Colin Mill

I fitted a Haze 28 AHr AGM battery that was only £40 and thus far I'm very impressed. As with anything that isn't a Banner there was some messing about to fit it - see here

Hi Rob I suspect part of the reason is that, as far as under-bonnet temperature is concerned it may not make that much difference. I did some sums on this thread and you see that the total heat exiting via the exhaust is only about 50% of that being dissipated via the radiator.

If it were a problem with toe-in it would have to be too little toe-in and if it has mainly been used on road the tendency is for the opposite wear pattern - more wear on the outside edge of the LH tyre and inside edge of the RH tyre (due to the predominant camber of the road). In road use not enough toe-in would be more likely to show up most as excessive wear on the inside edge of the RH tyre. So I would go along with your theory. You mention about a 2mm difference across the width of the tyre. 2mm in about 200mm width is about 1.7 degrees. The distance between top and bottom joints is about 220mm so increasing the length of the top by a couple of mm (which looks like about 2 turns on that thread) would be about right. Edited by - Colin Mill on 8 Jun 2008 10:39:56

Hi Jon Ah, yes, the dreaded joggle valve! From the descriptions from Rover this valve is a float valve that is supposed to open when there is air in the cylinder head and close, or at least seriously restrict the flow of the liquid out of the head once any air is expelled. On mine you can hear it operating and see small amounts of fluid coming through into the expansion tank when you squeeze and release the top hose violently. If it stick shut, and there are reports of ones doing this, it could trap air in the head. I can't find a really convincing explanation as to why this valve is a good thing. On this thread about the only thing we all agree about is that this valve is better removed 😬 Unfortunately, AFAIK it's an intake manifold off job.

I would do the measurements myself and sort it out. Someone with an interest in the outcome and a few straight edges, string and a tape measure will do a better job than some bored guy with the best kit in the world.

Is this the pipe that runs down from the expansion bottle to the 'submarine' section (the metal section between the rad and the pump intake)? If so, then it is certainly intended to be open.

I have effected a reasonably satisfactory repair using self-amalgamating rubber tape (similar to the tape in Alan's link) but reinforced with a wrap of fabric tape over the top. The only difficulty was that the self-amalgamating tape will not amalgamate if it is wet so you do need to get the hose dry first.

Hi Johnty I'm sure removing the joggle valve is a big step forward. I'm sure part of your success is that you have good instrumentation and warm the engine up carefully etc. Without moving the tail of the header to the bypass you would still have the problem that should the stat stick shut you would also loose the pressurisation because, as things stand, the stat inhibits the flow through the header until it opens. This is partly my worry over the standard system - it doesn't seem fault tolerant - designing it so loosing the stat looses you pressurisation as well is a bit of a house of cards. One of the reasons why I prefer the conventional systems is that they can pressurise to a pressure in excess of the vapour pressure of the fluid at the exit of the head. I see this as adding significantly to the tolerance of the system. In the existing system the pressurisation is, at best, only marginally above the vapour pressure of the exit fluid from the head. However, there are going to be variations in the temperature in the head so it follows that there must always be areas within the head that are at a temperature higher than the head exit flow. At these points the vapour pressure will exceed the pressure of the system and since vapour pressure rises rapidly with temperature (about 6% for every deg C!) some local boiling is almost inevitable.

I'm sorry if you find the calculations OTT but I find far too many engineers in the UK simply don't do the sums (unlike their German counterparts). My aim is to explain why I think we need something different to the system as it is currently installed in the K powered Caterham. I would personally be happy if the car had been fitted with a 'conventional' cooling system - I get the impression you would agree... I agree that removing the joggle valve and re-plumbing the expansion tank tail to the bypass would be a significant improvement to the existing system by allowing the expansion tank to more or less track the head temperature independent of the thermostat operation. Perhaps this will be enough though the Rover cap remains in my mind an unknown quantity.

Well, I just like driving (and driving the 7 especially) I don't feel I have to be thrashing it within an inch of its life to have fun. Sometimes it's good to simply enjoy how easily the car covers the ground, using its ability to carry speed through corners - another reason why it's economical perhaps.

On a run down from Manchester to Oxford and back taken fairly gently as the motorways were just too busy to do anything else I got something over 45 mpg out of my 1400 k powered 'Classic'. I'm sure the aerodynamic drag isn't low but since this involves the product of drag coefficient and frontal area and we can assume its not cutting edge on Cd it does at least start off with about the lowest frontal area of anything out there. Its also worth remembering that the Cd values are not so wide ranging - 0.3 is considered good for a production car while 0.6 is what 4x4s achieve.

Lets consider the stresses on the hoses and radiator at 3Bar (3 E5 Pascal) – enough for a 1Bar system allowing for a 50% overshoot and a safety factor of 2. The hoses are 32mm diameter. Consider first the circumferential tension trying to split the hose down its length:- Circumferential tension per unit length = Pressure x Radius = 3 E5Pascals x 16 E-3m = 4,800N/m Now consider the longitudinal tension:- Longitudinal tension = Pressure x cross sectional area / circumference = Pressure x diameter/4 = (3 E5 x 32 E-3) / 4 = 9600/4 = 2,400 N/m (note that its half the circumferential tension as mentioned earlier) The vector sum of these forces = 5,370 N/m Assuming we use polypropylene reinforcement with a tensile strength of 35MPa the reinforcement layer needs to have a thickness of 5370 / 35E6 metres = 0.15mm. So it is not much of a problem to make a hose to withstand the stress of 1Bar with a safety factor of 3. Consider now the stress on the top and bottom tanks of the radiator:- Taking the plan area of the top tank as 60mm x 400mm we have an area of 0.024 sq.m and, at 3Bar a force of 0.024 x 3 E5 N = 7200N trying to pop it off. This stress is carried by the periphery of the tank with a length of 920mm so the tension per unit length = 7,800N/m Even soft aluminium with a tensile strength of only 80MPa only needs a wall thickness of 0.1mm to withstand this! If the Caterham radiators leaked on you it can hardly be to excessive stress – just poor manufacture. We could go on but the bottom line is that the pressurisation to 1 Bar doesn't introduce a massive stress into any component. And of course none of this is an automotive novelty. The 'airless' system with a 13psi stant cap and overflow bottle was what you used to find under just about every bonnet (and hood) in the World – a system that ensures early pressurisation of the system to the cap rating. If the K series had a wonderful reliability record then I would accept that its cooling system design was fine but the truth is it has an absolutely stinking reputation and it is not at all clear if the HGF problems are not at least sometimes a result of the cooling system. From the number of modifications Rover and Land Rover tried on the system there is clearly a significant doubt in that quarter as to the ideality of the system! Edited by - Colin Mill on 5 Jun 2008 09:57:37

Well, I guess you could be doing 120 mph up a 1 in 4 and go over the brow to a minus 1 in 4 the other side in which case the zero g would last about 3 seconds but I reckon the fuel switch tripping at this point might be worthwhile in anticipation of the crash - the CAA would approve I'm sure 😬

I suspect that there may well be some resonance effect with the existing switch that could account for the rumble strips triggering it. I also wonder if the switch is directional and perhaps, as mounted, more prone to vertical g than horizontal g. Perhaps someone who has taken one apart could comment on that. I'm concerned that an accelerometer based unit should not have any loop-hole in the strategy and I would like to see an easy way of checking it periodically. The accelerometers themselves have a self-test pin so the micro can check if the accelerometer thinks its OK. Since they are used in safety critical applications this self test is supposed to actually apply some force to the accelerometer element so a proper front to back check is involved. So, in principle it could have a fail safe action with suitable flashing light warning so you don't end up stripping the car to its component parts trying to get it started when the switch has tripped 😬

Just to put some numbers into the electrolysis side of things:- A few useful numbers:- Elementary charge = 1.6 E-19 Coulombs Atomic mass unit = 1.66 E-27 kg So, 20 amps = 20/(1.6E-19) electrons/s = 12.5 E19 electrons/s since hydrogen is monovalent this liberates 12.5 E19 hydrogen atoms/s The rate of hydrogen evolution is thus 1.66 E-27 x 12.5 E19 kg = 20.75 E-8 kg/s or about 0.2 milligrams/s. The mass of oxygen evolved of 8 times greater or 1.6milligrams/s If the engine concerned is of 2 litre capacity then at 3000 rpm and a volumetric efficiency of say 50% will draw in 25 litres of air per second. This will have a mass of about 30grams and an oxygen content of 6 grams. So, if the product of the electrolysis is introduced into the intake we will increase the oxygen content by about 270 parts per million. The hydrogen concentration will be about 7 parts per million. It's not at all clear how this would make a dramatic change to the outcome of the combustion especially as it has cost about 1/3hp to electrolyse the water. Edited by - Colin Mill on 4 Jun 2008 18:39:54

Yes, that is the important thing and I believe would allow you to have a threshold that would ensure you don't get Robs problem of having a significant "off" that doesn't trip the switch. My guess is that the rumble strips are producing some rapidly alternating, mainly vertical g and this is easily suppressed in the software. I'm actually feeling quite good about my current little accelerometer project (I was pulling my hair out over it earlier today but it's sorted now!) so perhaps I'll be able to make a start on the code tomorrow.

As a bit of light relief from the current accelerometer job I have been giving a bit of thought to this. I guess the criteria would depend on the type of car as downforce would shift the g figures of interest. However, I think all we need to do is decide on a g-force and a speed of interest. To kick some numbers about I guess for any car without aero you would have to reckon anything above 2g in any horizontal direction has to involve nerfing something. So if you start integrating the accelerometer output (to give the change in speed) once this threshold g is exceeded and this integral exceeds the threshold speed - say 15mph - before the g value falls below the threshold again then the unit trips. So at 2.5g this would take about 0.2s for the unit to trip, at 5g about 0.1 seconds and so on. As for the vertical signal I guess if it goes to zero or negative g for say a second you have just parked it upside down or driven off a cliff so it can trip. Edited by - Colin Mill on 4 Jun 2008 13:41:32

If it only tripped in a real crash you wouldn't be very worried about resetting it anyway. 😬 Having said that, what objection would you have to pressing a switch on the unit to reset it just like the current unit ? Since I'm messing about with some 3 axis accelerometers at the moment (for helicopter applications) I might have a go at making something like this.

Because this is the only way of ensuring a reliable increase in boiling point of the fluid. Relying on the pump to provide a pressure in the head is fraught with problems. Firstly, the pump is only capable of providing a pressure differential within the system and so the pump pressure is going to be divided between a pressure increase on the outlet side of the pump and a pressure reduction on the input side. Secondly, the pressure generated by the pump is highly speed dependent. I calculated that the stall pressure of the pump will be about 0.7 bar at 4000 rpm but at 2000 rpm this will be only 0.18 bar - simply too small to significantly affect the boiling point of the fluid even if it was entirely available as a pressure rise in the head which with the bypass flow it won't be. (BTW the bypass has 25% of the cross-sectional area of the radiator hoses and, given the restriction of the radiator itself, the bypass flow is likely to be 20% of the total flow even with the stat fully open and generally more than this so it isn't negligible) A high system pressure is also useful in inhibiting pump cavitation. Remember that it is the absolute pressure of the water at the intake of the pump that determines the onset of cavitation. Once the absolute pressure at the intake of the pump falls to the vapour pressure of the fluid cavitation is very likely. And once you have cavitation you loose significant flow... bye-bye one heat-treated ali head. And to get a handle on this if the intake side of the pump is 0.3bar below ambient (i.e. 0.7bar absolute) then this will happen with the coolant 10 degrees below its normal boiling point! So perhaps I should turn your question around - given all the advantages of an elevated absolute system pressure why would you want anything else? 😬 Edited by - Colin Mill on 4 Jun 2008 08:38:37

Hi Jon You can check the engine is reasonably grounded by sticking the DVM on ohms or continuity buzzer and stick the probe on say the cylinder head and one of the ECU mounting screw on the scuttle. If that seems OK try measuring the volts between the same two points with the engine running and the headlights on. If the ground is poor you will see some voltage between them which will probably vary as the alternator output changes. As for doing the job more accurately it is very cheap. The sensor I'm looking at using, which is good to better than 1 degree C is about 40p and the PIC will cost about 80p. A hand-full of passive components, a PCB and it could be made for a couple of quid - a reasonable profit at 8 quid retail. Given that the standard sensor is about 8 quid retail its seems worth doing the job right. I'm up to my neck in R&D just at the moment (playing with accelerometers which is fun!) but when I get a moment I'll CAD a PCB for this and get a few made up by my tame circuit board outfit. I'm wondering if it's worth putting the sensor in the water flow or simply bolting it to the back of the head.

Well, there are three-axis accelerometer chips available these days that are designed for the airbag deployment market and are thus dirt cheap (about 3 quid in quantity and falling all the time). It would of course be no great problem to set an X g for Y milliseconds criterion and make the thing have a sensitivity that was right for the job.

The ideal thing would be for the inertia switch to tell the difference between a curb and a serious machine-bending arrival. I guess the inertia switch as it is will trip on a near instantaneous event over a certain number of g. I suspect that a modest amount of anti-vibration mounting would ensure that a longer high-g event is needed to trip the inertia switch.

Interesting testimonials! - and I thought the days of the snake oil salesman were over 😬

Hi Jon If you continue to have problems it could be worth checking that the ground voltage at the engine is the same as the chassis (a poor earth on the engine could cause there to be a difference). If the engine were, say, slightly negative with respect to the chassis it would cause the gauge to over read. Conversely, if it were positive WRT the chassis the gauge would under read. Just to complete the info from the testing of the gauge I found that it had an open circuit output voltage of 5.1v and an output resistance of 120 ohms. I guess that, as someone with a passion for instrumentation I shouldn't have started looking at this as I was bound to be underwhelmed with what I saw. However, given how important it is I think I want to improve this - it would be silly to cook the engine without knowing it. My thoughts at the moment are to use an LM35 IC temperature sensor (or similar) and a PIC to drive the gauge. Because this sensor has an accurate 10mV/degC output it would be possible for me to accurately calibrate the gauge at 40, 60, 80, 100, and 120C (with a set-up mode in the PIC to allow these adjustments) and a power-up check (gauge pausing briefly at each of these temperatures before going into read mode). An alarm drive output would be easy too (a switched ground for a lamp or buzzer at some preset temperature. There would be enough drive coming out of the gauge to drive the circuit so I don't need to mess with running a supply wire to it.

I think the problem we have with the standard system with the header filled to the standard depth is that damage may be done before the pressure cap comes into the equation. The air space in the header is so big that a significant void could build up in the head before the pressure in the system rises to inhibit it. The joggle valve in the head-to-header pipe prevents a large flow between head and header so the fluid in the header is going to be cooler than the outflow from the head - much cooler in my experience. This makes sure that the vapour pressure in the header is going to be significantly lower than the vapour pressure at the hottest points in the head. Since the compression of the air in the header is negligible unless you overfill the header we have a recipe for vapour-filled voids standing the coolant off from the hot points of the cooling ways in the head. I feel that the current system is only safe if the temperature at hottest points in the head are below the un-pressurised boiling point of the coolant. Peter's measurements in his original post pretty much prove this: he said :- "The pressure drops close to 0 bar and sometimes shows negative relative pressure after the thermostat opens" Exactly not what you want with a fully warmed up engine!! My temporary solution - by no means ideal - is to overfill the header so that:- a) water expansion causes significant compression of the air in the header causing a quicker initial pressure rise that does not rely on the header becoming hot. b) should localised boiling occur in the head the further compression of the airspace in the header rises the pressure more rapidly than it would with the normal level in the header thus helping limit the volume of the voids in the head. Edited by - Colin Mill on 2 Jun 2008 22:47:54