Projects '31 A FHC build (design?) thread

Extremely! especially as the Morris Minor will have priority, not to mention life intervening for the next year at least.

The thing has split into two streams of thought: one does away with the hydraulics entirely, and the other contains the entire springy part of the suspension system in a box on the RH running board, and is pretty much completely hydraulic. That involves a cylinder at each wheel, four spool cylinders, four slave cylinders, front and rear air springs, and left and right coil springs. I'm running with the other stream (if you'll forgive the mixed metaphor) right now, however, because I think it might end up being simpler.

 

Is Charlie Ware still making spares for those little Morris Minors in Indonesi ?
I saw his place one time out there, seemed to be little more than corrigated tin on a few scaff poles, and there they were building complete cars.

A school of thought on your FHC, if the track width is a least four feet, and the C of G only say a foot higher than the roll centre, even a 0.6 g turn only changes the load on each corner by a percentage compaired to the mass of the car.
So would be fairly easy to suspend the car completely on hydraulics with remote 'springs' and then valve the system to change the effective rate of springing 'seen' at each wheel station to control roll.
( Yes I know that sounds very Citroen, and I admit that it's only a stage on from the active control used on later Xantia's, but you know I am not adversed to the 'one bolt thinking' )

 

I am so sick of hearing this./..... If you dont like dont read.

 

Here's a complete rethink on the front suspension. I've done a little SketchUp model to illustrate the principle:

The axle is in two parts, so that the ends can rock forwards and back a bit. That's why it's blue and green. This is to allow the locating links to be non-parallel and of unequal length, as they would need to be if anti-dive geometry is used.

The red cradle is located longitudinally by the yellow links, and laterally by the short diagonal links with the big rod-ends at their lower/inner ends. I've shown the yellow links as arms because there are lots of ways to pick up springs from them, e.g. with different mechanical ratios or as bell-cranks, or some combination: I'm considering two options, either bell-cranks to a mechanical interlink or a hydraulic interlink system. If the spring mounts to the cradle pivot as shown here, the yellow links can be simple tubular bars. One could even run a transverse leaf between these points. There are all kinds of possible variations.

The short diagonal links define the roll centre, at the point where their axes cross, somewhere near the ground. I've plotted the roll centre's migration curve, and it doesn't walk very far, and only benignly towards the inside of the curve.

The intention is to effect different rates of mechanical advantage in bump/pitch and roll movements. In bump the cradle pivots about the two big rod ends, and the mechanical advantage at the springs is in the order of 3:1. In roll the cradle rotates about the abovementioned roll instant centre and, because the axle is laterally constrained to the cradle, the mechanical advantage is 1:1. The upshot of all this is that the suspension acts like the spring base is three times what it actually is: in fact wider than the front track; wider than the effective spring base afforded by ifs, which is necessarily equal to the track.

That means lots of effective roll stiffness without going into an uncomfortable range of natural frequency, and that without undamped anti-roll bars and everything that goes with them, like serious frame flex.

I get the impression that some people on here don't get what I'm trying to do. They would suggest that, if I want a traditional hot-rod, I ought to be willing to accept its limitations and not expect it to do what it can't do. If I want something else I ought rather to go for some product of the modern motor industry. But that doesn't work for me. My enthusiasm for old cars derives not from nostalgia, for that would be a longing for a past that is not my own; but a conviction that there is a sense, possible to define but hard to demonstrate empirically, in which old cars are better. Hence my striving to get the basic concepts of "obsolete" cars to outperform the basic concepts of modern cars.

Two aspects of old cars that interest me here are 1. that they are better suited to manufacture in a localized mom-and-pop economy, because of their body-on-frame construction and reduceability to small components, and 2. that they are less dependent on transport infrastructure and handle roads of marginal quality better. All this really asks for a suspension system that does not rely on a stressed-box structure to work properly. Then, I believe, it might be demonstrated that a car can ride like a Citroën DS and handle like a Porsche Cayman, at the same time, and yet does not need the resources of the global motor industry to make. That's what I want to show, but I'd like to enjoy it for myself, too.

 

This is great! Not that I fully understand the mechanical principles involved, but what you are doing is fantastic.

 

Thanks, James.

 

I'm lost. but fascinated.

I remember seeing a project with a solid tube front axle that was cut in half, then
put back to gether with a bearing between the two halves. (Same basic principle as
your axle design) I think he was using a more common 4-bar setup.
I'll try ad find it again.

Jeff

 

Here's a similar study around the rear suspension:


The differential offsets from lateral and longitudinal rotational axes is similar to the front suspension, but perhaps easier to see here. The two hydraulic cylinders are spaced far closer to the lateral axis than the longitudinal axis, and therefore different mechanical advantage obtains in the roll and bump components of any suspension movement.

Now, the theory here is that the vehicle is suspended on three virtual points and, as far as overturning stability is concerned, the vehicle acts like a 2F1R trike, moreover a trike with its centre of gravity in the right place. That is, the virtual suspension points are half-way between the rear wheels and about 67½% of the way from each rear wheel to the front wheel on the same side. Hence the additional air spring in the rear suspension: note that it counters no roll.

This means that the centre of gravity of the vehicle is likely to be at a third of the distance between the two side virtual support points and the single rear point.

Due to the differential mechanical advantage between roll and bump components of suspension movement at both ends of the vehicle, the virtual point of support does not necessarily coincide with the virtual centre of roll resistance on each side, but may be tuned at any other point along the wheelbase, from all at the front to all at the back. In practice these points will probably be about 55% from the back.

More, hopefully, a bit later ...

 

I got another part!

It cost me all of $101.61 at today's exchange rate, with badge, (damaged) bullnose, and stainless beading. These things are rare as hen's teeth in my part of the world. I'd been bargaining on having to import repro stuff from the USA, at about $550 plus taxes and shipping.

The plan is to mount a small, oval bicycle headlight in place of the badge, for which I'd have to modify the bullnose anyway. I can easily smooth it off where the bottom bit has broken off while I'm at it.

 

Always happy to see progress on this one.

 

I am trying to understand the point of this hydraulic roll resistance schematic.
It looks like you have designed a Hydraulic "split wishbone" and "ladder bar" system where both L & R move up and down together [ there is no roll compliance at all ,except the sprung chambers ] and maybe a ladder chassis will twist.
You will always have weight transfer during lateral accelaration, when the outside suspension compresses, the hydraulics will try to compress the inside the same amount [ she'll be real fun driving on 2 wheels, or the car will try to "compress down" ]

The best way to reduce "bodyroll" is to lower th CG closer to the Roll-Axis
This only reduces weight transfer through "overturning moment" or bodyroll , you will still get weight transfer at the tire contact patch.
[ even a Go-kart can tip over with no Roll-Axis ]
A good handling car has the Roll-Axis at ground level [which you have here ], the bodyroll is controlled via roll stiffness.[ which this doesn't have because it is solid via hydraulics]
If the CG is below the Roll-Axis, under high lateral accelaration ] the car will "Jack" because the lateral forces can exceed the vertical weight [ Pontiac Tempests, VW Beetles, Early Jags were examples of this
The Air Bags and rockers [posting #1] could easily be swapped for longitudinal torsion bars with a tension link between the bar anchor points [ it could be height adjustable ]
but this method still needs roll stiffness.
You can design all the fancy "Bell-cranks" "Rockers" "Pushrods" etc , but Geometry is Geometry, all the car cares about is suspension stiffness at the wheel [ antiroll, antisquat ,and antidive included ]

 

Hi Kerrynzl

I've moved away from the system in the illustration for the purposes of this project, but I still believe it is sound as such. Note that it is a roll-control system only and not a complete system of springing.

I am aware of the points you raise concerning weight transfer as determined by CG, roll axis, etc. as regards the vehicle as a whole: all you say is true. The important thing, however, given any value of weight transfer at any corresponding lateral acceleration, is the distribution of the weight transfer between the front and rear ends of the vehicle. That will determine how the vehicle handles, i.e. what the driver can do with the total amount of grip that is available under any given circumstances. Usually this is done by varying the relative roll stiffnesses at the two ends of the vehicle, a simple approach that is easy to understand and easy to implement but has the drawback of being only as effective as the vehicle structure is capable of resisting any torsions resulting from differential weight-transfer inputs.

I submit that a mechanism that allows equalization between front and rear weight-transfer inputs subject to certain parameters can maintain a set proportion of front-rear weight-transfer inputs - independently of fore-aft weight distribution - without affecting the torsional state of the vehicle structure. It's hardly new ground: the Citroën 2CV, BMC ADO16, and '55-'56 Packard all made use of this principle, but I don't believe the possibilities have yet been explored properly, especially in a high-performance context. I'm not quite sure why so few have got their heads around this stuff: the design psychology involved is worthy of a study on its own.

One point about jacking, though: in most of the worst cases the geometry is such that the conditions you describe are exascerbated by the onset of jacking. Thus when a swing-axle begins to tuck in under the car it further raises the roll centre, which intensifies the jacking effect, which raises the roll centre even more, etc. Hypothetically it should be possible for a high-roll-centre system to react to such conditions in a self-correcting way, which one would hardly describe as jacking. But there are other advantages to a low roll axis, most importantly it increases the effect of anything one does with the roll stiffness.

 

A little bit of progress. It did involve revisiting the Mumford Link front end with which I'd started before returning to the set-up most recently described. That was about interference between the steering drag links and the interconnect pullrods, leading initially to a configuration with a rack and pinion over the radiator and short, high drag links.

It's complex to explain, but the concept behind all this weirdness is to resolve the vertical movements of four wheels to be equivalent to the three wheels of a trike, because that configuration is inherently stable. That is, when placed on uneven ground a body on three supports is not subject to any flexing due to the unevenness of the ground. A body on four supports will either raise one of the four supports clear of the ground or bend so that all four touch the ground. We think of a car having four wheels for lots of very good reasons, simple packaging in plan, simple two-axle construction, and, most importantly, the facility to control handling characteristics by determining the ratio of load transfer at one axle to that at the other - in addition to an instinctive analogy to quadruped animals, though they are really tripod structures with a variously substitutable fourth leg to enable walking.

Tripod stability pertains to four-wheeled vehicles in that it allows the base-line for weight transfer to be set consistently and independently of road contours or frame flex. The idea is to resolve pairs of wheels to be equivalent to the respective wheels of a trike by arranging them in interconnected pairs. Initially (and, as it turns out, subsequently) the arrangement was set to be equivalent to a "reverse" trike, i.e. 2 front 1 rear, because that configuration makes for a better-handling trike, even though the lateral weight-transfer issues don't apply in the design as developed. The equivalencies are as follows:
LF + LR = Left
RF + RR = Right
LR + RR = Rear
It is possible to represent these three "virtual wheels" graphically on a plan view of the chassis: "rear" being half-way between the rear wheels and "left" and "right" being at a point between the respective wheels on either side, the position of which determined by the mechanical-advantage geometry involved. In this case "left" and "right" were placed 37.7" back from the front axle, i.e. 32½% of the wheelbase. This places the centre of gravity at the centre of the triangle described by the "virtual wheels".

A lot of the detail of linkages and locating devices here is about differentiating displacements in the bump and roll components of suspension movements, so although the car will always sit stably on any contour of road like a trike, the centre of lateral weight transfer does not necessarily coincide with the "virtual wheel" positions, but may be determined independently. I've initially got it at 52.2" back from the front axle, which is equivalent to a conventional chassis with a 55/45% distribution of roll stiffness.

I wished to reassure myself that the opposite arrangement (i.e. conventional 1 front 2 rear trike) would not be better in this situation, as it occurred to me that it would be advantageous in the case of a light truck, which will have a variable load on the back wheels. It should be possible to devise a mechanism by which the load bed may be levelled, and the "virtual wheels" moved back and forth, to ensure that the centre of gravity is always where it ought to be regardless of how heavily the truck is loaded. In principle it's a very simple system, as the same mechanical-advantage adjustment achieves both goals. In that arrangement the Mumford front end with its single spring held a distinct advantage. But in the end the other worked out better.

And, with apologies to some of you, I've ditched the hydraulics at long last. I don't know for sure, but there are enough question marks over seal durability for me rather to decide against hydraulics.

Here are some updated drawings:



The rear suspension, De Dion axle which rotates, in bump, about an axis close to the axle tube, which means that unsprung mass for bump motion is quite low, therefore comfortable ride. The "leading-arm" geometry should make for a solid - and self-correcting - hook. I stuck with the "Shockwaves", smaller ones though, not for "laying frame" but for self-levelling, which is something quite useful when one wants a soft ride and a low ride height at the same time.

 

The steering, with Foose-style dual drag links actuated by a firewall-mounted rack-and-pinion.



The unit I've got in mind is Alfa-Romeo Alfasud.

 

I've finally got to that, or rather got back to that, as I'd probably drawn this car fifty times. Here's the latest. The resemblance to my avatar should be obvious, as that was another iteration along the way.

The wheelbase has come down from 116" to 111", in response to more realistic tyre sizes and the decision to deviate from the stock A cowl length as much as I want, so I can place the A-pillar where it should be without that causing the wheelbase to be a function of the stock body length.

There's been significant progress on the frame and suspension, too, though I haven't finished drawing that up yet.

 

Inspiring ideas.

 

Having pretty much given up on hydraulics for this I've nevertheless been thinking about Citroën DS etc. hydraulics the past few days. Our upstairs neighbour in the new flat has a DS and his garage is by our front door. He had it out the other day, giving it a once-over after a 3000-mile tour of southern Africa. That got me to thinking about how they work. I never really understood why they worked that way, until now.

I was worried about how long hydraulic seals last: that the system would work as designed when the seals are new, and then for a long time not work quite as intended while fluid slowly leaks out; and finally the system would fail and the car would sit on its bump-stops. That isn't a concern with the Citroën system, as it isn't designed according to a hypothetical "new" condition wherein the seals don't leak: the seals are designed to leak. The entire system slowly leaks fluid, not onto the road but into a low-pressure recovery circuit, and the loss is constantly made up for by the main pressure pump. That's why DSs etc. slowly lie down when parked. That's also why the system is quite durable and reliable and diy-friendly once one understands it properly.

I'd not really thought of the Citroën system for two reasons. Firstly, interconnection was never really a feature of the system as applied to Citroën cars, though technically the wheels of each axle are interconnected side-to-side and the effect of that counteracted by a substantial anti-roll bar. That, especially the reliance on anti-roll bars, makes the system as it stands unsuitable for my purposes, which seek in interconnection freedom from torsional inputs onto what may consequently be a light and shallow ladder frame. However, a moment's reflection reveals that effective front-rear interconnection may very easily be introduced. The ease by which the hydropneumatic spheres may be separated from their respective suspension cylinders simplifies this considerably: mounting the spheres remotely is a mere matter of adapters.

Secondly, looking at hydraulic suspension for purposes other than "laying frame" I did not like the idea that the car would slowly settle onto its bump-stops when parked. But this, too, is soluble. All that is needed is a very soft spring acting in parallel to each suspension cylinder, of a rate that it will bear the car's weight at the desired ride height but no more. The car will drop when one gets in, or puts something significant in the trunk, but will return to the design ride attitude once the engine is running. Higher ride-height settings will obviously not be affected by the extra springs. Again, converting the suspension cylinders to a coil-over configuration should be quite simple.

I think it's the psychological thing that suggests that the Citroën suspension is best suited to a relatively low-powered front-drive car with the accent on a luxurious ride - though DS handling is not to be sneezed at. In fact the system was used on a number of cars whose layouts and performance envelopes suggest more commonality with hot rods and custom cars, e.g. Rolls-Royce and Mercedes-Benz (450SEL6.9). We'd easily lift the engine out of a '70s Shadow or V8 Mercedes for use in a hot rod; why not the Citroën suspension as well?

I need to do some head-scratching around how to make this work with my effective-spring-base manipulations, but it shouldn't be too hard to do so, nor with the virtual-tripod concept. I can see some simplified axle locations happening.

 

thats what my Grandad said in 1959 when we cut up a model A and put a Lincolin engine in it----we ruined it and it would never work

 

I was being ironic. It was aimed at the "this isn´t trad" comments, my point being that hydraulics and bell cranks are hardly new technology, by any stretch of the imagination, even if applied in an unusual way here.

 

This is turning out to be one sweet build. I just found this thread and am in love. No need to take anymore classes at college! I'll just study your work.

And the link in the signature has some pretty sweet stuff too!

Subscribed.


iPhone - TJJ App

 

Something that's been happening a lot on this project is sort of the theoretical equivalent of sitting down a short distance away to have a beer and a scratch of the head and a good long stare at the work. There were a few tangents leading up to my last posts in April.



Similar pondering around the Morris Minor led me to consider torsion bars, in particular a development of the "Torsion-Level" system used by Packard in 1955-6. It had struck me that my objection that the damping wasn't interconnected like the torsion bars could be resolved by running small hydraulic lines between the front and rear dampers. Doing that with the Minor's lever-arm dampers would be really cool, very much in the spirit of the project. I tried the same idea for the '31 with the torsion bars inside the frame rails, but in order to achieve the roll stiffness I wanted the bars would have had to be heavier than the frame rails It's fine on the Minor: the wheelbase is shorter and I can live with more roll, so 28mm bars would do the trick.

That was just as well, because that was about the time I started musing on Citroën hydropneumatics, which led to the following arrangement:



This was the first one on the 111" wheelbase. Note also the return to the Model A frame shape. Both those came out of trying to get the torsion bar idea to work a bit better. All this worked out quite well, and indeed allowed a lot of simplification: so much greater the irony that I should come across a locally developed interconnected hydropneumatic suspension system shortly afterwards.

These guys seem to have got the sealing problem licked (if not a Citroën-style leak-recovery system should not be too difficult to incorporate). I'm not going to go into a lot of detail at this stage, as I've promised the guys that I'd keep it largely under my hat for the moment, except to say that the inventor of the system had been thinking along the same lines as I had been for many years, and that he has made the breakthrough in his thinking that I'd been looking for all along. With any luck this is coming your way some time soon, and it's radical stuff.

Gone, consequently, is the virtual-three-point idea, and a return to the front-rear-left-right approach with which this thread started. This new development allowed me a simple and basically conventional four-corner installation. Thinking that through led to further simplifications.

Perhaps it's worthwhile to list my requirements on this project:

1. Complete free articulation in warp, so that the front-rear distribution of lateral weight transfer may be fine-tuned without thereby incurring a need for a torsionally-stiff vehicle structure. This is consistent with a requirement of about 10" of single-wheel bump displacement without the suspension bottoming.
2. Out of sheer pig-headedness I want to do this with solid axles.
3. Minimal unsprung mass, hence a DeDion axle at the back, moreover one that rotates in bump about an axis near the relatively heavy tube.
4. Roll at 0.5g limited to 1.00°-1.25°.
5. Low roll axis. This is mainly about widening the range of lateral weight transfer distribution tuneability. I'd been aiming for at least the front roll centre on the ground, but my thread about roll centres gave me some perspective. Roll centres up to 3" or so above the ground will do the job.
6. Sharp rack-and-pinion steering, and as close to an ideal steering wheel position as possible. Hence my pursuit of the idea to mount a rack-and-pinion fairly high on the firewall, without losing too much of the benefit to lost motion in bell cranks, idler links, etc.
7. A decently small turning circle.
8. Launch behaviour one would more expect from a drag car than from a road racer. Hence, optimized effective swing-arm geometry in the rear suspension, and enough front droop travel to assist rearward weight transfer effectively: front travel 3" up / 5" down; rear 3" up / 3" down.
9. What one might call copyability; that is, the ability to transfer solutions to other contexts. The benefit of all this should apply as much to, say, a work truck as to the rakish fixed-head coupé here under discussion.

Searching for simplicity led to this idea: the diagonal Panhard Bar, with diagonal cross-steer, coupled with four-bar location. It gives me my fairly low roll centre if the four-bars are angled as shown. The same angle gives effective anti-dive braking. The drawback is a little bit of Panhard waddle over large bumps.

Making the steering arm and the corresponding lower arm of the steering crank as long as practical in relation to the upper arm of the crank should limit lost motion to an absolute minimum. The rack-and-pinion is an Alfasud centre-steer unit: I've since been able to estimate the size of the thing fairly well, and it's a bit bigger than I thought. I also considered a VW Golf Mk1 end-steer rack, but the Alfasud worked out better.



The front axle is back to a simple tube as one might find on the front of a T-bucket, without rotating joints or anything like that. The suspension cylinders are quite long in this case, to give me my 5" of droop travel. They can be made much shorter but for that requirement.



The rear axle is the same two-axis-located DeDion tube, held by two stubby A-arms in such a way that the lateral axis of rotation is considerably higher than the longitudinal axis of rotation. That gives me my launch geometry together with my 2½"-high roll centre. The rear axle is considerably simplified by suspending it by its ends only: no more "ShockWave" in the middle. The suspension cylinders attach to the axle by Vernier plates, so that the lateral weight transfer distribution can be fine-tuned by adjusting the rear spring base width.

The control unit of the hydropneumatic suspension system occupies a box behind the right-hand front fender, the car battery occupying a corresponding box on the left. It is actually an extremely simple system, in essence entirely passive, no pumps, no electronics. There are convenient places for the front and rear hydropneumatic accumulators, behind the bottom-left corner of the radiator and between the rails behind the Jaguar diff.

If I had any cash I could go and buy the steel now. As it is it'll have to wait. At least I can start figuring out how to construct the bodywork ...

 

I think I may have to see this completed, before I can start to get to grips with understanding it. Do you have any 3d views of your ideas?

 

Dawie,
I'm pretty sure you've delved into this, but I'd be careful with that diagonal Panhard design. That setup is quite common on drag cars over here. It's called a track locator and only allows a very limited amount of roll in the 4 link. I know part of your design brief is freedom of motion, so I thought it was worth bringing up.

Let's say the panhard is installed from point A to point B...as the 4 bar suspension articulates (rolls), point A and point B want to be closer together, but the Panhard bar has rigidly triangulated the 2 points, and will not allow them to get closer or further apart. You've done enough of this that I'm hoping that explanation makes sense; and if you've taken that into account, I'll be glad to be proven wrong.

 

I like it. Even if I don't understand it or even want to apply any of these fancy-pants engineering principles, there are plenty of cool and clever designs and build ideas I can liberate from you, Ned-ster! Gary

 

I've been pondering this for a while, a problem i haven't come to grips with is that the front corner pressures will be greater than the rear corner pressures say a 60/40 split so how will compensation occur for a rear corner bump/rebound motion? i can't see how the diagonal front will react since it is already under greater pressure than the rear.
Lets look at an event: the right rear tire drops into a pot hole and drops 2", it is a solid axle and has 150lbs/in spring or resistance so what generates the pressure to support the vehicle with the left rear? it needs 300lbs to support the car? The diagonal front corner (left) rises by a proportionate amount that the right rear drops.
Now the right rear hits the leading edge of the pot hole and rebounds 2" up into the wheel well, that generates 300lbs pressure that diagonally transfers to the left front (proportionally) that the left front will have to counter act and what happens to the left rear? will it rise in sympathy will the right? it cannot react and go down since it is on the ground.
I could make more sense out of this system if it were powered, something other than the vehicles own weight were creating pressure, but it is a passive system and unless you can get all 4 corners of the vehicle to weight the same i don't see how it will work. But then, back in the day, i thought a video rental store was the dumbest idea i ever heard of and a shop that just sells a cup of coffee? So the candles on my cake are sputtering a little.
Thanks Ned, oj

 

Ray, you raise an interesting point. I wasn't quite sure myself, so I built a quick cardboard-and-string model to test my logic. I think the enabling factor here is that the four-bars are of the same length and initially parallel to one another in all planes. As each of the four-bars therefore describes the surface of a sphere, the axle assembly can adopt a range of positions without any change to its orientation, as long as no roll is involved. At any position within that range the four-bars will be parallel to one another in all planes, and from any position within that range roll motion is possible by the bars ceasing to be parallel. Therefore the range of possible positions the axle end of the Panhard can assume is large and will probably describe the annular space between two non-concentric spheres.

Now, the Panhard bar likewise describes the surface of a sphere, and if the intersection of that with the above annular zone is something other than a single point, motion is possible; and if it is something other than a circle, roll is possible. It would appear that the intersection might be something like a part-spherical ring or piece thereof, like the skin of a slice of tomato. In particular, the axle end of the Panhard bar will describe a different arc with roll motion for every given overall ride height; conversely it will describe a different arc with vertical motion for every given angle of roll.

Had the four-bars been parallel to the ground (i.e. parallel to the Panhard bar in side elevation) these arcs would have been predominantly vertical. As it is there will be a tiny little leftward scrub over large bumps; the price of my roll centre location.

 

Think of it this way. Imagine a vehicle like this:



Any corner input is resisted partly by the spring on the axle and partly by the spring on the side. In roll, only the side springs come into play, and the distribution of lateral weight transfer between the front and rear will depend on the positions of the side springs, irrespective of the weight distribution of the vehicle itself. The side springs do help to carry the vehicle's weight, though, so roll stiffness will have a bearing on ride quality.



Here is a further elaboration: the side springs have been replaced with an anti-roll bar. Now roll stiffness is independent of ride quality - except as regards asymmetric bumps, to some extent. Lateral weight transfer distribution depends entirely on where the ends of the anti-roll bar attach.

 

After searching in vain for several days for free software that could model the mechanism accurately, it occurred to me that a close approximation could be had with no more than a few simple Pythagorean functions. I crunched, therefore, a number or two.

To achieve the maximum single-wheel travel I want, axle angular displacement of a tad over 6° is required. As this corresponds with a backward/upward axle displacement of a mere .040", we can say that, in roll, the diagonal Panhard bar will behave for all intents and purposes like a conventional Panhard bar. The only difference would be a ±.040" lateral shift at extreme axle angles.

This lateral shift is much more pronounced during simple bump and droop motions: the abovementioned "Panhard waddle". The axle is shifted about 0.4" to the left at maximum bump and about 1.2" to the right at maximum droop. This is, however, a relatively long-travel suspension system, and it is anticipated that most driving would happen at around 1" either side of normal ride height, over which range the lateral shift is less than 0.4". Expressed differently, around ride height the movement of the axle with bump/droop is around 10° off vertical. Moreover, as a large portion of typical bumps are single wheel bumps and therefore involve an angular component, the typical real-world range of lateral shift is likely to be closer to ¼".

Now, I could get all quantum and claim that this is all to do with involving time warps in the front end behaviour during a good, solid launch: "It's just a jump to the left ..." Seriously, it's not really that big a jump.

 

Here's a 3D model of the rear suspension:

Also posted on the DeDion axle thread.

 

Thanks Dawie. I´m still having trouble seeing how that will allow the car to rock side to side, as it seems to be held solid on the two triangular plates near the centre line. I dare say i´m missing something blindingly obvious though!