Electric fan switch hysteresis?

Yup. ;-)

Again, agreed.

There is an extra bit I can add to that. Initially I had a heat deflector (a piece of ally sheet that sat on the thermal block and under the carb ... or it might have been under the thermal block, can't remember) that helped deflect the heat from the water heated manifold, away from the carb float bowl. A while back that snapped off (work hardened from vibration) and now I see this carb overheating slightly more when the engine is generally running too hot, more than I did when it was still there.

Well, it isn't 'buggered' (in that it is new and still 'works'), but yes it could be buggered in that it doesn't have the hysteresis it is

*supposed* to have (for good or bad) and so never lowers the coolant temperature to the lowest range in it's designed settings. I think that switch did exactly the same as this new switch and why I bought *another* new one. ;-)

Yes, both a good and less_so_good ones. ;-)

Ok.

I can't quite remember the coolant path here (till I go and look) but I think the manifold shares the path with the cabin heater somehow and so isn't in the main 'cooling' path as such?

Whilst I don't know for sure, the rad isn't that old (certainly less than a few hundred miles since being worked on by a rad specialist) is double core and looks and feels very clear when you drain or flush (the water just drops out of it!). ;-)

The switch in question is actually mounted on the top of the thermostat housing (like an early Fiesta, same engine but transverse) and so even if there wasn't much 'flow' through the rad, would likely be the first thing to get hot, if anything does, ;-)

When I first built the kitcar, I turned up a short ally tube and internally threaded (M22 x 1.5 I think) it to take one of these std fan switches. I then put a new (Mk II Escort) thermostat housing on the faceplate of the lathe a bored it out to be an interference fit with the 'tube'. I then used 'Lumiweld' to put a fillet weld around the outside. I think that lasted about 10 years but when changing a tight fan switch, it twisted the tube out of the housing so I cleaned it all up and put it back together with a 'Liquid metal' type two part epoxy. It's been fine ever since (even when changing the switch recently). ;-)

Anyroadup, thanks very much to all who have offered thoughts and advice so far. You never know what things one may have forgotten or overlooked so it's all being considered and even checked out.

I'll try to take a picture of the actual thermostat housing and switch assembly as I'm sure that will help people visualise it all better. ;-)

Cheers, T i m

Yeah, that would be nice. ;-)

I wonder (OOI) if it actually has the hysteresis they say it has and if they would confirm that my tests were valid? I'll ask. ;-)

Cheers, T i m

[snip]

The mechanical thermostat is irrelevant to this discussion. When the car is stationary and the mechanical thermostat is open, there is negligible cooling because there is no air passing through the radiator - as you have explained.

What happens is that there is significant thermal capacity in the system that even with the cooling fan running some parts of the engine continue to heat up. So the temperature rises above the set point of the "thermostatic fan switch" which should ensure that the fan continues to run. As the engine cools the temperature at the "thermostatic fan switch" drops to its "off" value and the fan stops. There may now be a body of cooler water in the radiator which is pumped around the engine, causing further cooling.

Thus the temperature at the "thermostatic fan switch" oscillates about a value near its set point - could be either above or below. If the "thermostatic fan switch" had some hysteresis the oscillation frequency would be lower, and the temperature would oscillate about a lower value.

Your initial tests suggest that the new "thermostatic fan switch" does not acually have hysteresis. So it is maintaining the water temperature at a value quite near its "on" point. Unfortunately this is too hot for the comfortable running of the engine. It's not clear what this temperature is, or whether there is a temperature difference between the sensor for the gauge and the "thermostatic fan switch"; but the most obvious solution is to fit a "thermostatic fan switch" with a lower set point.

Given that these are apparently not available, artificially applying hysteresis will help reduce the average temperature.

Ok. I was just trying to consider how hysteresis was present without the thermostat (or thermal fan switch etc). Obviously there would be some with either and at certain points in time.

Agreed.

Understood.

Ok.

Yup.

Ok. Because in this case (and I know you aren't talking about this case specifically) the fan doesn't cover the entire radiator (and it isn't cowled like some are so they do effectively cover the entire radiator, even if not lineally) there will always be some mixing of water that is cooled by the location of the fan and some that isn't. This plus the thermal mass of the hardware means there probably wouldn't be such an obvious 'plug' of cooler water found anywhere (I'm not suggesting you said their would be, just refining your concept a little in my head). ;-)

Ok (where there was no hysteresis in the switch).

Wouldn't it fluctuate around it's hysteresis range? Once the switch had gone off, wouldn't the temperature would have to rise significantly (~9 degrees) before it turned back on again (on a switch working as it should I mean)? ;-)

So it seems? ;-(

Agreed, when the vehicle is not moving or subject to a breeze. ;-)

Well, that is certainly the case for the switch currently fitted and I believe this new switch (as yet unfitted) has the same (lack of hysteresis) characteristic, even if it has a different (possibly lower) operating temperature.

Well, (FWIW etc), one (from memory atm ... the temperature sender, also fairly new) is on the side of the block, just under the thermostat housing and the other, the fan switch is just above it, over the thermostat in the thermostat housing. Once the stat was open constantly I would expect the temperature both were experiencing to be fairly similar.

Agreed, and both times I bought them recently that's exactly what we looked for (at the car spares place).

Yes ... or finding a switch that actually does what it says on the tin at least. ;-)

I believe that IF the new switch I have here was to switch off at the lower temperature it suggests it should, I wouldn't have a problem. 88 DegC is quite a way off of boiling (even on an open system) and the lower 79 degrees may be too low and crossover with the range of the thermostat. However, what sounds likely in theory isn't necessarily how it works in practice. ;-)

Cheers, T i m

[snip]

The maximum temperature may be some way above the "on" temperature, and the minimum temperature may be some way below the "off" temperature, so that the average value will be somewhere between the two. The exact value will depend on the rates of rise and fall so the waveform might well be an asymmetric sawtooth.

My point being that the temperature will oscillate about the average value, but you can only generalise about the exact maximum and minimum. They may well be further apart than the hysteresis range of the switch

- because of the hysteresis in the rest of the system.

I car I had a while back had its temperature gauge marked - I would not claim "calibrated". With ordinary rural driving it showed about 80 degrees. Driving at 70mph for many miles on a motorway in hot weather would cause it to rise by a couple of degrees.

Inching in London traffic in hot weather would cause the temperature to rise to about 95 degrees, then the electric fan would run and the temperature would drop rapidly to about 90 degrees (in less than a minute), then rise slowly to 95 in about 2 minutes. The temperatures are an estimate - the gauge showed 80, 100, and 120, I think.

Ok. ;-)

Ok.

Ok.

I'd be happy with that sort of variation. ;-)

Again, that range would be more than acceptable to me. The last time I played with all this I did actually have a thermometer in the top of the rad and felt for when the thermostat first opened etc. I may do the same thing again soon (with the addition of the IR thermometer) and make some notes. ;-)

Cheers, and thanks again for all your input Graham. Very interesting. ;-)

T i m

[...]

Slightly OT, but most cars made in the'ECU era' do not really have temperature gauges; the indication is just tri-state.IOW, when the gauge indicates 'normal', the engine temperature is within a range that won't cause damage.

The reason for this is that modern engines operate satisfactorily over a much wider temperature than those of old. If a gauge indicated the extremes of that wide range, owners would be forever taking them back to the dealer as over- or under-heating.

Chris

Are you talking about things that look like conventional linear temperature gauges Chris (but aren't)?

Makes sense ... assuming they bothered to look at them in the first place (and why I thought some cars just had a 'Too hot' indicator)?

Cheers, T i m

[...]

Yes, or look like conventional dials as fitted to cars of yore.

I doubt that an actual device showing *temperature* has been fitted to a mass-market car in the last two decades.

Chris

Ok.

So do you include even indirectly in that? eg Temp send or > ECU > Instrument CPU > Temperature gauge with 'realtime temperature display'?

When our 04 Meriva had it's ECU connection faults we would intermittently lose the temp, revs and speed indication (the fuel gauge was fine). When playing with the OpCom OBD diag tool thing I was able to increment the temperature gauge through small steps (can't remember what) so I think ours was a direct electronic translation of the real-time temps, rather than something 'simplified'?

Not 'direct drive' that's for sure. ;-)

Cheers, T i m

Well, even a tri-state led could be said to be 'showing' temperature.

The sender is probably going to be something reasonably linear like a thermistor. The ECU output to the gauge won't be; it will be whatever the software writer has decreed the gauge should show as OK.

The fact that it is possible to drive the ECU output to whatever state you want would have no bearing on how the ECU software translated the input into a gauge reading.

Chris

Well quite , but not in the same way as we were talking about above? I was trying to qualify your suggestion re cars with non-direct-temperature-reading gauges and what you were actually including under that heading.

Sorry, I wasn't questioning the *actual / scientific* linearity of the system ... but what systems were displaying actual temperature rather than a dumbed-down version of it?

Sorry, I think we are at crossed purposes here. I was asking, of the vehicles your referred too 'over the last couple of decades' ... how many actually showed the temperature (or as near as) versus those that took what looked to an observer to be a traditional temperature gauge and modified that into a dumbed-down version (cold / normal / hot etc, if that's how it works)?

I was asking because if I was trying to determine if (say) a thermostat was going off spec, I couldn't do that if the 'system' was not truly representing the temperature (as long as it stayed within the preset constraints for what was considered acceptable for each zone etc)?

Our Meriva for example (made within the last two decades) seems to have a 'straight' temperature gauge (no matter what process is involved to achieve that)?

Cheers, T i m

[...]

None that driver could see. The ECU would 'know' and no doubt being able to access data further then mandatory EOBD codes would let you read the actual temperature according to the sensor.

Many recent-ish Fords have the ability to show additional dashboard information using the trip mileage display using a 'trick' intended for service staff. That will show a wide range of values on my Focus, all whilst the gauge stays dead centre.

If you drive a variety of cars, you can start from cold, and within a couple of miles the gauge shows normal. It can't be up to full operating temperature in that time. If you then go for a brisk drive, then leave it idling till the fan cuts in, the gauge will still be in the same place. It's implausible that the actual temperature will have remained constant.

Early 90's Escorts certainly used the method of temperature display I have outlined.

You could not use the 'temperature' gauge on any car that used such a system to tell you anything significant about the thermostat other than that it had failed completely in one state or the other.

Sorry, I couldn't tell you anything about a Meriva in that sense.

Some base-level Vauxhalls don't have a gauge at all however.

Chris

Ok.

Quite ... and that supports your 'simulated readings' idea. ;-)

Ok.

Ok thanks, that was what I was wondering.

Ok. Well, FWIW, I think it must have a straight / electronic temperature gauge as I was able to determine a faulty thermostat with ours. Light throttle, medium speed and cold day and the temperature gauge showed lower than normal. Stop in traffic and the gauge climbed as expected till the fan cut in.

Yes, I have heard of some cars where they only have a red 'Engine too hot' indication but the Engine Management takes over (and goes into limp mode or shuts off a cylinder or summat) in any case. ;-)

Cheers, T i m

[...]

Modern (as in the last 15 years!) Fords are quite sophisticated in that respect.

They have a temperature 'gauge'. An increase in temperature will attempt to use the radiator fan(s). If that doesn't reduce coolant temperature, the 'gauge' will go into the red. If temperature stays high, the MIL will illuminate, and alternate cylinders will be prevented from firing every other power stroke. RPM will be restricted to 3,000. Continued rise in temperature will cause the MIL to flash for 30 seconds, then the engine will be shut down.

Chris

No because the thermostat would prevent it happening

Agreed

Yes once they get something into he electronic brain its a royal PITA to sought

That's what I thought. ;-)

I now seem to have trapped a nerve in my spine so I'm not doing anything for a while, especially leaning over the bonnet of a car outside. ;-)

As an aside (and in spite of what I may have said elsewhere), I've ordered a cheap programmable temperature switch off eBay as at least (in principal) I can set the upper and lower temperatures to whatever 'works' for my particular car in the real world. There is already a relay driving the fan so the relay on the PCB won't have that much to do. I may even be able to modify one of the existing fan switches to accept the new temperature probe (pull all the guts out and 'pot' the probe in some thermal compound of some sort). At least then I'll only have to keep a spare switch and adjustable to be able to go back to the 'old way' at the roadside. ;-)

I think the board I've ordered uses a tiny microcontroller so we will see (if I ever fit it) how well it holds up in such an environment.

Cheers, T i m

But they have coolant systems with electric pump(s) that continue to run with the engine off.

And for proper stop start it needs a lot more mods.

A Pinto has an engine driven coolant pump. Switching the engine off will reduce pressure in the head created by the restriction of the by-pass and T-stat to coolant flow. That is typically 1bar greater than the cap pressure. Without flow there will be local boiling of coolant in the head.