Unintended results...

My manufacturing partner came up with the idea of combine reservoirs for packaging reasons. Instead of having individual reservoirs for each damper, he wanted a 'front' reservoir and a 'rear' reservoir on the basis that it's easier to route pipes around an engine bay than to find homes for two large reservoirs.

Now this idea has a couple of distinct advantages. The first is that if we keep the combined reservoir vertical and large enough, we won't need to gas pressurise the fluid at all which will make life inordinately easier for the end user, although there would be an increased risk of cavitation. If you do decide to pressurise, it'll be a lot easier to bleed air out of the system with a single reservoir at the highest point of the hydraulic circuit.

The second advantage came to me a little later. When the car rolls there will be effectively no change in the working volume of the dampers as what goes down on one side will go up on the other. Only when the suspension pitches or heaves will there be a change in the volume. Now, if we replace the gas spring with a stack of belleville washers, then we're effectively adding an extra spring which only works in heave. If we restrict the flow into the reservoir we also have extra controllable damping of that heave motion.

So the rendering above is a revised system that does just that... those out there who like lateral thinking might like to consider what would happen if we connected the dampers diagonally rather than side-to-side. And if you're really interested check out this link for some Spaniards who've been doing pretty much the same thing.

Halfway there...

There... that looks a little prettier than the original 'let's take a big hunk of billet and turn most of it into swarf' valve block. All it needs is a pair of pipes to connect it to the cylinder and we'll have a working damper. I've change the design slightly in that instead of using a hex and a slot to do the adjustment, we just have nicely knurled knobs..

As in all these things, the devil is in the details and I've just spent a useful hour best part of a morning making sure these things can actually be built and put together. There's nothing quite like realising that your O-rings will never seal because the bore is too large or you can't actually insert the springs and valve seat into the canister because the hole is too small. For those without a copy of thread tables and the like to hand, I can thoroughly recommend the RoyMech website (http://www.roymech.co.uk) as somewhere to find all those useful little diagrams and measurements.

Anyway, just got the last parts of the other half to sort and then the tedious job of creating engineering drawings for manufacture.

Valves...

Although I've still got 4-way adjustment (high and low speed bump and rebound), there's still going to be times when there isn't enough adjustment available within the bleed and blow-off valves to match into the car. So, instead of the solid piston and through rod design of the original, the picture above is of a standard shim valve and orifice piston. This will be the ultimate determining factor for high speed damping - which gives the option of using the high speed adjusters to control the 'knee point' of the damping curve.

Virtually every damper uses shim valves somewhere along the line as, although they're horrendously non-linear (although a quick study of the various text books will give the suitable formulae - sadly the IMechE site is down at the minute, so I can't tell you if it's in Roarkes or not) they're a nice consistent non-linear, which needle valves tend not to be. The disadvantage is that, other than by supplying a pre-load (and a lot of cheaper dampers provide adjustment this way), you can't change the characteristics except by cracking the damper open and changing the shim diameters, stacking arrangements and thicknesses. For a top end damper, you often get a choice of 30 different shims to build into your stack, which gives an eye-watering number of options and excessively long times spent playing with the damper dyno trying to get the characteristics you want.

The basic design is still going to be twin tube, with all the flow passing through the valve block. The problem with this is that as the damper moves in and out the volume of the working chamber decreases and increases. Now hydraulic fluid is not renowned for it's compressibility, so we need a reservoir to hold this extra fluid. In 'proper' twin tube dampers, the annular ring around the working chamber does the task, but I want these dampers to work irrespective of orientation - the classic way of telling that a twin tube is a twin tube is to invert it, pump air into the working chamber and feel the lack of damping that occurs. So we'll need a second floating piston somewhere, with a gas pressure chamber on the other side.

Gas pressure can be a useful thing, in that you can use it to support some of the weight of the car and you get a slightly (more if you don't have a big enough chamber for the gas) rising rate as well. WNTL? It's another non-linearity which may need tuning, and I don't have huge amounts of free time...

Back to the drawing board...

OK, so I got a price back from my tame manufacturing partner for the design of the six way adjustable damper which made my eyes water a little and, notwithstanding the fact that it was for an engineering one-off, even with economies of scale wouldn't really be viable in the sense of being able to sell it to any target group other than rich idiots.

So we're off on a voyage of discovery into how to make something cheaper. The first thing to go is the plethora of adjusters. I'd envisaged lots of precision drilled barrels which would give consistence between units. Of course if you take a cylindrical component that can be turned out a huge rate and then have to carefully mount them in a dividing head and drill 16 holes (8 for the indexing mechanism and 8 different sized ones for the orifices) in exact positions, then it's going to cost a lot more to build one. If you're building thousands then you sort out jigs and fixtures, but I can never foresee this being a mass production item, not matter how bling it is...

Now, I still want separate adjustability of high and low speed damping at a sane price, so that I can take a unit off the shelf and valve it for most applications. If we can't have miniature drilled orifices to squeeze hydraulic fluid through, then we'll need a needle valve that you can screw in and out to change the size of the orifice. Not as repeatable and you'll need a damper dynamometer (a few thousand pounds) to do setup.

High speed adjustment will be similar to the original design... OK, so a short CAD session later we have mark II of the adjustment valving:

I've shrunk the component count down from eight individual bits down to three (not including the coil spring and various sealing O-rings) and there's no nasty indexing and drilling required. With the exception of an exhaust hole in the blue component and the hex adjuster on the top, it's all lathe manufacture and thus reasonably rapid (and hopefully cheap)

Away from the tiny precision bits, I'm rationalising the design of the main body. Instead of having a massive boring job, I'll use standard off the shelf tubes interfacing with turned caps. Minimal cost, and bar a few threads, no machining. It'll all be fine...

...tumbleweed

has been blowing through this blog for far too long. I could blame pressures of work (somehow I've ended up with four jobs at the same time) or the fact that some of my creativity has gone commercial. Anyway, the pretty picture above is one of the fruits of all that labour.

One of the four jobs has been developing suspension kits for various people. The problem with doing this is that you often have to make do with whatever a damper manufacturer has sent you, particularly in terms of adjustability. Sometimes you get lucky and you can get the dampers set exactly how you want them and other times you just can't get the rates you want.

So, in finest engineer fashion (i.e. slag off the competition, insist you are the only person on the Earth who can actually do the work properly... have no dress sense or sense of rhythm), I've decided to fill my gap by designing a damper to work just the way I want it. I've even got a manufacturing partner lined up, which is a degree of organisation I'm not renowned for.

So, quick run down of the features:

1) It's a through rod damper. Virtually all dampers aren't, so as the piston moves into the body, the shaft it's attached to changes the working volume, which has to be accommodated. Two common methods are employed. Either you have a second concentric chamber as a reservoir (a twin-tube damper) or you have a second piston with a gas spring to one side which can be compressed to take up the slack (a monoshock). This one has the piston in the middle of the shaft and a seal at each end of the working chamber. As the piston moves, there is no change in working chamber volume. Very useful as it means the design can work on low pressures and therefore react quicker to changes in velocity. It also means that the seals can be off the shelf O-rings and DIN seals.

2) No internal valves. The attractive block bolted to the side carries all the valves. So if you want to change something fundamental, you don't have to strip the whole thing down, you can simply depressurise the unit and swap the valves about. Fluid can only flow from one side of the piston to the other via the valve block. This means you get a large flow of fluid and it's far easier to control a large flow than it is a small one.

3) No shims. Commonly, dampers use thin sheets of metal blocking holes at provide the damping force. These tend to be horrendously non-linear. This makes tuning an interesting affair and lots of rebuilding tends to go on. I've engineered this using simple tappet valves and adjustable orifices.

4) Six way adjustment. OK, this is probably overkill. As well as controlling the gradient of the Force-velocity curve in high and low speed situations, you can adjust the knee point where the two curves switch over. All of that adjustment comes without stripping down valves, which makes it ideal for someone with very little time. Of course having six different knobs to twiddle makes it a nightmare for someone with no idea what they're doing.


I've also tried to make the unit as manufacturable as possible. Similar units from Koni and Ohlins retail at around £500 each, and I'm trying to get the price point to something around 40% of that and if any machine shop with a CNC mill and lathe can churn these out in their thousands, then I'm laughing.

First New Bit...

In finest bottom up design methods, I'm starting at the end and working my way forwards. The nicely rendered bit above is the output shaft for the reversing box. It'll sit coaxially with the input shaft (hence the big hole in the middle) and can be driven two ways. The first (and most common) way will be for dogs to engage in the slots on the front and drive the output shaft directly. These dogs will be on a sliding collar splined to the input shaft. In this fashion there will be a non-geared direct connection between engine and output shaft.

The second way will be for a gear to turn the output shaft. If we take the drive from the input shaft via a pair of gears to a layshaft and then via two extra gears to the gear on the output shaft then we'll reverse the direction of travel of the output shaft relative to the input shaft and hey presto - reverse gear. Now given that bike engines tend to major more on power (by virtue of stratospheric rev limits) than torque, a degree of speed reduction (and hence torque multiplication) might well be a good thing as the driver won't need to necessarily slip the clutch mightily to get the thing moving in reverse.

Now for all this to work with a minimum of nasty mechanical graunching noises, all the gears will have to be in constant mesh. This will mean that the gear on the input shaft must float and only be connected when we want reverse gear. If we use the other end of the sliding collar to do the connect then we have a workable design. Everything else is metallurgy and calculations...

Oops...

I know it's been a bit of a while but somebody pressed my 'get a life' button again and I've changed jobs (while still teaching as well) and a home (re)construction project has been filling what is euphemistically termed spare time. Never mind, I'm back now and I suppose I had better get on with some non-paying real work.

One of the requirements for RGB racers is a working reverse system, something that the majority of bikes (and all bikes if you don't class a Honda Goldwing as a motorcycle) don't seem to have fitted. If you've gone for a longitudinal installation of your engine then it's relatively easy to have an extra gearbox between output cog and differential. Unfortunately, I seem to have plumped for a mid-engine layout which makes for a much simpler differential layout (using chain and sprockets just like the donor bike) but does hamper the ability to go backwards under the influence of the engine. A lot of racers use a second starter motor acting on the drive chain but these seem to have a relatively large failure rate (and failure when tested is an automatic disqualification) not to mention issues with how the torque is delivered. The fact that I'm a mechanical engineer rather than an electron herder seems to be pushing me down a purely mechanical option.

So knocking out a quick specification for a black box, I get the following list of desirable features:

1. Lightweight and compact
   
2. User switchable between forward and reverse
3. Minimal transmission losses in forward mode

And that's about it... I'm still at the brain storming stage but I can envisage a system that is basically the reverse gear and top gear from an old gearbox. With one of these you have direct drive from input to output in forward mode and a small geartrain in reverse.

Adjustment...

Sorry there's not been a post for a little while but life has been a little frantic around here, with a mixture of the start of term (my beloved leader is currently tearing his hair out as we've got about 50% more students than we needed/wanted/expected) and the need to rebuild our outhouse before it collapsed taking the house with it. Even worse than that, I haven't got a SolidWorks model to throw up as a picture. Normal service will be resumed soon.

Anyway, getting back on track, I want to make my suspension design adjustable for two reasons: firstly, I have no doubts that any chassis that gets manufactured is going to have dimensional accuracies measurable by thumb widths and the suspension will need to be tweaked back so that it matches what was originally designed. I have a friend who makes recycling lorries for a living and he works to plus or minus 5mm. Secondly, you will want to fiddle with camber and caster settings to get the right grip balance for corners (why you'd want to tweak them will come later when I do the spring rates).

The classic way of performing suspension adjustments is to screw your balljoints (If you're in a cheap formula) or Rose/Heim joints (as stolen from the Luftwaffe at the end of World War II and gifted to either of those two fine engineering companies - one UK and one US) if you're working with cash. If you happen to follow Formula SAE, you'll now that one thing that makes judges tear their hair out is the REIB problem - 'rod end in bending'. Formula SAE goes for light weight - very light, as any car over 200 kg doesn't get through to design finals - which means that the rod ends would look ridiculously tiny on remote control cars. If the rod end isn't fully screwed home the stresses on it cause horrendous distortion and early failure.

I'm unlikely to be using 6mm joints on any design soon, so we can simply spec up the joints to account for any excess bending - in effect over-engineering the joint. If it does bend, it's probably better that it deforms rather than the suspension arms or, god forbid, the chassis. Rod ends are relatively cheap (and as an aside, don't engineer anything in brass for the foreseeable future - I've just ordered a load for the workshop and it's twice the price of Aluminium) and more importantly, easy to replace.

By screwing the joint in or out we can effectively change the length of a suspension are and thus the camber angle of the wheel. If we just have a simple threaded suspension arm, we have to dismantle the suspension to adjust it. This is not a trivial exercise, and more importantly it at least quadruples the effort required to get a car squared away as you have to go through vast cycles of disassembly-adjustment-reassembly-measure to get the numbers you're looking for. Many cars have adjustment ladders - effectively a turnbuckle with a left hand thread at one end and a right hand thread at the other. These add to the length of the arm, and more importantly add extra weight as you have to over engineer to potential joints in bending.

There is another option - a coaxial turnbuckle (I promise to forget about students tomorrow and CAD one up for your general delectation). All the threads are co-axial so adjustment is made by turning the middle portion. Clever thinking also tells you that you can play silly buggers with the thread pitches to give yourself significantly finer adjustment than a normal fine pitch thread.

As for caster, I considered all the sane and insane options and decided to go with having spacers on the inner mounting arms and thus the ability to shift arms longitudinally to angle the upright. SusProg suggests that I'll need around 20mm of movement in the top arm to get 7.5 degrees of caster. Any more than that and the driver will need to go to the gym a lot to be able to turn the wheel accurately - even with the low weight.

Making sense of the Regulations...

Now, I know that I'm supposed to make sense of regulations, or at least try and find all the useful loopholes in them, but the RACMSA 'Blue Book' makes about as much sense as a bucket of chocolate frogs. I'm trying to work out how small I can make the tubing for the spaceframe chassis, and depending on which interpretation of the rules you like. Section Q covers all the safety features you could or should fit on a car and paragraph 1.3 mandates CDS steel tubing, with a minimum yield strength of 350 MPa and either 45x2.5 mm or 50x2mm diameter and wall thicknesses. Yet paragraph 1.5.2 for Sports racing cars says a minimum of 48.3mm with a wall thickness of 2.6mm and 1.5.4 (b) for non-standard cars says 32x1.5mm. Which is right? I think we should be a Sports racing car rather than a 'non-standard' car, but there are significant weight penalties for using thicker than necessary tubing.

So if we look at the cross-sectional areas of the four options we get values of 333mm², 301mm², 373mm² and 143mm². The maximum compressive force that these members can take before starting to permanently deform (for those who remember A-level physics, the equation is of course Force = Stress x Area), varies between 50 kN for the lowest member and 130kN for the largest area. So which is adequate and which will give you a few short microseconds of terror before a basilar skull fracture puts you out of your misery for good?

The human resistance to acceleration was found out in nicely empirical fashion by Dr John Stapp of the USAF who strapped himself to a rocket sled and pulled 46.2G in a frontal deceleration. Using Newton's second law (Force = Mass x Acceleration), if we get our car down to the minimum weight of 560kg, we get an impact force of 254 kN. If that impact force is suddenly applied (and you can bet it is), the initial impact stress is actually double that (so effectively we can only use half the limiting impact force if we want no margin of safety). I'm vastly oversimplifying the reality, but SolidWorks doesn't come with a copy of DYNA3D or PAMCRASH, which is what the automotive industry use for this sort of thing.

Anyway, if we have two roll hoops, with extra front and rear stays, we'll have 8 struts resisting an impact force.
If we use the smallest diameter tubing the cage will yield under such loading, whereas if we use the largest tubing we have a factor of safety of around 2. Is it really worth saving weight if you get one of those once in a lifetime accidents - and you want to be around to tell the grandchildren all about it? I'd be inclined to use the lightest of the large tubing (which has other advantages, such as being less likely to buckle than the thinner diameter - so that's 50mmx2mm thickness round tubing.

For frontal impacts we can trade deformation for force - i.e. limiting the force and hoping the car stops before the metal tubing makes contact with our delicate feet. F1 expects a peak rate of deceleration of 10g for the first 150mm of deformation, and 20g for the first 60kJ or energy absorption. For our car 20g and assuming that our non-CDS metal tubing (so a yield strength that could be as low as 275 MPa) will yield under such an impact, we need an impact force of 110 kN and thus a cross sectional area of about 400mm². There will probably be 4 frontal force members here (and a nice big chuck of energy absorbing honeycomb to keep the force at those levels for at least the initial phase of the impact). So each tube needs a cross-sectional area of 100mm². So it's a case of running through the sizes to work out what tubing will work in these areas. A quick bit of maths lets me know that I need tubing that weighs around 780 g/m. In circular tubing that means 7/8"x16g, in square that means ¾"x16g and in rectangular we can use 1"x½"x16g.

I finally got out to measure someone's chassis and bodywork at the weekend and the picture at the top is a quick revision of the chassis, less the back end and using thinner tubing. With a 50mm diameter cage on the top, but using only ½" round tubing, the whole lot weighs around 34kg. The chassis was using 1" tubing and had a lot of extra tubes knocking around (it was originally a prototype chassis), so weighed considerably more. Even with a ramping up in tube size, I reckon we can get significant weight savings in place to get the car close to (or even below) minimum weight.

Safety

Now no racer would ever imagine that they're going to have a really BIG accident, and certainly one where they impact the rollbar at a really nasty angle. I spent the thick end of a decade investigating RTAs and I've seen the impossible happen quite a lot. So, I wanted to see what my allegedly natty rollbar design really would do in the event of a shunt. I've loaded the top part of the rollbar with 10kN (effectively a tonne) and at the same time applied the same level of force horizontally - sort of landing upside-down while sliding into a brickwall with just the rollbar taking the loading. This is nowhere near as sever as F1 for example, which expects loading of 90kN through the rollover structure.

With this loading the rollbar only just survives - the factor of safety, i.e. the ratio between yield stress and max stress, is 0.97 so some parts of the rollbar would just start to bend. Dynamic crush testing is way beyond my knowledge of Cosmos, but I'm more certain than ever that I'll want either front stays on that roll bar or a full roll cage tying the rollbar to another roll hoop around the steering wheel area. Once I've got the rest of the chassis, I'll repeat the analysis and see what the whole car will do. I'll also see if I can get some dynamic analysis done using Cosmos.

The other thing I want to do is have sensible impact protection to the front and sides. Aluminium honeycomb is your friend here and I've got both Hexcel's design document and the F1 regulations as an input value:

"For the purposes of this test, the total weight of the trolley and test structure shall be 780kg and the velocity of impact 15.0 metres/sec.The resistance of the test structure must be such that during theimpact:
- the peak deceleration over the first 150mm of deformation does not exceed 10g;
- the peak deceleration over the first 60kJ energy absorption does not exceed 20g ;
- the average deceleration of the trolley does not exceed 40g ;
- the peak deceleration in the chest of the dummy does not exceed 60g for more than a cumulative 3ms, this being the resultant of data from three axes.
Furthermore, there must be no damage to the survival cell or to the mountings of the safety belts or fire extinguishers."

There's similar for side impact as well. I'll see what sort of volume of expanded honeycomb is required and see where it can be fitted in. I've seen some nasty intrusions into spaceframe chassis over the years and it's a risk I'd like to design out as much as possible.

Making progress...

As promised, I've got back onto SusProg to work out important things like position of pullrods, spring rates and the like. First problem was that pull rods simply weren't going to work as a triggering mechanism. Where I've put in the zero roll suspension there was effectively bugger all motion translated from the pullrod into the monoshock assembly to get any meaningful translation of a coil spring. This is not a good thing. I actually want the suspension to move, and more importantly, move under control and 10 mm of spring travel could be taken up by stiction, clerances and poor manufacturing.

So, pushrods it is, albeit at the expense of a little increase in centre of gravity height. That said, SolidWorks is reporting that the current unit is around 10kg, so I'm not unduly concerned with that minor growth. I used SusProg to work out suitable spring rates. I've gone for a ride frequency of 2 Hz which requires a spring rate of 310 lbf/in in this instance. 2 Hz is about the upper limit for ride frequency until you start getting ridiculous amounts of downforce, and I'll probably stiffen up the rear ride frequency by around 10% more. That can wait until I've actually got some ballpark weights to play with, rather than the 'pick a number out of the technical regulations and use that' option.

The picture at the top shows the layout of the front suspension in SusProg. SusProg can't actually calculate values for a monoshock, so it's really two separate bellcranks with coincidental spring and damper units. Still, it looks pretty and gives me some values, which is all-important. The key value is that for 50mm of bump and droop travel, I actually need around 85mm of damper travel, so the 40mm travel damper I've been playing with will need to be replaced. Creating a new damper is not of itself a major issue as all I have to do is change one number in Solidworks, but I now have the lead on issues of changing everything else to fit the central guide rail. The only ray of light is that the equivalent suspension travel in roll is going to be about an inch either way, so I'll be able to keep the guides very short and limit the bending moments on the shafts.

So, I've pretty much got the front sorted out now and we can think about putting the chassis together. Unlike certain 3D solid modellers, SolidWorks has a built-in facility for 'weldments'. That's spaceframes made out of random tubes to you and me. You simply have a separate drawing of the tube profile and a 3d wireframe sketch of where the tubes have to go, marry the two together and you get something like this:


You can even add mounting plates, weld fillets (there's a few on the drawing where I've joined said plates to the tube) and trim tubes so they don't appear inside each other. All-in-all, something pretty much designed to make a chassis designer's life easier. Of course until I measure the bodywork, I can't really put a design together, but that mockup meets the MSA requirements and weighs 12.3 kg. The basic concept for the chassis is to use that heavy rollbar to link into the engine mounts at the front and the differential and suspension pickup points at the rear to reduce weight as much as humanlymachinely possible. While I'm waiting to go and measure (work is intervening for the rest of this week), I'll get the modal suspension finished off...

More integration...

and a lazy way to do calculations. The Educational edition of Solidworks has the full-on CosmosWorks designer built in to it, and FloWorks, but I'm fairly sure I don't have enough processor cycles in my PC to sanely do any meaningful CFD studies. I thought I'd check out the possibility of integrating the body-side mounting of the heave spring into the big lump of metal that slides from side-to-side when the car rolls. A quick analysis later (and it took less than a minute to cue it up and run it - far, far better than when I used to use FEA in the days before Pentiums - I first used PAFEC on a Vax and IBM CAD on a mainframe) and you get the picture above. This is the von Mises plot, which is effectively a combinational stress criteria for yield (if you get yield when you combine the stresses in all three dimensions using the von mises formula, it'll probably yield in real life) when 5000N (a bit more than 500 kg) of loading is applied to the bolt hole. If you click on the drawing you'll get a better view, but the safety factor (i.e. the ratio of actual stress to peak stress) is just over three, so even if shock loading was applied, you'd still not cause that lump of metal to yield.

So, knowing that I'm not overloading the lump in question, I can clean the design up a bit, which gives the picture you see below:


I've shortened the guide rails as much as possible (I'll be running the geometry again to see what motion ratios I have and how much roll travel I actually need) to limit the bending moment on them and put a simple framework around that can be bolted onto the chassis in the most appropriate place. I'm now almost ready to start designing the chassis. Adrian has offered to let me measure his bodywork, so I know what space I have to play around with, and I've got a copy of the Blue Book to tell me what I can and can't do, so I'll start with the rollbar and go on from there.

Coming Together...

OK, the picture above is the central gubbins of the modal isolation suspension system. You have a central block which carries a normal coil spring on one side and a coil-over unit on the other. For the purposes of mock-up, I've modelled a Sachs Race Engineering damper, although at the eye-watering costs of these units, I can't imagine I'd ever actually use one. This central unit slides left-to-right on linear bearings (You could probably get away with plain bushes, but I'd want as little stiction in my suspension linkages as I can get away with, hence the posh guide bearings) when the car rolls. The big bearings at the bottom of the picture are to support a bell-crank that will rotate around the lower guide rail to give displacement control in bump. At the moment these are standard roller bearings, although it'll make more sense for them to be tapered to allow for the axial loads as the roll displacement works.

The only outstanding issue is to work out the sort of springs rates needed for this unit and to find someone prepared to do a 1:1 bump:rebound damper. Because of the mounting angle, we need a mono-shock damper and it's pretty rare to get a 1:1 ration for rebound forces as you normally want softer bump (which controls the wheel mass relative to the body) than rebound (which controls the body relative to the wheel mass). If it proves to be completely unfeasable (unlikely), there's nothing stopping you using two coil-over units mounted back to back across the central bar.

Update
I've put the bellcrank (I haven't put the bolt holes for location the two pull/push rods yet) in position with a second coil-over unit to show the heave spring mechanism as well. I was originally planning to attach the heave unit to the body directly and put some sort of guide bush in to deal with the side-to-side movement, but it seems possible to locate it onto the same central bracket that holds the roll coil-over. The only issue is that you want the guide rails then to be as short as possible to manage the bending loads they'll be undergoing as a result. If the loading all looks a little severe then I can still attach the heave coil-over to the body, I'll just have to tweak the mountings so there's sufficient articulation in the bushes.

Time to hit the calculator and work out if the attractive looking method will work...

Monoshocks and modal isolation...

Running on from yesterday's slightly cryptic post about modal isolation and the joy of being able to get exactly the right amount of damping in particular modes (getting the right amount of spring stiffness is usually easy). I thought I'd better elaborate. The picture above is from a Force hillclimb chassis (you can tell it's a hillclimber by the absence of just about everything in the search for stuff-all weight) and is about the best photograph I could find of a monoshock suspension. You can see that the two pushrods operate a single bellcrank. When both pushrods are displaced upwards, the crank rotates and moves the coilover unit in bump. When there is differential movement, the crank displaces along its 'axle' against the action of stacks of Belleville washers.

The design shown has two major design issues. There's very little roll movement possible and that movement is undamped. This isn't an issue in a wing-equipped car because you actually want very little roll else your expensive carbon fibre is going to be none-to-gently abrading itself on the tarmac. The absence of damping is an issue. At the end of the day these cars are going to be driven by amateur racers (gifted or not) and it's worth trading a few tenths of a second for driver confidence in the car. A car vibrating around in roll does not normally give a driver confidence.

So, how can we get around these two problems. Firstly we can replace the Belleville stack with normal helical coil springs. With more movement we can get more roll at less loading. With more lateral movement comes more side loading of the coilover unit which is bad, but this is easily curable by using a wider yoke rather than a clevis connect the coilover to the bellcrank. The only way to introduce damping is to fit a damper unit for the lateral movement. I'm thinking of having a Y-shaped bellcrank with a secondary inner arm. This inner arm would be static relative to the rotation of the bellcrank and move linearly with the roll movement. You can then fit a damper to the inner arm, connect it to the body and hey presto, properly damped, modally isolated suspension. and with little weight penalty over conventional two coilovers plus anti-roll bar.

The only other thing I'm considering is to put all this gubbins on the floor of the chassis and operate it by pullrods rather than push rods. A saving in centre of gravity height. In fact the only downside I can see is that corner weighting will be slightly harder as you'll have to preload one side of the roll springs relative to the other to transfer the weight. You do however get very easy ride height adjustment.

So I have a mental picture of how it'll all go together (I don't do visuals except in CAD becuase I'm quite possibly the World's worst sketch artist. I've got to sketch up a couple of dampers and some linear shaft bearings (well I say sketch, when I really mean drag them off of 3dContentCentral - thank you OnDrive) and then I can do a concept model for your general delectation.

Modal Isolation

Sorry for the absence of a post yesterday, but I was flat out on the CAD preparing for my evening class. They're going to be making a pipe vice, a mill stop and a plumb bob (or at least parts of them) as assessed pieces, so they all have to be modelled in CAD and then dimensioned up as 2D drawings before being cross referenced to the specifications of the qualification units. Anyway, here's a couple of quick renders of the two main items:
But getting back on with the car, I've been thinking long and hard about modal isolation, mainly because this month's issue of Racecar Engineering has a review of the Formula SAE/Student/FISITA World Cup events - annual competitions for university undergraduates to build their own deathtraps 600cc racing cars. Back when I was just a sprog in ride and handling terms, I worked closely with a certain British sports car manufacturer on fitting an active suspension system to a tank. The system worked on the principle of modal isolation. In other words instead of each wheel station having a spring and damper rate associated with it the overal vehicle was considered to have four specific spring and damper rates: roll, pitch, heave and warp. Warp was effectively torsion of the chassis, but that's probably better known to racecar engineers as either Roll Moment Distribution (if you read Milliken & Milliken) or Magic Number (If you've been trained by Claude Rouelle). Sitting in a car with a laptop you could change any of these rates and make either a dream handling vehicle or one that would disappear off into the undergrowth at the prod of a key. In fact if you had a half-decent egotist test driver you'd prod the key mid corner while trying desperately not to laugh as you sailed off into the undergrowth.


So, where does this come in for us - I'm hardly likely to specify an active ride system (especially as they cost around £500k - Moog servovalves tend to be around 10K each and you need at least eight per car). Well, on most cars the spring and damper units have to do all the work for all four modes and as a result you tend to have a bit of a pigs breakfast when it comes to getting an optimum solution, especially in damping of modes other than wheel heave. If you could have separate springs and dampers for everything, you could make large gains in handling without compromise.


Such systems are fairly popular on racecars - F3 have been using monoshock systems for over a decade, although these don't have any roll damping whatsoever. I'd be looking for a system that replaces the two coilover units and an antiroll bar with a pair of coilover units and a set of linkages and bellcranks. I've got a couple of ideas that I need to get modelled up so I can check the kinematics with Solidworks. I would ordinarily start designing my own rotary roll damper but I reckon that would be overkill so an extra bellcrank might be the order of the day. The basic idea is to have no ride springs and a T-type anti-roll bar connected between a pair of bellcranks operated by pushrods (or pullrods, for the benefit of my other reader). A coilover unit would be connected between the chassis and the centre of the T-type bar and thus be operated in heave. A second damper would be connected between a floating frame on the anti roll bar and the bar itself to give separate roll damping. It's the latter part that's causing some grief at the moment, mainly because the whole application lends itself wonderfully to rotary damping but chucks up a load of interesting kinematics problems when it comes to linear dampers.

Oh well, time to get the back of the fag packet out and start sketching...

Something other than cars...

I'm running an evening class for wannabe machinists, starting in September. One of the would-be master craftsmen asked a question about Stirling engines, which then segued into the little beast shown above (click on the picture for a larger view). For those who know their Stirling engines, it's a 'Ringbom' type that uses air pressure to move air between the hot and cold sides of the displacer rather than running the two pistons off a common crank arrangement.

Actually, it's probably a little complex to use for assessment pieces (not least because it's only around two inches high) - the qualification is at 'level 2' which is sort of equivalent to GCSE. That said, I'm hoping there'll be plenty of time available during the year and we can build one anyway. If we do, I'll re-engineer it (lets just say the person who designed it doesn't expect anyone trying to make it to have access to CNC machinery) for quicker manufacture.

Now back on with the car. I'm debating whether to install the spherical rod ends for the uprights in a pseudo-vertical or a pseudo-horizontal plane (by which I mean the axis of the hole). If I install them vertically, I'll have no issues with steering lock, but I might have articulation issues under bump. For a horizontal installation, it's vice-versa. I've been using the Aurora Bearings catalogue to try and make my choice and their rod ends would have around 35 degrees of articulation. If you look at yesterday's picture, you'll see that the joints are already relatively inclined, so I might have issues with wheel travel. I doubt that I'd need to use 35 degrees of articulation with turning. The tightest bend I know (the hairpin at Mallory) needs around 10 degrees of steering lock at the wheels to negotiate for a neutral steered car, which leaves plenty in reserve. I'll do some calculations and let you know the results.

Back to Geometry

With the information garnered from the upright model, I've been able to return to SusProg and try and finalise the suspension geometry, so that I can then engineer the chassis hardpoints and then get on with the design of the chassis itself. The picture above shows the (theoretical) final positions for the suspension arms, together with the best location for the steering rack to give minimal bump steer. For those who want a feel for the scale of the drawing, the thicker grid lines are 200mm apart and you're looking rearwards from the front centreline of the car.

I've rolled and bumped the suspension and got some feel for the geometry. As expected, there's very little change in roll centre height during roll, something of the order of 0.08 degrees inward for both wheels at 2 degrees of body roll. As I've said earlier, the net result of putting the instant centre on the centreline of the car is that you get a relatively shorter equivalent swing arm length (695mm in this case) compared to having less inclined suspension arms. According to SusProg, at 40 mm of bump travel, there's 3.5 degrees of negative camber. The roll centre doesn't move at all during roll, and is fixed at 132 mm above the ground.

Using SusProg to do all the hard maths for me has resulted in a total bump steer of 0.15mm toe-in at 100 mm bump, and around 10 microns over the range of +/-40mm wheel travel from static. One thing that I havn't considered up to now is the amount of Ackerman that the car will have. Ackerman is the method by which the inside wheel turns more than the outside when cornering. It's especially useful on circuits with tight corners, such as the hairpin at Mallory Park. At the moment the tie rod end is effectively outboard of the virtual kingpin, so the car has anti-Ackerman (277% for those who like meaningless numbers) and the outside wheel turns more. If I had sensible tyre data, I could actually work out if this was a bad thing or not. I don't and I doubt anyone will give me £200,000 to do some testing at CALSPAN, so I'll put it in the 'to be determined' pile for now. It's not an enormous issue, as I can do the same job on the steering arm as I did with the top rose joint mounting to offset it inward. In fact doing it that way would enable a range of Ackerman to be easily engineered and you could have a different geometry for Mallory than for Thruxton or Silverstone.

All done for now...

I think the upright is finished. One side effect of building the upright in two main parts, which I'd forgotten, is that there is no need to have two separate designs for nearside and offside. Those of you who look carefully at the finished design will spot the method for joining the two bits together. The aim is to mill a round-ended slot in the upright and have an identical mortice in the top extension. Ideally this will be a light interference fit (one of the joys of CNC is that you can do this sort of thing without the need for extensive grinding) which is further located by a pair of dowel pins for added strength. The cutaway below shows the joint a little better:



According to SolidWorks, that complete assembly weighs around 4 kg (and in reality probably less because the calliper has more volume than the real-life item. That's not too shabby, especially since the FEA results are looking fairly reasonable. I'm tempted to use a bit more mass and reinforce the upright slightly by reducing the size of the central hole in the purple bit. Now all that's together I can hit SusProg and get on with firming up the suspension geometry. After that should come the suspension hardpoints on the chassis and the chassis itself.

Back on track...

OK, 24 hours later and things are looking a bit better. I've changed the design for the upright by rotating the mounting holes through 45 degrees and then sending spurs off in the right direction for the various attachments. I've also come to a decision about manufacturing. While the idea of machining the whole thing out of solid billet has great appeal, the inward spur on the upright for the top wishbone mounting is going to be the thick end of 80-100mm inside the rest of the upright. This means that were this to be made out of billet around 85% of the billet would be converted into swarf. It seems to make a lot more sense to have a multi-part upright so that anyone with a milling machine can make the bits cheaply. I need to do some analysis on the loadings but I imagine that a properly designed joint is going to be no weaker than the billet method, and will be quicker and easier to manufacture, and cheaper to boot.

I decided to extend the method to the brake calliper mounting as well, as you can see in the complete assembled upright. This has the added side effect that other callipers can be easily accommodated for the unit without having to design and build new uprights every time. Now everything fits together and doesn't clash. I just need to do some FEA work to make sure the upright is meaty enough (the Mk I eyeball says it is, but I want to see what sort of safety margin I have when I've got 1.5g of cornering acceleration, a bit of downforce and 1.5g of braking applied too) and isn't going to fail like a US road bridge at the first sign of abuse.


Once I've satisfied myself with that, the I can get back to SusProg and finish the front suspension geometry off. I definitely want the zero camber change in roll, combined with zero bump steer (SusProg can calculate the position for the steering rack for me to achieve this)

Reasons not to be cheerful...

part one! Following on from the stress-free design of the front hub unit yesterday, I cracked on with the design of the upright that's going to glue all those miscellaneous parts to the rest of the virtual car. So, combining the basic dimensions I'd clubbed together from SusProg for my default upright, I set to work. The first, and most obvious problem was that I'd need to shift the lower mounting inboard by around 50mm so that it wouldn't foul on the brak disc. Susprog works on individual points and doesn't allow for such piffling considerations as the diameter of the rod ends that will actually be in use. Not a big issue, but I'll have to revisit the location of the upper mounting as well to keep the zero offset kingpin inclination I want. Still, it's only software and time rather than swearing in the garage 'cause you've mucked up another set of calculations.

So, next job was to make a mounting for the hub unit - nice and easy and there's a nice big hole in the middle for weight saving purposes. After giving it a nominal 25mm thickness, I did a quick and dirty check with CosmosWorks (the inbuilt FEA) package which reckoned on a safety margin of 8 with a 250kg load on the mounting bolts. I then added extensions for the lower mounting and the steering arm which gives the object shown above.

Final task of the day was to start building up a complete upright design so that I can be sure that I'm not doing anything silly with clearances. Yep, you've guessed it - I've manage to clash the steering arm with the brake calliper itself, and not by a small amount either:

So I have two choices:

1. Move the steering arm or calliper to the front to remove the clash - nice and easy, although I did flirt briefly with the idea of installing the calliper at the bottom of the wheel for c of g lowering purposes until I realise that I'd then be clashing with the lower mounting eye instead
2. Move the steering arm further inboard to avoid the clash - again easy to do in virtual space but a right bugger for anyone who actually wants to make the thing

I think 1 makes more sense and I'm more inclined to move the calliper than the steering arm because I want to leave the far front of the car free for a crushable structure and the radiator (if I can't get enough cooling using side ducts) rather than filling the space up with a steering rack.

Oh well, back to the coal face...