RR26 Steering

The steering project was my introduction to leading a system design from packaging constraints all the way through to kinematic justification, and it taught me just how interconnected those decisions really are.

 

Packaging came first, and it was less "design freely" than "find the only slot that works." Between the T.1.2 internal cockpit template, the T.1.1 cockpit opening template, the front hoop, the front hoop bracing, the steering column drop angle, and the desired Ackermann behavior, the rack ended up with essentially one viable location: as low as possible to help with T.1.2 compliance and to cut down the bending moment on the rack mounting tabs, and forward of the front roll hoop because of the column drop needed to reach the steering wheel collet. Working through that made it clear that packaging isn't a constraint you satisfy after the fact, it's a constraint that actively shapes the geometry you're allowed to pursue.

 

The Ackermann analysis was the core of the project. Using a bike model and predicted weight transfer, I worked out target steer angles for each wheel across a range of corner radii, then focused on optimizing for 25, 55, and 70 ft radii based on our autocross layout. The bike model showed that because our tires have low load sensitivity in slip angle and the car doesn't generate much load transfer, peak lateral acceleration actually comes from the tires being parallel to each other, not from classic Ackermann steer. That was a genuinely useful reframe: parallel steer doesn't mean parallel wheels, because static toe (2 degrees toe-out on this car) creates a 4-degree heading difference between the wheels that ideally should be closer to zero in steady-state cornering. So the real objective became closing that 4-degree gap, which meant designing in anti-Ackermann rather than the pro-Ackermann most people default to.

 

Getting there meant deciding between a few levers: moving the rack further forward, moving the tie rod pickups on the uprights further inboard, or increasing clevis-to-clevis distance with a larger rack to increase the angular gain of the tie rods through steer. I modeled this with a 3D kinematic sketch in SolidWorks integrated with the existing suspension geometry, which worked, but in hindsight was more resolution than the problem needed. Because the upright stock was already purchased and the front roll hoop brace fixed the rack's envelope, a 2D sketch would have gotten me to the same answer faster. That's a lesson I've since carried forward: I built a Python tool to run Ackermann kinematic sweeps, which lets me iterate through pickup point and rack geometry options far faster than rebuilding a 3D SolidWorks sketch every time, and reserve the higher-fidelity 3D modeling for when I actually need to check something like bump steer interactions rather than pure steer-angle Ackermann.

 

The final geometry ended up as far toward the anti-Ackermann target as the packaging allowed but not all the way there: the tie rod pickups on the uprights as far inboard and forward as the stock uprights would permit, the rack pushed as far forward as possible, and a 17" Sensata/Namco rack chosen over the 14" option specifically to increase the tie rods' angular gain through steer. Bump steer was handled with a simpler geometric method, drawing lines from the tie rod pickup through the rack to the instant center and setting pickup height at that intersection, which isn't a complete bump steer solution but is a reasonable time-efficient approximation, and it landed at 3 degrees of bump steer.

 

Overall, the project reinforced that steering geometry is really an exercise in constraint satisfaction under uncertainty, not an idealized optimization. You rarely get your literal target numbers; you get the best compromise the packaging and prior sourcing decisions allow, and the value is in understanding why you ended up there well enough to defend it and to know where the next design iteration should push.