage divider. I can figure out how to read the voltage between the gauge and the sending unit using a buffer Mosfet or op-amp. I can figure out how to determine the turn-on and turn-off temperatures/voltages. I am stuck trying to gain enough knowledge to put together the rest of the control functions, i.e. Soft-start, variable speed and in effect a 100% duty cycle if I choose to go the PWM route. If I use a PWM-to-linear voltage controlled function, I am stuck trying to figure out pretty much the same stuff.

Any 1st hand experience/help or links to circuits that could be a part of this controller design would be greatly appreciated.

 

[alec_t]

Don't know which country you are in, but here in the UK almost any mod to a vehicle would have to be approved by your insurer; otherwise the insurance would be void.

 

[KeepItSimpleStupid]

I've used this https://www.jameco.com/webapp/wcs/stores/servlet/Product_10001_10001_120539_-1 as a starting point for a few projects. Take a look at the manual. It already has soft-start built in to it.

In the US, BTW, they don't care much except probably in California (CA). In my state all we have to do is pass an emissions inspection using the OBD port. We go through an inspection lane and "hit the brakes" as the last test where stopping is measured on all 4 wheels.

Initially,, you have to have "proof of insurance"; your lights have to work; turn signals have to work; Tires are "looked at"; the driver's side window has to work; wipers have to work; wiper blades are inspected; and other cosmetic stuff.

Illegal stuff includes a muffler cut out and front window tinting.

As long as it passes the emission test for the particular vehicle, your good to go.

Other states have a physical inspection that occurs at an approved shop. They will take off the wheels and examine the brakes, for instance. Emissions are much stricter in CA and the cars have to have stricter emissions from the automobile factory.

Car mileage at inspection is always noted. The actual inspection takes about 15 minutes. You pay for it at another area while still sitting in your car.

To make life easier, they have a re-check line for all problems encountered except brakes. I think, it's an initial 5 year interval (car new) and then every two.

 

[tyrebyter]

Hey, thank you very much "KeepIt" ! This might be just what I need. I downloaded the manual and data sheet for the heart of this kit. I will see if I can decipher the schematic and internal workings of the SG3525 ic enough to grasp how and what it does. Maybe I'll be able to duplicate some of it's functions as needed.

Alec, not to worry. I live in the US and this car is too old/is not tested for emissions in my home state. Modifications are allowed with only a few restrictions, not including the radiator and cooling fan.

 

[3v0]

Here the lady who sells you the plates goes out to see if you really have a car. I think she looks at the vin number. So it varies.

Whatever solution you come up with make sure it can handle enough amps. I would put a header on the board so you can move a jumper to power the fan if it dies.

 

[tyrebyter]

OK, I have run software based simulations on a conglomerate of components and wires. The software is a freebie spice type simulator and I had to learn a fair amount about how to use it, just to get to where I am now.

The circuit I have so far, taps into the wire that connects to the temperature sender. The car, a 1976 Triumph TR-6 uses a heated bimetal piece to move the temperature gauge needle. The heating element is in series and presents around 60 ohms to the circuit. There is a mechanical voltage stabilizer which will be replaced with a standard ic voltage regulator rated for 10 Volts. These "modern" solid state 10 Volt regulators are available specifically for these cars. The temperature gauge is then connected to the temperature sending unit which is a thermister that goes from around 1Kohms cold (below 70 F) to about 53 ohms when the engine is overheating. The thermistor is grounded via the cylinder head that it is screwed into. The gauge and the sending unit make up a voltage divider.

I have determined that the voltage range which I am interested in from this voltage divider, is between 6.2 and 7.2 Volts. I do not want the temperature gauge circuit to be affected by the radiator fan control circuit if at all possible so I selected a MOSFET (2N7000/PS ) from what was in the model library of the freebie simulation software. The extremely high impedance of the 2N7000/PS gate is hopefully what I am looking for in order to buffer the gauge circuit's voltage divider from the rest of the fan controller circuit. The output (source) of the MOSFET runs from about 5 volts when the thermister resistance is at it's highest (1K ohm, engine stone cold) to about the full 10 Volts when the thermister resistance drops to zero ohms (engine super heated to the point of vaporizing).

The MOSFET is powered via another voltage regulator with 10 Volts connected to the 1N7000/PS Drain. Thus the 10 Volt maximum output at the 2N7000/PS source. Also powered by this 10 Volt regulator are a pair of OpAmps (LM741/NS) again from the simulator library. The two 741/NS OpAmps are setup as single supply on-off switches. They are both connected in parellel to the MOSFET output (source). The MOSFET drain is connected to each OpAmp's negative input. Each 741 has it's positive input connected to it's own voltage divider.

What I have now is OpAmp #1 turns off when the Mosfet output(drain) voltage reflects when the thermister reaches a preselected temperature. OpAmp #2 has a slightly higher preselected temperature when it turns off. Next I need to put these two OpAmps outputs to use to drive the fan power curcuits. If you are following this thread closely yo uwil lremmeber that I have a two speed (two windings) fan motor. I want the lo-speed to kick in at one preselected temperature and the hi-speed to kick in only when needed, but at a temperature that is several degrees higher than when the lo-speed turns on.

If I could figure out how to post the schematic, I would. Any suggestions how ?

 

[tyrebyter]

OK, I just surprised myself. Here is the schematic. I need help obviously. Make any suggestions you want.

 

[tyrebyter]

OpAmp #1 output is green, #2 is red verses Thermister R2 resistance in the schematic.

 

[tyrebyter]

Here the lady who sells you the plates goes out to see if you really have a car. I think she looks at the vin number. So it varies.

Whatever solution you come up with make sure it can handle enough amps. I would put a header on the board so you can move a jumper to power the fan if it dies.

Not a bad idea 3PO. I could leave the mechanical temperature switch in place but set at a higher temperature, just in case.

 

[KeepItSimpleStupid]

Not bad, just hope you don't use a 741.

I need help obviously

. Either that or go back to biting tires.

 

[tyrebyter]

OK KISS, I'll bite . BuwaHaHa haaaaa

Why not use a 741 ? Maybe you don't realize how little I know about this stuff. I know a LOT about cars, thus the "tyrebyter" handle. Electronic components and circuit design are not something I am very fluent with. Is there a good substitute for the 741 ? Is there a better choice for the MOSFET ?

Ultimately I'm hoping to reconfigure OpAmp #1 to provide a variable voltage which will control a PWM for variable speed control and a soft start for the lo-speed winding in the fan motor. I figure I better get the on-off stuff to happen at the correct temperature first.

Oh yes ... about that LM317 voltage regulator. I'm sure I can find a 10 volt regulator to replace the adjustable one, once I get to the point that I am soldering these pieces together.

 

[KeepItSimpleStupid]

The 741 may not even operate on a single supply. It's like one of the first or the first op amp commercially produced. You should use a comparitor rather than an op amp anyway.

Look at the LM339 http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0CC4QFjAA&url=http://www.onsemi.com/pub/Collateral/LM339-D.PDF&ei=oG54UpqBEoTgsASB6oCIBg&usg=AFQjCNH1YWwCBZAaisOO27k9Nr28KcoMtQ&bvm=bv.55819444,d.cWc&cad=rja It's a dual comparitor that could be purchased as an automotive rated part.

The outputs are open collector.

Here is another possible chip: http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&ved=0CC4QFjAA&url=http://www.ti.com/lit/ds/symlink/lm613.pdf&ei=vHB4UvWdOJDKsQS9wYLACA&usg=AFQjCNHJD4zfg4b8V6kjDyWBpd3MYRgqLQ&bvm=bv.55819444,d.cWc

It's hard to work with, so I'm not going to suggest using it. Generally, you can probably get the chip soldered to a DIP adapter if required from www.protoadvantage.com

So, you get a real reference and you could use both op amps to create two variable references and then the two comparators to create a window comparator.

FWIW: There are lots of missing parts such as bypass capacitors. The circuit will need some extra components to make it automotive environment friendly. See: http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0CDsQFjAA&url=http://www.littelfuse.com/~/media/Electronics_Technical/Application_Notes/Varistors/Littelfuse_Suppression_of_Transients_in_an_Automotive_Environment_Application_Note.pdf&ei=hXJ4UtfbEOvhsAThrYHYBg&usg=AFQjCNFtLnWTgd2nP7kFajjBuwe-dCcHBQ&bvm=bv.55980276,d.cWc&cad=rja

Your doing OK. Your circuit may work on paper and it would probably fail in a practical sense. Don't give up.

You may want to consider the circuit behavior when starting the car, so the fans won't come on. It might not come on anyway because some circuits in the car are disconnected when the car is started.

PS: I thought Tyrebyter might have meant tired of programming. But tyre is Brittish spelling for tire, I think.

 

[alec_t]

IMO you don't really need the FET to buffer the temp sensor voltage. Providing the circuit input impedance is > ~50k the temp gauge reading will be affected < ~2%.
Here's how you could use a single opamp with hysteresis to determine whether the voltage is in the 6.2-7.2 range.
Trim 1 sets the lower voltage trip point and Trim 2 sets the hysteresis amount, hence indirectly the upper voltage trip point. There is some interaction between the two trimmer settings.

 

[tyrebyter]

Thanks again Kiss, I read the articles regarding "transient protection" from Littlefuse and I have to agree. Something needs to be in the circuit to do that job. It will probably be added towards the end of the design and testing phase. I'm trying to do this one little step at a time and it's still mind boggling to me. I also understand that it will probably take numerous burnt components before I get this thing even close to operating properly.

I also checked and there is no LM339 in my simulator software model library. BUMMER. I am thinking it might be better (for me anyway) to use bipolar transistors instead of the two OpAmps since I will be hand soldering this circuit together and to avoid any possibility of chip "creep" down the road. I believe the pair of transistors could then be used to drive the turn-on relays directly, without going to a third stage or 2nd buffer.

One more of many things I do not understand is the need to use capacitors. This circuit is reading a DC voltage that s-l-o-w-l-y rises over a period of 10 or 15 minutes as the engine coolant warms up. Except for extreme voltage transients, as mentioned in the Littlefuse articles, I don't see where ripple or "noise" would be produced by the circuit.

Your suggestion to use a window comparator has me puzzled. I am trying to turn on a fan when the coolant temperature is getting too high which occurs at a preset voltage from the (temperature gauge/sending unit) voltage divider. The fan will stay on until the coolant temperature is reduced sufficiently and then shut off as the temperature/voltage drops below the preset. Don't window comparators turn ON at a preset voltage and then OFF again as the voltage rises above another preset ? I do not think a control on hysterisis will be needed because the hot coolant that just turned on the fan will have to navigate through the radiator, getting cooled and then back up to the top of the engine getting heated, before reaching the sending unit/thermister again. In effect, hysterisis is built-in. The fan won't be switching off and back on again repeatedly, in rapid succesion, I don't think so anyway. I hope I'm using the nomenclature properly i.e. "hysterisis".

One of the overriding features I am looking for in designing thsi circuit is the ability to control the turn-on and turn-off temperatures, independantly. Have the fan turn off sooner, before the coolant gets too cold. The typical aftermarket adjustable fan switches have a 15 F degree range - on at 195 F, off at 180 F or on at 185, off at 160 and so on. The fan either turns on when the engine gets a little too hot for my liking (the carburetors do not tolerate temperaturefluctuations very well) or it runs much longer than needed. Sometimes it won't shut off until the car is moving again. Duty cycle is in the neighborhood of 75% and higher when it's hot out.

The entire electrical system of the TR6 is based on 4 fuses and no relays except for the optional overdrive wich operates via a switch/relay/solonoid controlled hydraulic valve. Battery power is fed directly - unfused - to the lighting and ignition switches. Woefully inadequate by today's standards but that's what it is. Oh yes, there is also an inline fuse for the radio. I can see where the wiper, washer and heater motors could steal current when they start up. The starter motor especially takes a lot of current when it operates. I guees the fan control circuit needs to be able to tolerate a worst case scenario and not malfunction when the supply voltage drops below 10 Volts. I'll double check that and see if I can find how it might be a problem with the circuit as it is now. Thanks for bringing that up.

Meanwhile I'm heading back to the simulator.

 

[tyrebyter]

IMO you don't really need the FET to buffer the temp sensor voltage. Providing the circuit input impedance is > ~50k the temp gauge reading will be affected < ~2%.
Here's how you could use a single opamp with hysteresis to determine whether the voltage is in the 6.2-7.2 range.
Trim 1 sets the lower voltage trip point and Trim 2 sets the hysteresis amount, hence indirectly the upper voltage trip point. There is some interaction between the two trimmer settings.
View attachment 81791

Hi Alec thanks for the help. I'm not sure 2% is an accurate measure for how much this circuit would affect the gauge reading. I think adding another resister/path to ground between the gauge and the thermister (trim 1 in your schematic) would throw the gauge off by the value of the resistor. Here is some more info that maybe I should have included previously ; at 176 F the thermister has 142 ohms resitance, at 194F the thermister has 106 ohms resitance. I think I did the math correctly with 60 ohms from the gauge (R1 in my schematic) to arrive at the 6.2 to 7.2 Volt range. Give or take a few degrees. The 820 ohm resistor (R2 in my schematic) is the thermistor at 77 degrees F. I've been using the sweep funtion of the simulator to run R2 from zero ohms to 1k ohms, in order to see what happens as the engine coolant/thermistor warms up.

 

[alec_t]

I'm not sure 2% is an accurate measure for how much this circuit would affect the gauge reading.

In giving the 2% figure I was assuming a worst-case scenario of 1k for the thermistor and a 50k load in parallel.
Trim 1 is 100k and the opamp has a much higher effective input resistance, so 100k in parallel with 1k and the opamp input gives a change of only ~1% in the effective value of the thermistor resistance. You wouldn't observe any change to the gauge reading unless you have Superman vision .

 

[tyrebyter]

Alec, I need to look at the math of what you are saying. I'll incorporate it into just the voltage divider R1, R2 (60 ohm gauge in series with the 0 - 1k ohm thermister), portion my schematic. If I add a 100k resistor in parallel with the sending unit/thermister, I believe the resitance will go from 820 ohms for instance (as shown for R2 at 77 degrees F) to a different resistance according to the formula for two parallel resistors. please pardon my lack of using k in the following paragraph.

R(total) = (820*100,ooo)/(820+100,000) = 82,000,000/100,820 = 813 ohms. 820-813 = 7 ohms. 820*.01 = 8.2 ohms. OK, I see it now. Very little difference. If on the other hand, I use for instance 106 ohms (R2 at 194 degrees F) then the math goes like this R(total_ = (106*100,000)/(106+100,000) = 10,600,000/100106 = 106 ohms, actually a squosh under. OK, I get it. Very little difference. Less than 1% in both cases, just like you said.

Resistances in parallel have a funny way of doing that I guess. What can I say ? DUH ! During my mental contortions putting together this circuit, I did not realize that well enough apparently, when I decided a MOSFET would be a good choice for the buffer. Presumably the trimpot will always show as 100K being in parallel with my R, regardless where I set the turn-on temperature. Am I correct in making that assumption ? If so, then I agree. The discrepancy is so slight that it will hardly be noticeable if at all ... even if I put on my 3.0 "Geezer Goggles" to look at it.

I'm still not convinced that using an OpAmp is the best solution however. If I had never had to push memory chips back into their sockets inside my 'ol 386 PC, then I wouldn't be concerned. But I know chip creep is a problem which I'd like to avoid. Also, I'm going to have to study your schematic more thoroughly to see how/if it can be used to turn on the fan and then turn it off again when the temperature falls.

I also have not looked at how to get the initial buffer stage would operate two different switches/devices at two different presets. The more I find out, the more I find out there is to find out. Sounds pretty typical to you guys, I bet. At least now I know I can stop worrying so much about effecting the gauge reading. That alone is worth a lot more than the price of admission.

Now to divulge even more of my ignorance regarding electronic circuits, I'm still floundering around trying to figure out why a window comparator would be helpful. Please bear with me. I realize you and KISS know what you are doing but I need to know WHY these suggestions work better as you saw when I did the math for parallel resistance.

 

[KeepItSimpleStupid]

Window is somewhat subjective here. I tried to show an alternative to getting two adjustable references to compare to.
"Window" could refer to OFF, variable speed range, FULL ON, although it looks like you don;t want to do the variable speed range. The "window" here is "variable speed".

I gather you may want to do something different. Maybe, the fill the container problem. Pump starts at 1/3, and stops at full, but does not start again until it reaches 1/3.

I need to re-read my post, but with comparators with open collector or open drain outputs, it's easy to "wire OR" them together.

Your learning. You don't know about the gotcha's yet.

 

[MrAl]

Hi,

A few notes about fan controllers...

There are at least two types which you seem to be interested in. One is the limit type and the other is the variable type. The limit type only requires one comparator that sets the upper and lower limit, and two set points and the motor only turns on or off. The variable type requires PWM but only one set point.

The two comparators wired as mentioned before in this thread is usually referred to as a Window Comparator. The window is set by the two set points, one lower than the other. The fan turns on when the upper limit is exceeded, the fan turns off when the lower limit is reached. Pretty simple circuitry, but with two pots for adjustment.
With one comparator instead of two, we simply build in a lot of hysteresis. This is a little harder to adjust though than the two comparator design.

The variable version also requires a triangle generator and comparator to generate the PWM. There is only one set point with this version however, so you only have to set one temperature. The fan speed varies in a way that keeps the device being cooled at a constant temperature. So if you set it to 200 degrees F, the fan will keep it at 200 degrees, period. It will turn off if the device normally runs lower than that such as during the first few minutes of engine operation. Once the temperature gets near 200 degrees, the fan starts to run slow, then faster if needed. If the fan works right the temperature differential will be quite small unless the fan is not powerful enough to cool the device during certain run times, in which case it will run at full speed.

The 741 should be abandoned for this project. It is really only needed when the input offset has to be trimmed. Also, it's input spec's are too limited and so is the output. A good substitution is the LM358 which is much better for these kinds of applications, but for the comparators the LM339 would work very well.

If you are worried about the impedance of the sensor then buffer it with the LM358. Build the window comparator with two sections of the LM339 or get the two section version. Use a nice heavy duty driver transistor like a high current MOSFET so it can handle the current and have room to spare, and probably a heat sink for it.

The wiring of the two comparators:
The comparators are wired with their outputs setting or resetting a flip flop. Without using a flip flop the wiring could be very tricky and not as simple as tying the two outputs together. Two of the inputs are tied together where one input is the inverting terminal and the other input is the non inverting terminal, one from each comparator. The other two terminals of the two comparators go to the pot arms, and the pots may include upper and lower set point limit resistors. The voltage for the two pots comes from a well regulated supply or better a reference diode based voltage reference.
Without using a flip flop, the wiring is a bit different as the two comparator inputs (one from each comparator) end up being the same (like the two inverting inputs tied together). The output is then a pseudo analog voltage that is sensed by a third comparator that does all the controlling.

The single comparator version simply has a relatively low value hysteresis feedback resistor. This causes the comparator to stay 'on' once the upper threshold is tripped, and it takes a much lower signal to turn it back off again. This means the fan stays on until the lower threshold is reached.

So you might want to decide which version you want to use. They are all pretty interesting.

Motors already have a built in natural slow start mechanism but yes the surge current may be higher during the initial startup, so some current limit might be a good idea. This would most likely require PWM for motors with normal run current around 10 amps.

 

[tyrebyter]

KISS,

Actually yes, I would like variable speed. Just need to fathom the basics "on/off" first. I'm sure it seems uneccesary but I'm a slow yet thorough learner, that way.

Ideally a vairable speed range from the lo-speed fan motor winding would be desireable. This fan motor has two basically identical windings each drawing 15 Amps at 12 Volts. Hi-speed is achieved by supplying power to both windings simultaneously. Splice the two positve leads (one from each winding) together and conect them to 12 Volts then ground the negative lead and you have hi-speed. Lo-speed is by connecting either (but only) one of the windings by itself, to 12 Volts.

MrAl,

All good points and well taken. Here is some more background that may give a better understanding of what I am looking for regarding an electric fan controller for this Triumph TR6 automobile.

I have done testing with a mechanical dial type "coolant" thermometer made specifically for this task. I removed the radiator cap and inserted the thermometer through the opening directly and down into the coolant that is in the header space above the tops of the tubes. It's a downflow type radiator. There is clearly a steady flow of coolant past the thermometer when the engine is fully warmed up and idling. The coolant flow turns into a rush when I rev the engine, as expected. The thermostat is rated at 180 degrees F and the thermometer verifies the operating temperature of the engine to be 180 degrees F regardless how much AC powered fan I put in front of the radiator. There is a slight fluctuation of only a degree or two +-.

The design of the relatively very large grill opening, duct work directing the airflow into the radiator and the heat rejection capability of the radiator has me convinced there is no need for the fan to turn on at any roadspeeds above approximately 5 - 10 MPH. The coolant gauge drops rapidly as soon as I get the car moving when it happens that the engine gets hotter than "Normal" while idling at a standstill. It drops somewhat if I increase the RPMs at a standstill but only for a while. Without getting into too much detail it has to do with the radiator capacity and volume of coolant in the engine. It also has to do in large part with the airflow through the radiator.

My target operating temperature for the fan controller would be adjustable (with a knob accesible from under the hood) from a low of 160 degrees F and a high of 200 degrees F. Reason being, there are times when it's beneficial to use a thermostat that is rated at a different temperature than 180F. There are various thermostats available that fall within this range. With the 180F thermostat I would like to achieve a target fan control temperature of 185-187 degrees F, if possible. I'd rather the fan stays off entirely until the coolant reaches at least several degrees above the 180F operating temperature.

I realize it might be impossible to achieve such a narrow band of control. Working for me in this regard is the diameter of the fan blade which covers nearly 85% of the radiator fins. It is a modern "reverse S blade" design which I believe should be more than adeqaute to deliver enough CFM to do the job. Also the motor is comparatively a MONSTER as far as electric radiator fan motars are concerned. This fan is the largest I could find that will physically fit in the avaiable space and has the most powerful motor avaiable.

Working against me is my weak knowledge of electronic circuit design. If I found a controller already built to do what I am looking for, I would have bought it several months ago. After extensive research, I do not belive such a controller exists. I would have had to buy one and some of them are hundreds of $$$ and then modify it or break into the probably glued together case to figure out how/if I could make it do what I want. The way I see it, if I have to design or modify a current design then, I may as well just start from scratch. The deeper I looked into how to design the circuit, the more I realized I was in very deep and obviously getting in over my head, which is what brought me to this forum.

I got started on the PWM route when reading into stock fan controllers that are already used on various, more modern cars. In particular I owned a Dodge Neon that used a PWN driven fan controller. I didn't realize it at the time but there is was, a rugged solid state relay that was PWM controlled via the ECM. I had a problem with the radiator fans not turning on and of course the engine would overheat. I found the problem was a broken wire in the fan control circuit, deep within the wiring harness. The journey I took to fixing this had me on a Dodge forum asking about where the fan relay was. Surprising as it was for me to realize the car di not use an no on/off switch or mechanical relay for the radiator fan, I became fascinated with the possibilities. When This TR6 electric fan project came along, I could not ignore the modern fan control circuits that are already in use, as a good starting point.

There are many aftermarket fan controllers available with all kinds of features. From the most basic non-adjustable mechnical "on at 195-off at 180" with a capillary tube connecting to a sensing probe you stick into the fins of the radiator. The next step which is basically the same but adjustable albeit keeping the 15 degree range. This is what has been in the TR6 for the past 10 years and it works but is barely adequate. The electric fan is noisy and runs almost all the time when the car is at a standstill. Mostly because of the low turn-off temperature. It's time for a better fan setup. They get better and better all the way up to some very sophisticated, programmable controllers. One thing every fan controller I've found lacks though, is the inclusion of of every design element listed here :

1) the ability to adjust the target temperature
2) the ability to have a tight range of control that has the fan turn OFF very quickly (or immediately) as soon as the coolant temperature drops back down to the target temperature
3) soft starting
4) variable speed
5) dual motor winding control
6) sensing the voltage by sharing the same temperature sending unit that the temperature gauge does, without effecting the gauge reading
5) remote mountable on or close to the fan motor

I'm going to be looking at window comparators and related circuits to see if I can get a better understanding why they are such a popular choice. In my mind I had decided that two controllers are needed.,One for each motor winding. Besides the small signal level buffer transistor at the temperature sending unit, it would be only two more transistors. One for each winding. The lo-speed controller would be PWM. Once the coolant gets hot enough that the PWM controlled transistor is at 100% duty cycle, then the other transistor would come into play. It would control a relay that is wired to both windings. Maybe it would not be too difficult to use PWM on both transistors and have variable speed all the way up to full 30 amps of draw, maximum motor speed. That's beyond me at present but I'm trying to learn.

I'm not sure why but I have the basic feeling that using transistors instead of ics, would end up being a more "tolerant" circuit as well as more rugged in the high vibration and generally abusive environment of an automobile engine compartment.

One overriding factor to keep in mind is that the alternator is capable of only a measly 36 Amps ! There will probably be times when battery reserve will be called on for brief moments especially if a relay is used. Before anyone wants to tell me to put in a higher capacity alternator, let me add that it's going to stay small/as-is, in order to NOT add parasitic losses to the power output of the drivetrain. This particular TR6 is highly modified already and the engine SCREAMS ! I defintley do not want to detract from that aspect of the car. Not even in any slight amount, if possible and still get good electric fan control.