Kenlowe Cooling Fan switching on and staying on all the time (Triumph Spitfire Mk III)

[Adrian Girling]

50 minutes ago, Rosbif said:

You're driving on the wrong roads in that case.

Just living in the wrong part of the country....

[PeteStupps]

5 hours ago, Adrian Girling said:

One thing I learned early on is not to trash other peoples ideas

The first thing I discovered in engineering was that your ideas are going to be challenged by every colleague and client so you've got to be ready to defend them!

But anyway, if the voltage from the stabiliser is too ripply (which it may or may not be) what about adding a simple low-pass filter on the input to your control circuit? Sounds like a diverting project, go for it.

Incidentally the stabiliser is on the back of your speedometer, it's a dinky little thing. The regulator is the big black box on the bulkhead.

[Adrian Girling]

13 hours ago, PeteStupps said:

The first thing I discovered in engineering was that your ideas are going to be challenged by every colleague and client so you've got to be ready to defend them!

Point taken.  In hindsight, I overreacted to a series of posts which were actually trying to be helpful. Apologies.  I've cooled down now - in spite of the heat!

 

13 hours ago, PeteStupps said:

Incidentally the stabiliser is on the back of your speedometer, it's a dinky little thing. The regulator is the big black box on the bulkhead.

I am so embarrased that I didn't realise this distinction.  I should have looked at the wiring diagram - which calls the regulator a "Control Box".  In my defence I'm literally working with a complete wreck - regulator along with a thousand other parts in four big plastic boxes - it looked so bad inside I decided to change the dynamo to an alternator.  Mea culpa

Found the little blighter on the back of the speedometer (in another box).  Doesn't look pretty, I wonder if it still works...

13 hours ago, PeteStupps said:

But anyway, if the voltage from the stabiliser is too ripply (which it may or may not be) what about adding a simple low-pass filter on the input to your control circuit?

That now makes sense to me.  My ignorance was a major contributor to the miscommunication, sorry again.  I wonder how this device actually worked in the 1960s?  A quick google suggests that modern replacements are probably made along the lines of the linear voltage regulator I described, with a pretty clean output but a low pass filter is good too.  I need to check the characteristics of the temperature gauge

 

13 hours ago, PeteStupps said:

Sounds like a diverting project, go for it.

Thank you for the encouragement.  As usual, the detail is always more complicated than the idea.  The temperature curve of the NTC sensor looks so flat where you need it most - it's a wonder that temperature gauges consistently show a useful reading.  A diode would be so much easier as a sensor for my solution.

[RobPearce]

41 minutes ago, Adrian Girling said:

The temperature curve of the NTC sensor looks so flat where you need it most - it's a wonder that temperature gauges consistently show a useful reading.

The characteristics of NTC thermistors are interesting to deal with, sometimes. As they're a resistance device, and the thing you're feeding tends (these days) to be voltage sensitive, the designer picks a load resistor to get the response they need around the temperature of interest. Of course, the gauge is more of a current sensor, but it still gets chosen to match. And if nothing much happens in the region of interest - at least the region you're interested in for fan control - then that's no problem at all for the gauge, as the driver actually doesn't care whether the engine is near the bottom of the normal operating range or the top. The driver only cares when it's outside that range. Which is why a lot of modern cars deliberately map the whole "normal" range to one central position on the gauge - the needle won't move from 60C to 95C.

[Adrian Girling]

33 minutes ago, RobPearce said:

The characteristics of NTC thermistors are interesting to deal with, sometimes. As they're a resistance device, and the thing you're feeding tends (these days) to be voltage sensitive, the designer picks a load resistor to get the response they need around the temperature of interest. Of course, the gauge is more of a current sensor, but it still gets chosen to match. And if nothing much happens in the region of interest - at least the region you're interested in for fan control - then that's no problem at all for the gauge, as the driver actually doesn't care whether the engine is near the bottom of the normal operating range or the top. The driver only cares when it's outside that range. Which is why a lot of modern cars deliberately map the whole "normal" range to one central position on the gauge - the needle won't move from 60C to 95C.

Nicely explained, thank you.  I'm about to check the characteristics of the temperature meter.  I've read somewhere that it's a moving iron meter, not a moving coil but that would probably need more current whereas a moving coil with a series resistor (and maybe a parallel one as well) could surely be made to work with less current so the current through the thermistor has negligable heating effect.  But I'm speculating - you sound like you know!

[Kevin R]

On a lighter ( or cooler note ) does an electric fan and associated housing stop the annoying accumulation of insect debris on the inside of a Spitfire bonnet ? 🤔

[glang]

Now that could be the clincher to end forever the elect/mech fan argument👍😁🙏

[RobPearce]

2 hours ago, Adrian Girling said:

I've read somewhere that it's a moving iron meter

I believe that was true of very early cars - the ones without the "voltage stabiliser" - but on most it's actually a bimetallic strip with a heating element. It's all very low power - just a few milliamps needed, which is much more than a moving coil, of course - but the current heats the wire which heats the strip which makes it bend which pushes the needle across. That's why they're so slow to react (most noticeable when you turn the ignition off and the gauge ambles slowly down to cold).

[Adrian Girling]

1 hour ago, RobPearce said:

it's actually a bimetallic strip with a heating element.

Spot on!  Took it out of the instrument assembly (Temp and Fuel gauge, beam and indicator light) and can see the bimetalic strip and return spring.  Measured a resistance of 60 ohms - same as the fuel gauge, presumably the same core instrument with a different scale.  Wacked 9 volts across it, watched it climb to the red line - the resistance went up to close to 65 ohms - i.e. not a lot.  An elegant way of building in latency for measurements that don't change rapidly - less prone to jump about than a moving coil and presumably less expensive.

With a sensor which may nominally have a low temperature resistance of 500 ohms and a high temperature resistance of 40 ohms I can see how this would work nicely giving a reliable indication over the range.

With 60 plus 40 ohms at high temperature the maximum current would be 100mA - easily achieveable with a 10V linear regulator (instead of my sad looking voltage stabiliser), probably without a heatsink.  Sniffing the voltage between the gauge and the thermistor through an op amp comparator should provide a signal to turn the fan on.

For now, I'm not even going to mention electronic hysteresis - may not be necessary as there will probably be enough lag in the system.  And I'm certainly not going to entertain a PWM drive for the fan - for now....

 

[Adrianb]

Thank you Gentlemen this has been quite an exciting episode in a Day of the life of a Triumph Enthusiast it beat’s Mills and Boon any day! Any chance of some photos to show us less technical minded where we can advance our cooling potential!

[Clive]

16 minutes ago, Adrianb said:

Thank you Gentlemen this has been quite an exciting episode in a Day of the life of a Triumph Enthusiast it beat’s Mills and Boon any day! Any chance of some photos to show us less technical minded where we can advance our cooling potential!

First things first, have you actually got a cooling problem? too many people panic about temperature, when the engines are happy to reach 100 degrees with no damage. 

And the first thing is to make sure the cooling system is good and clean, and the radiator core is clear (blocked cores usually won't flush clear as the water will bypass the blocked core). Only then start thinking about electric v mechanical fans. 

[glang]

Yes as Clive says the temperature gauge, unlike a modern car, will go up and down and thats correct. Then I suppose the best test you can do is wait until the hottest day of the year for a high speed run and after just leave the car idling for half an hour. If it keeps running smoothly and doesnt boil over your cooling system is perfectly adequate...

[Adrian Girling]

1 hour ago, Adrianb said:

an exciting episode in a Day of the life of a Triumph Enthusiast it beat’s Mills and Boon any day!

Adrian B, I wonder what the Venn diagram looks like - overlap of Triumph Enthusiasts and fans of Mills and Boon 😂

1 hour ago, Clive said:

First things first, have you actually got a cooling problem?

Clive, the answer is no. If I ever get this car on the road I'll find out!  Doing this for fun whilst I wait for the GRP bodyshell to come back from the genius who is copying it into aluminium.  I'm using the 13/60 to build a replica of a 1936 BMW 328 using a body made from a buck made from an original car.

21 minutes ago, glang said:

the temperature gauge, unlike a modern car, will go up and down and thats correct.

I'm sure that's true - I suppose I'm trying to reduce the variation but I'm also sure it's actually not necessary.  As Clive said, need to get the basics right.

[Clive]

My response was to the other Adrian.... there are few secrets about our cars, no matter what some will say. But essentially a well maintained triumph will be reliable (caveat is to avoid poor quality parts, and that is the hardest thing when so many old trusted names are now tat)

[Adrian Girling]

20 hours ago, Adrianb said:

Any chance of some photos to show us less technical minded where we can advance our cooling potential!

Thanks for asking Adrian.  See my back of an envelope (literally) sketch.  Need to think it through, and I'd welcome anyone who offers simplification (or sympathy if I've got it completely wrong) before lashing up a prototype.  I'd want to add trimpots to be able to adjust the trigger voltages - I'm thinking maybe turning on the fan at a sensor resistance of 50 ohms and off again at 100 ohms but would welcome input on this too.

Happy to explain how (I think) it works if anyone is interested.  I don't have a working vehicle to test it with right now so if anyone would like to beta test it I'd make one up...

[Howard]

These are the temp sensor resistance figures for a TR7

Howard

Edited July 9, 2022 by Howard
spelling error

[Adrian Girling]

Thanks Howard, your numbers suggest that switching at lower resistances would be fine. I believe it’s better for the engine to run hot, just not too hot!  I don’t know what the optimum temperature is, but I’ve seen temperature gauges that have a full scale of 170 degrees C

I think my poor old herald might complain at anything much more than 100C 😂 

[RobPearce]

1 hour ago, Adrian Girling said:

I'd welcome anyone who offers simplification

One simplification to offer - I don't think you need two transistors. Remove TR2, R5 and R6 and put the relay coil between Vbat and TR1 collector (where R5 and R6 are at the moment). TR1 would have to be a decent spec - gain of 100, max Ic of 200mA - but that's not hard to find. A typical automotive relay has a coil resistance around 90 ohms.

52 minutes ago, Adrian Girling said:

I think my poor old herald might complain at anything much more than 100C

It should be good for 110C or a bit more, depending whether it's a 7psi or 13psi radiator cap. Fan on at 105 (your 50 ohm ballpark with Howard's measurements) should be OK but making R2 a trim pot (or adding one in series if you prefer) would be worthwhile.

[glang]

Dont most original fit electric fans come on at 95 so I would have thought thats about the temperature the manufacturers are aiming for to maintain optimum engine running...

[RobPearce]

53 minutes ago, glang said:

Dont most original fit electric fans come on at 95

I don't know about "original fit" electric fans for '70s cars but the ones I worked on in the '90s all came on at 105. The threshold will be lower if it's sensing the middle of the radiator, and lower still if it's bottom hose, but Adrian's proposing to use the gauge sender so it's in-engine temperature that's being measured.

 

3 hours ago, Adrian Girling said:

(or sympathy if I've got it completely wrong)

Further to my earlier simplification, there is one change you need. Fit a diode across the relay coil (anode to negative side, cathode to positive) to quench the flyback voltage, otherwise the transistor will be fried.

[Adrian Girling]

59 minutes ago, RobPearce said:

One simplification to offer - I don't think you need two transistors. Remove TR2, R5 and R6 and put the relay coil between Vbat and TR1 collector (where R5 and R6 are at the moment).

Good point, thank you!  I was thinking the relay should be near the fan rather than behind the dash, screwed to the chassis with just one low current wire running to it from the collector of TR2 but no good reason really. 

 

1 hour ago, RobPearce said:

Fan on at 105 (your 50 ohm ballpark with Howard's measurements) should be OK but making R2 a trim pot

I'll do a bit of maths but I agree that a trimpot instead of (or in series with) a fixed R2 would allow adjustment of the turn on temperature.  I was thinking another instead of (or in series with) a fixed R3 would allow the on/off hysteresis window to be varied.  This would introduce the dreaded effect where changing one affects the other - don't see an easy way round that...

 

1 hour ago, glang said:

Dont most original fit electric fans come on at 95

Supporting the need for adjustment!  Should probably add a dashboard LED to show the fan is on.

[Adrian Girling]

12 minutes ago, RobPearce said:

Fit a diode across the relay coil (anode to negative side, cathode to positive) to quench the flyback voltage

Thank you 👍

[RobPearce]

4 hours ago, Adrian Girling said:

I was thinking another instead of (or in series with) a fixed R3 would allow the on/off hysteresis window to be varied. 

You certainly can do that but I reckon the hysteresis is less critical. Ten degrees is a typical figure, but mostly it's just masking the signal variation, really, because there's a fair bit of delay between the fan's action and the sensor's response.