AW11 Setup

Hello everyone,

I'm attempting to build up an aw11, however the most common autox setup for these cars is stiff enough to warrant trailering it to an event rather than drive itself (6hz front 4hz rear). This seems very dependent on the rules of your class, aswell as a smooth track. So with minor considerations taken about setup not forcing me into a higher class(no rca's, suspension pickups, large tire size changes, or engine swaps) I'll avoid some easy solutions. If they legitimately fix an issue I can't solve elsewhere(or I blow my 230k mile engine) I have no issues doing them.

Anyway, lets avoid built to rulebook for now, and look at some parts. https://imgur.com/a/9B9gDok

I'm lucky enough that my 32 year old car has camber, caster, and toe adjustment from the factory, however the factory range of adjustment is somewhat limited. About 1.5 degrees of caster and camber adjustment. While it's enough to potentially avoid needing adjustable tophats, 32 years takes a toll on anything rubber. Also, for strut suspension adjustable tophats include a very small amount of kingpin, or steering axis inclination adjustment.

However, these adjustments do absolutely nothing if we don't know what settings we want. The inherent nature of the car due to rear weight bias, is to have a "floaty" front end. Initial Caster and KPI settings were simply to max them, 10 degrees caster and 12 degrees KPI(KPI is an almost nothing adjustment for self centering unless at large steering angles, it's maxed to place me in a more aggressive camber gain portion of the curve with regards to bump). For the wheels to turn, they rotate onto the edge of the tire, lifting the entire car up slightly. This causes the weight of the car to want to return them to straight ahead (Also mechanical/pnumatic trail). I have a manual steering rack so the steering got notably heavier, but not so much to be problematic.

https://imgur.com/a/UTYNptP

Here's the method I used to adjust caster, it's only applicable to strut suspension, not overly accurate, but can be done with things I already had in the garage. I have a real caster gauge now, I was only .2 degrees off with the strings(but still consistent left to right of the car). If you can afford it, a quality camber caster gauge makes your life significantly easier. Anyway, I placed 2 21mm sockets on the strut nuts, and a straight edge across body. This gives me a straight line 1 inch above, and .5 inches behind the actual strut pivot(for my car). The center of my wheel is 2 inches above my ball joint. So I'm measuring fixed points that aren't actually at my pivots, but dependent on them for their location. Measure the distance(or angle if you have the capabilities) between the 2 strings, and match them on both sides of the car. Throw it in any cad program, or even google triangle calculator, adjust for your offsets, and you have caster.

I'm intentionally avoiding discussing springs here, that'll be a long post if I go over it, however changing my spring rate, changes how much body roll happens. Body roll is what forces the suspension to compress or extend while cornering inducing camber change. Additionally, the increase in both caster and king pin inclination will change my camber with steering angle. So it becomes necessary to test my camber(and tire pressures) before I can set this. I set it to -1 degree as a baseline before testing.

https://imgur.com/a/r9DpBh6

I'm not cool enough to own a pyrometer and take tire core temps, mounting an IR camera aimed at the wheel would have been a good idea, but I just used an IR temperature gauge, and took 3 measurements across the tire. From there adjust camber and tire pressure to equally heat the tire while cornering. Trying to effectively collect as much data as possible, I setup an arduino, accelerometer, and LCD display to show yaw and lateral force. Pitch, and braking/acceleration longitudinal forces would be sent to a laptop, but that bounced around too much affecting the readings, so they weren't logged.

Programming the arduino I had one additional realization, physics and the reality of the car differ. Because the my accelerometer has no input from the car(Steering angle or speed) It has to calculate everything from forces acting upon it, and the car. For a set radius, at a set speed, the car will be under a set amount of lateral force. Ultimately, The steering wheels angle(and thus the radius), and my speedometer define out how much lateral force the car is seeing. The arduino is calculating this. It sees the centrifugal force and the yaw rate. Ac*Pi^2/Yaw^2=R | Lateral G's * Pi Squared / Yaw(rad/s) Squared = Radius, yaw*radius = speed. In testing I have to turn the wheel more than is theoretically necessary here, travel at a slower speed, or a larger radius. The difference between physics and my actual car? Slip angle! I can confirm the steady state load on the tires defining my steady state over/under/neutral steer bias in the measured slip.

I'll update this thread with some of the issues in the cars design that may need addressing, and possibly my current setup for springs and roll stiffness, and why it was chosen in a later post, this is long enough to cover half an alignment. Nothing interesting happened with toe so I left it out.

Testing the car and trying to measure slip angle proved to be exceedingly difficult. Any change in throttle, steering angle, or even bumps in the track would register on the arduino, and take time to settle. I expected to see an increase in radius as speed increases due to the slip angle in the tires, for neutral steer however it shouldn't require any change in steering angle. Between my general feeling, the constant need for an increase in steering angle as speed increased, and averaging multiple laps on the arduino it became quite apparent that my car was understeering nearly the entire time.

So I disconnect the front sway bar. My front roll resistance is 830 ft/lb per degree of body roll. The swaybar is 351 ft/lb/deg of that. Issue resolved... sort of. The car now oversteered up to 1g lateral force(I didn't test beyond that). The spring rate had been chosen to progressively shift weight forward as lateral force increases. At 1.3g lateral it should be neutral steer(equal slip angle front and rear) and understeer beyond that, I would have been happy with anywhere after 1g being neutral steer. Something was increasing the stiffness of my front end beyond what the springs and roll bar were, but not by a large margin, turns out it's the bushings.

https://imgur.com/a/chusQjJ

This is my front control arm setup, and the tension control arm is what replaced the original bushing piece. The control arm will pivot around the blue axis. A virtual line between both connections of the A-Arm. With a rod end this isn't an issue, it will rotate freely even if it is off axis. The bushings on the other hand were forced to deflect to allow any suspension movement. It's positioned exactly how it was connected to the car through a hole in the subframe. The next picture is the permanent deformation of the polyurethane bushings. If it's permanently deforming bushings there won't even be consistency in the amount of force it takes. Obviously, I switched to an after market rod and removed the issue.

The same general principle is on the rear, however in a far less problematic orientation.

https://imgur.com/a/kRDYA8h

Here I have a 3rd bushing to help absorb the off axis bushings, aswell as a far less problematic layout. The rear bushings bind significantly less than the front of the car. If anyone knows how to calculate the spring rate of bushings, or why that's not possible please let me know. For full racecars, bushing deflection is enough to switch to rod ends, let alone an unknown "spring rate". Out of curiosity I took a luggage scale to the control arm and pulled up. It took 50 lbs for about an inch and a half of travel on the front. The rear is around 20 for the same distance. For 200 in/lb and 220 in/lb front and rear springs, the front makes up a large portion, the rear less so. A stiffer car would lessen the impact of these bushings on the spring rate.

Swapping the rear control arms out for a set with rod ends seems likely on my future upgrade list, but I'm still reconciling a few things. Under lateral loads, the bushing will compress and give me toe in(on the outside wheel), toe out on the outside, a tiny bit of roll understeer also exists on the car. All of these aid in stability under almost any situation. Removing the bushings for rod ends gets rid of 1 of these(I suppose compliance still exists in rod ends, but its nothing compared to bushings). Replacing them may require additional rear toe in(and thus drag, tire wear, and heat) for the same stability. With them going to same direction, I don't expect many issues here, but if I change wheels I'll need to pay closer attention to the offset.

https://imgur.com/a/zTrzfBz

This is an oddity of the car. I'm not really sure why the LCA chassis mount slants upwards. Anti squat is a consideration, but that depends more on the trailing rod's height. It could just be to keep the bushings in tension by always having the LCA try and pull backwards, while the trailing rod pulls it forward. Or it's just be to remove the bolt without colliding with the tie rod. I don't know of any issues it's causing, but it seems odd to me.

Hi Robbie,

This is an awesome thread you've created, so much detail and great information! This was the type of thread we hope to see more of on this forum, people can also easily apply this to there build without any expensive alignment tools.

It's clear to see you have a real understanding of alignment and geometry. I've also seen you give great answers to peoples questions here on the forum, thanks for that!

Keep us updated on this project, I'm interested to see how it develops.

-Brandon

Thansks Brandon, hopefully it sends people in the correct direction to research further. If nothing else I catch mistakes or estimations I had used when typing these up to correct for. I had a broken woodruff key and completely forgot to update this; Anyway, I've hinted at overall roll stiffness enough, I think it's time to explain it, Pictures might be a challenge for this as it's all imaginary lines that represent your suspension arms to simplify calculations.

So, looking at a car, unless this is an insanely high end f1, mclaren, koenigsegg etc with roll specific springs and dampers, we don't have roll as an individual parameter. The springs that control ride, aswell as the anti roll bars are our source of roll stiffness. I'm sorry in advance for the mess of my calculations as I did most of them using radians and ride frequency for easier comparison of cars.

So to convert to roll stiffness the basic equation is - Roll Stiffness Front = Front wheel rate * Track Width ^2 / 1375. KΦsf=Krf*Tf^2/1375 | Track width is in inches, 1375 is a conversion. For my car, 199*57.4^2/1375 = 476 front rolls stiffness due to springs. 199 in/lbs is my wheel rate(its really 200), this is after the motion ratio from the wheel, not the literal spring rate.

We're still missing a link in the chain however, the chassis. Roll center height explanations are just about guaranteed to be incorrect without an accompanying textbook, so I'm just going to use the simplest, and then why I don't like it. Roll center height is the leverage Lateral Acceleration has on your suspension. The distance from your center of gravity height to the roll center height. An amount of force goes in, and a leverage, ft/lbs, comes out. This is your roll Roll moment. MΦ/Ay=hRM*Ws/12 | Roll Moment / Lateral Acceleration = Roll moment leverage(distance from cg to roll center) * Sprung Weight / 12

2412 = 15.8*1930/12 Ft/lbs. If I choose 2 degrees of body roll as my target, 1206 ft/lbs needs to be my total roll stiffness.

With those 2 out of the way, we can either change our springs here, or correct for the current ones. Front springs are 476 ft/lb roll stiffness, rears are 528. I need another 202 ft/lb of roll stiffness to hit my 1206 for 2 degrees of body roll. Alternatively I could also change my ride height(and thus roll center and cg height) to stiffen the car without changing anything. Seeing as I still have a stock front anti roll bar, I'll just use it. The Anti Roll bar is 230 ft/lbs, which if I didn't round all my numbers would be correct, so we'll pretend it's 202. This makes my front roll stiffness 678 ft/lb and rear at 528 ft/lb.

It's important to remember this is only the stiffness, after the leverage has been applied, we can't quite translate it to under or oversteer yet. What we're focused on this far is the distribution of roll stiffness, 678:528 | 0.56:0.44 | Next we add in the actual lateral load transfer. Weight * CG Height / Track width. simple enough. 2350*19.1/57.4 = 781 lbs/g (pounds per g of lateral force). This gets distributed based on our roll stiffness. 343 Lbs go to the rear, 437 go to the front. Finally(at long last) We take the static weight of the front of the car, 1100/2 + 437 = 987 lb, 1200/2 + 343 = 943.

Finally we have the outside wheel loads. Rounding all my numbers appears to have made my car understeer heavy. As lateral acceleration increase, we follow the same trend upwards and load more and more to the front wheel and we've hit near equal at 1g. Ideally I'd want to target neutral steer(equal loading) at higher lateral than just 1g. However this is a good setup still. Understeering, in steady state conditions beyond 1g lateral should coincide with the apex and corner exit. Having additional load on the front leaves more available for the rear as lateral load reduces as we exit the turn, and allows the driver to shift it rearward by accelerating, it's just happening too early. This is my already modified setup, stock its similar, but the crossover happens around .4g, which might explain a few memes about this car.

https://i.imgur.com/m78WxC1.png

When I graph the results you can easily see the switchover(Red is understeer, green oversteer) Where blue and yellow represent the amount of load being taken by the front and rear, and how it gradually climbs as lateral acceleration increases. So the end goal of the whole setup is to push the crossover to approximately my max cornering capabilities, and tune out the entry and exit transients without severely impacting the steady state conditions. (I'm running too stiff of dampers in the front to increase the load transfer forward while lateral acceleration builds). Taking it further would involve geometric changes, likely longer rear axles control arms and track width to get the roll centers changing at different rates, so I'm not settling for oversteer entry understeer out, and can setup for fairly linear handling, however obviously involves more expensive changes like custom axles or knuckles.

Anyway, I'm not a fan of this description personally. As always, there's a deeper rabbit hole you can dive into, that involves separating roll couple(roll center height front and rear combined into a line), into front and rear. Then you can plot that dynamically as the suspension goes through its travel. For newer/nicer cars with different suspension configurations front and rear, multilink rear and macpherson front(lookin at you porsche) as an example, you can have drastically different roll center height changes with body roll. In my case the front and rear are near mirrors, so I can simplify without any major consequences. You can go another step further and seperate inside and outside tire's roll center height from eachother, as the inside tire generates far less lateral force(grip) than the outside and treating roll center as anti-roll(anti dive or squat but 90 degrees off). However the only references to this I can find involve computer simulation and tire data, both of which are outside my capabilities.

https://i.imgur.com/N2Gy4VA.png

Disclaimer - I had to pull most of my numbers from my actual calculations that are done in ride frequency, radians, and has compensation for cambered roads. It's incredibly likely I forgot unit conversions and motion rations as they were done elsewhere on my actual notes. It's worth noting that while I didn't explain it, roll centers should be near ground level, as you go further away from the ground you end up with jacking effects.

Seems like I have a thing for bushings, as my next project is to replace a few engine mounts. Hopefully some of the engine tuners might have estimates or theories regarding this. 2 of the 4 mounts on my car I can get, however for the trans, and timing belt side mounts, I would have to order them from the EU. They're also rather expensive for what is ultimately a brick of polyurethane.

So I'm going to 3d print them instead. I know this part has been done before, however I believe he simply printed a solid part without much thought into the bushing itself. So, as is tradition for me, I wildly overthought the possibilities of what I can do with them, and half expect them to be minor influences at best. In the interest of saving filament, I printed some little guys to test, and destroy. Forgive lack of any real accuracy in testing, as I don't have the equipment to do so. However, all of the tests were the same, regardless of the unknown weights/springs/leverage.

https://imgur.com/a/F4iZefH

So, clearly I wasn't overthinking for no reason. There is some notable differences between parts. The air spring idea at the top was effectively worthless as a test. By forcing all the air cylinders into the part, the outline surrounding them simply made the part stronger than most of the intentionally weaker comparisons. Poking a hole into each chamber produced near identical results. I have very little knowledge of gas laws, but pv=nrt is fairly straight forward. If I make the cylinders larger, the pressure change for the same movement will be smaller, resulting in a less stiff air spring, however would allow for a weaker part, possibly making it noticeable on a test? For now I think I've scrapped the idea of a built in air spring. The increasing spring rate characteristics seemed useful, but I can do that with the geometry of the X looking patterns. Spring rate will increase with displacement as their angle changes. I didn't cut them in half to show the inside as the plan is to squish the life out of them when I figure out a way to measure it, so I used the slicer images to show the internal patterns.

Onto the part where I hope one of the engine guys might provide some useful insight. Is there a benefit to setting it directional in any orientation? I can effectively mirror the oem bushing, in that it has very soft movement for the actual vibrations of the engine. My best estimation is that each firing of the cylinder produces a torque, so 1/2 the RPM*Number of cylinders/60 should be about the frequency of vibration? 6000/2*4/60, so 200 hz for a 4cyl at 6k?

The secondary consideration, and probably the more important one, is the initial torque when releasing the clutch. If the crankshaft cannot spin for some reason, the only alternative is for the entire engine to rotate around it instead. So there would be an initial surge of force into the bushings while it overcomes the inertia of the drivetrain and wheels? Following this logic, would it benefit me to stiffen the bushings in the direction of crankshaft rotation to restrain the engine during launches, while leaving the opposite side softer to reduce vibrations on the bottom half of the sinusoidal motion?

I may try this with suspension bushes depending on how it works out, but thats a very high force application, so I want a way to put 1000 lbs of force into these before I'd feel brave enough to do so.

If anyone has suggestions they cost a whopping 10c of material for the little test models, so I'll try just about anything out of pure curiosity. For reference, https://i.imgur.com/KqTd2RE.png this part squeezing from the sides is about as stiff as a marshmellow.