Validate the wheel path before the car moves
Generated from
content/lms/suspension-and-chassis-design/03-compliance-and-kinematic-validation/01-kinematics-rig-testing.md; edit the source file, not this page.
Source path: content/lms/suspension-and-chassis-design/03-compliance-and-kinematic-validation/01-kinematics-rig-testing.md
Course: Design suspension geometry that actually wins races
Module: Test the suspension before the car turns a wheel
Estimated duration: 60 minutes
The rule
Before you let the car prove anything on track, make the wheel path prove itself in the shop. The job is not to admire a suspension drawing. The job is to confirm that the real wheel, on the real chassis, moves through bump, droop, and roll the way your model and your setup sheet say it should.
Suspension kinematics is the relationship between suspension motion, the vehicle, the ground, and the dynamic behavior that follows. For you as an intermediate builder or setup engineer, that becomes a practical question: when this wheel moves vertically, does camber go where you intended, does steer stay under control, do the left and right sides create the roll-center behavior you modeled, and does the hardware expose slack or hysteresis before the driver has to discover it at speed.
That last part matters. A road-racing suspension is not just a collection of arms and joints. Its purpose is to keep the tire in a useful attitude to the ground while chassis motion, load transfer, and tire deformation are trying to change that attitude. The tire has a limited total force budget, and it works best only when its operating state is kept inside the useful range. If the wheel path quietly adds rear toe-out in bump, gives away the camber you expected, or moves the roll center somewhere your spreadsheet did not predict, the driver will feel the symptom later as scrub, instability, poor grip, or a balance problem. Rig testing lets you find that cause while the car is still quiet.
The principle is simple: model, measure, sweep, compare, correct, then track-test. The model gives you the intended wheel path. The shop measurement gives you the real geometry. The rig sweep shows how the real wheel moves. The comparison tells you whether the disagreement is a measurement error, a model error, an assembly error, or a real characteristic you must design around. Only after that do track development and data logging get to do their job.
What this lesson is and is not
This lesson is about validating the kinematic wheel path before the car runs. It sits between two sibling skills. It is not mainly about the alignment your bushings create under load; that is a compliance question, and the related lesson in this module handles that boundary. It is also not about predicting chassis torsional stiffness before cutting tube; chassis stiffness matters to suspension behavior, but this lesson assumes you are validating the suspension path itself rather than proving the frame.
The useful phrase is pseudo-static testing. You are not reproducing a full lap. You are holding the chassis in a controlled shop condition and moving the suspension through its travel so you can observe geometry, rates, slack, and hysteresis. That makes the test narrower than track development, but also cleaner. On track, springs, anti-roll bars, aero, tire heat, driver input, and surface variation all speak at once. On the rig, you can ask a more disciplined question: with the body held in a known position, where does the wheel go.
That narrowness is a strength. The source material separates shop geometry development from skidpad and track development. Shop work can do a great deal when you understand the approximate targets and the theory behind them. Track work supplies the proof of the theories and the final degree of adjustment. Neither replaces the other. If you skip the shop test, the track becomes a noisy place to debug basic geometry. If you stop at the shop test, you have geometry that has not yet survived real load, tire behavior, driver input, and the last setup compromise.
Start with the questions you need the rig to answer
A kinematics rig test can produce a pile of numbers. A useful validation pass starts with a short list of questions. Do not ask whether the suspension is good. Ask whether the real wheel path matches the intended wheel path closely enough to let track development begin.
The first question is camber control. Unequal-length, non-parallel independent suspension is common in purpose-built road-racing cars because it can reduce positive camber of the laden wheel in roll and let the designer choose a roll-center and camber-control compromise. Your rig test should confirm the camber-change curve you believe you built. If you expected the outside tire to gain useful negative camber as the car rolls and the rig shows the tire standing up instead, the track session will not fix the geometry.
The second question is steer through travel. The ideal toe condition is close to zero across bump, roll, acceleration, and braking, but the practical target is more nuanced. Toe-in creates tire scrub and drag. Toe-out, especially at the rear, creates instability. When toe variation cannot be eliminated, the safer bias is usually slight toe-in rather than a hidden toe-out condition. For roll steer, the general consensus in the source material is nearly zero, or a tendency toward roll understeer. Your rig test should tell you whether the car is moving toward that condition or away from it.
The third question is whether the left and right sides make sense as a pair. Independent suspensions cannot be validated honestly by assuming that the left half and right half behave symmetrically. That assumption can create large errors. The complete suspension pair determines the roll-center behavior you actually care about. If you measure one front corner beautifully but never analyze the two fronts together, you may miss the transverse behavior that matters when lateral force goes through the car.
The fourth question is whether the physical car agrees with the kinematic model. A complete 3-D model can calculate camber, steer, roll-center location, and related parameters as chassis position changes, but only if the input geometry is real. If the measured car does not match the model, you do not yet know whether the model is wrong, the measurements are wrong, or the assembly is wrong. The rig is where that gets separated.
The fifth question is whether the suspension contains slack or hysteresis that should not be hidden inside a clean curve. Pseudo-static tests can identify poor characteristics in the mechanism because they evaluate how the suspension moves on the real vehicle. A curve that traces one path upward and a different path downward is telling you something. A wheel that hesitates, jumps, or produces inconsistent readings is also telling you something. The answer may be a joint, a fixture, a bushing, a loose plate, or a measurement process, but the point is that the car told you before the driver had to.
Build the measurement foundation before sweeping the wheel
The rig test is only as good as the geometry you feed it. The source material describes a detailed chassis geometry process in which two people can typically measure the car in one day, recording the X, Y, and Z coordinates of 48 chassis and suspension points relative to a level surface plate. That level reference matters. You are trying to compare physical geometry to a model. If your reference plane is vague, every later curve is built on guesswork.
This does not mean every club-racing shop has to own the same tools as a professional engineering program. It does mean you should be honest about the quality of your inputs. If pickup-point positions are estimated from a tape measure hanging in space, do not treat the resulting roll-center curve as a truth source. If the chassis is not sitting squarely on a known reference, do not pretend a left-right difference is a design feature. If a spindle pickup location is inferred rather than measured, do not chase an arm pickup point until you have checked the upright.
The source material calls out spindle geometry because it is easy to underestimate. The front suspension components in the example process are measured off the car on a fixture that locates key spindle features, especially the upper and lower ball-joint centers. That is not a decorative detail. A double-wishbone model depends on the real inboard and outboard pivot locations. A small error at the upright can look like a major error in camber gain or instantaneous-center location.
You also need to know what you are trying to predict. If you are calculating tire normal loads or steady-state behavior, spring rates and anti-roll-bar torsional stiffness also become part of the required measured data. That is adjacent to wheel-path validation, not a substitute for it. The wheel-path test answers where the wheel goes. The rate measurements help explain how load and attitude may develop once the car is moving.
A practical measurement foundation has four parts. First, establish the chassis reference: level surface, known supports, and a fixed ride-height datum. Second, document suspension pickup points and wheel-center positions in a coordinate system you can reproduce. Third, measure the spindle or hub geometry well enough that ball-joint centers and wheel-face planes are not guesses. Fourth, record the setup state of the car: static alignment, installed components, ride height, and any parts removed or disconnected for the sweep. Without those four pieces, your rig results may be interesting, but they are not clean validation evidence.
Fixture the car so the wheel movement is the only story
The physical setup described in the source material is deliberately plain. The chassis is held firmly on blocks or jack stands. A large flat plate is bolted to the wheel or hub flange. Dial gauges are mounted on a portable but rigid stand and positioned so they contact the wheel face about a diameter apart, either horizontally or vertically. The wheel is then moved vertically through its total travel while the gauges track the change in the wheel face.
The important words are fixed, rigid, flat, and repeatable. If the chassis support moves, your wheel path now contains chassis movement. If the gauge stand flexes, your reading contains stand movement. If the plate is not flat or is not tight to the hub, your reading contains fixture error. If the wheel is not moved through a controlled path, your curve contains operator noise.
The flat plate gives the gauges a clean surface. Two gauge contacts placed about a diameter apart give you angular information from the face movement. With one orientation, you can read camber change. Rotate or reposition the measurement arrangement as needed and you can read steer change. The exact fixture can vary by shop, but the discipline cannot. You need a known face, a known reference, and enough repeatability to trust differences between positions.
Move through the whole travel you intend to evaluate. Do not only measure static ride height. Static alignment is one frame from a moving picture. The failure you care about may appear in bump, droop, roll, acceleration squat, or braking dive. The source material is explicit that toe ideally would be zero under bump, roll, acceleration, and braking conditions. A car can look respectable at ride height and still generate a bad curve once the wheel moves.
A good rig operator watches both the numbers and the mechanism. If the gauges move smoothly but the joint binds, note it. If the wheel face repeats upward but not downward, note it. If a curve changes after you tap a rod end, note it. Pseudo-static testing is valuable because it can show rates, geometry, hysteresis, and slack. Do not reduce it to a single spreadsheet value.
Compare the physical sweep to the model, not to memory
A complete 3-D kinematic model can calculate wheel camber, steer, roll-center location, instantaneous center, roll moment arm, and jacking-force related behavior as chassis position changes. The model is powerful because you can study different components and attachment-point changes quickly. But the model is not self-validating. It becomes useful when the car you measured and swept agrees with it well enough for the decisions you are making.
The comparison should be visual and procedural. Plot camber versus wheel travel. Plot steer or toe versus wheel travel. Compare the left side and right side. Mark static ride height and the travel range you expect the car to use. If you are studying roll, analyze the suspension pair instead of one side in isolation. If the roll-center path is central to the design decision, compare the complete pair behavior rather than only the inboard pickup sketch.
When the curve disagrees with the model, do not immediately change parts. Recheck the hierarchy. First, confirm the chassis coordinate data. Second, confirm the spindle or hub geometry. Third, confirm the wheel-face fixture and gauge readings. Fourth, confirm that the model uses the same pickup points, ride height, and component lengths as the car. Only then treat the disagreement as a real design or assembly issue.
This is where many teams waste time. They use a model to make a change, use a rough measurement to dislike the change, then use the next track session to decide which guess was less wrong. A cleaner process is slower at the front and faster overall. Measure the real car. Sweep the real car. Compare the real curve to the model. Fix the reason for the disagreement. Then decide whether to change geometry.
The model is also where left-right assumptions get exposed. Early suspension models often assumed symmetric behavior between the left and right halves. That can be acceptable for non-independent suspensions, but it can create large errors for independent suspensions. Your rig process should make symmetry a measured result, not an assumption. If the left front and right front create different camber or steer curves, the next question is whether the car is built asymmetrically by design, assembled incorrectly, damaged, measured poorly, or modeled lazily.
Read roll center as a result of the suspension pair
Roll center is one of the dominant kinematic factors in suspension analysis. It is not a magic dot that fixes handling by itself, but it shapes the roll moment arm and jacking-force picture that the rest of the setup has to live with. The horizontal location of the roll center and jacking forces can produce additional rollover moment that many simplified treatments miss. That is why you should not validate roll-center behavior from a single-side sketch.
For an independent suspension, both sides must be analyzed together. The complete pair lets you study the resulting roll center. The rig test gives you a way to check whether the modeled pair behavior is plausible on the real car. If a left-front sweep and a right-front sweep look tidy in isolation but the pair produces a strange lateral roll-center migration, you have not finished validating the wheel path.
The practical lesson is to treat camber, steer, and roll-center movement as connected. A suspension design often trades among roll-center location, camber control, roll moment arm, and jacking effects. If you change an attachment point to improve one curve, you may move another curve. The rig cannot decide your compromise for you. It can tell you whether the compromise you think you chose is the compromise sitting on the car.
For an intermediate driver-engineer, this is the level of roll-center discipline that matters before track time. You do not need to turn every club car into a research paper. You do need to know whether your model says the front roll center does one thing while the real measured suspension pair does another. If that disagreement exists, the driver feedback from the first session will be filtered through an unknown chassis attitude problem.
Use toe and roll steer as safety and confidence gates
Camber curves often get the attention because tire attitude is easy to visualize. Toe and roll steer deserve equal respect because they can turn a geometry issue into a stability issue. The source material gives a plain practical target: ideal toe-in would be zero under bump, roll, acceleration, and braking. Because any toe-in creates scrub and drag, you do not want to add it casually. But any toe-out, especially at the rear, can cause instability. When you cannot eliminate the toe change, leaning toward slight toe-in is the safer direction.
That is a validation gate. If the rear suspension sweeps into toe-out in the travel range the car will use, do not hide that result under a static alignment number. The car may feel nervous on entry, vague in roll, or unsettled over bumps. The driver may describe it as a confidence problem, but the rig is showing a mechanical reason. You may still choose a static setting that works around a small curve, but you should make that choice knowingly.
Roll steer gets a similar gate. Nearly zero is the clean goal, with a bias toward roll understeer when the design cannot be neutral. Roll oversteer is a dangerous surprise because it can add yaw as lateral load builds. A driver may not separate that from tire limit, brake release, or steering input. The rig gives you a chance to see whether the suspension is contributing yaw by itself.
This does not mean every measured toe change is unacceptable. Real cars compromise. Racing regulations may limit pivot relocation. Production-based suspensions may not give you enough adjustment directions to make every curve perfect. The point is to know which compromise you are carrying. A small, known, stable toe-in tendency is very different from an unseen rear toe-out bump curve discovered when the car gets light over a crest.
Separate validation from development
The source material draws a useful boundary between shop geometry work and track development. In the shop, theory and approximate goals let you design and adjust a great deal. On the skidpad or track, the proof of those theories and the final degree of adjustment appear in the real situation. Your kinematics rig test belongs on the shop side of that boundary.
A passed rig test does not mean the car is optimized. It means the wheel path is understood well enough to begin dynamic work. The later dynamic work can use objective testing and data logging to validate vehicle response. Data logging can reveal events and support intelligent changes that improve behavior. But data logging is more valuable when it is not being asked to identify a basic geometry error that the shop could have found.
Think of the workflow as a chain. Geometry measurement supports the kinematic model. The kinematic model predicts wheel path and related parameters. Pseudo-static rig testing validates the real suspension movement and exposes slack or hysteresis. Track testing and data logging validate the dynamic behavior. Development changes then feed back into the model and measurement record.
If you break the chain, the meaning of your evidence gets weak. A track balance problem without a validated wheel path can lead you to springs, bars, dampers, aero, or driving style when the rear toe curve is the root cause. A model without a measured car can make elegant predictions from false coordinates. A rig sweep without a model can tell you what happened but not why a pickup-point change might improve it. The skill is not any one tool. The skill is making the tools check each other in the right order.
What improving looks like
You know you are improving when your rig work becomes less dramatic. Early on, every sweep may produce surprises. A wheel-face plate is loose. A gauge stand moves. A pickup point was measured from the wrong reference. The model uses the wrong ball-joint center. The left and right sides do not match. That is normal. The improvement is that each surprise gets captured and corrected instead of carried into the next test.
The first calibration cue is repeatability. Sweep the same corner again and the curve should come back. If it does not, the problem is fixture, slack, hysteresis, or process. Do not tune from non-repeatable data.
The second cue is curve shape. Camber and steer curves should make mechanical sense for the linkage you built. A sharp jump in the middle of travel is not a subtle setup feature until proven otherwise. It is a reason to inspect the joint, plate, stand, bushing, or measurement sequence.
The third cue is model agreement. The model does not have to predict every physical imperfection, but it should match the measured kinematic path closely enough that a pickup-point change in the model teaches you something about the car. If model and rig diverge wildly, keep validating inputs before making design conclusions.
The fourth cue is paired-suspension sanity. Left and right behavior should either match or have a documented reason not to match. The resulting roll-center behavior should be something you can explain from the measured geometry, not a surprise produced after the fact.
The fifth cue is cleaner track development. When the car finally runs, the first sessions should be about the remaining dynamic compromise, not basic unknowns. If the driver reports instability, you can compare that report to known toe, camber, and roll-steer curves. If data logging shows an event, you can decide whether it fits the validated wheel path or points somewhere else. That is the payoff: fewer ghosts, better decisions.
Worked example: rear toe sweep before the first track session
Imagine a production-based road-racing car with enough rear suspension adjustment to change pivot locations or link lengths, but not enough freedom to make every curve perfect. Static alignment looks sensible. The rear wheels show a small toe-in number at ride height. If you stop there, you may think the rear is safe.
Now put the car on stands or blocks, bolt a flat plate to the rear hub face, set the dial gauges on a rigid stand, and move the wheel vertically through the travel range you expect the car to use. The important output is not only the static toe. It is the shape of the steer curve. If the rear wheel moves toward toe-out as it bumps, especially in the region associated with roll or braking attitude, the rig has found an instability risk.
The source material gives you the decision logic. Zero toe through all conditions would be ideal, but real cars compromise. Toe-in costs scrub and drag. Rear toe-out costs confidence and stability. Roll steer should be nearly zero or tend toward roll understeer. If your sweep shows rear toe-out in the useful travel range, you do not solve that by telling the driver to be calmer. You first recheck the fixture and repeat the sweep. If the result repeats, you inspect assembly, pickup points, and the model. If rules and hardware allow, you adjust the geometry. If they do not, you choose a static setting that keeps the car away from the dangerous part of the curve and document the compromise for track validation.
The success condition is not a perfect line on the plot. The success condition is that no one is surprised later. The driver may still tune bars, springs, and dampers on track, but the team already knows whether the rear suspension is adding an unwanted steer input as it moves.
Worked example: model validation on a Formula Ford style suspension
The modeling paper in the corpus names Formula Ford examples among the race cars used for design specifications and modeling results. That is a useful mental model for this lesson because a Formula Ford style passive unequal-length suspension rewards careful geometry and punishes lazy assumptions. The car is light, the suspension is exposed enough to measure, and small geometric errors can change camber and steer behavior.
Start with the measured chassis and suspension coordinates. The referenced procedure measures 48 points relative to a level surface plate and treats spindle geometry as its own measurement problem. That matters because a double-wishbone model depends heavily on the real upper and lower ball-joint centers. If those are guessed from visible casting features, your model can predict a camber curve that belongs to the drawing, not to the car.
Next, build or update the 3-D kinematic model from the measured data. Use it to predict camber, steer, and roll-center movement as chassis attitude changes. Then perform the physical sweep with the hub plate and dial gauges. If the rig camber curve is close to the model, the model earns trust. If the rig curve is different, do not immediately move the suspension pickup point. Recheck the spindle centers, inboard pickup coordinates, ride-height datum, and gauge fixture first.
A common good outcome is not that the first model is perfect. A good outcome is that the disagreement leads to a better model and a better car record. After correction, the model and rig agree closely enough that future component or attachment-point changes can be evaluated quickly. That is when computer modeling becomes a development tool instead of a polished guess.
Common mistakes
The first mistake is validating static alignment and calling it wheel-path validation. Static camber and toe are only one position. The suspension has to move through bump, droop, roll, acceleration, and braking attitudes. Good looks like a sweep: the wheel is moved through travel and the curve is recorded.
The second mistake is measuring one side and assuming the other side matches. For independent suspensions, that assumption can produce large errors. Good looks like paired analysis: both left and right sides are measured and used together when studying roll-center behavior.
The third mistake is trusting the model before measuring the car. A 3-D kinematic model needs real coordinates. Good looks like a measured reference plane, documented pickup points, measured spindle geometry, and a model built from those inputs.
The fourth mistake is blaming the driver for a rear-steer problem the rig could have found. Rear toe-out through travel is an instability risk. Good looks like using the rig result as a safety gate before the car runs hard.
The fifth mistake is tuning from a non-rigid fixture. If the gauge stand moves, the chassis rocks, or the hub plate shifts, the numbers are not the wheel path. Good looks like a fixed chassis, a rigid stand, a flat plate tight to the hub, and repeatable sweeps.
The sixth mistake is treating hysteresis as noise to average away. Pseudo-static tests can reveal hysteresis and slack. Good looks like investigating the reason the upward and downward sweeps differ before using the curve for design decisions.
The seventh mistake is expecting the shop rig to replace track development. The shop can validate geometry and expose mechanism problems. Track work still supplies the real-life proof and final adjustment. Good looks like using each test at the right stage: rig first for wheel path, data logging and track work later for dynamic behavior.
Drill: the one-day wheel-path validation pass
Run this as a two-person shop drill before the car attends its next serious test. The count is one car, two people, one shop day, three passes, and one written decision sheet.
Pass one is the geometry foundation. Establish a level reference and record the chassis and suspension coordinates that your model needs. Use the professional example as the benchmark: a full procedure can measure 48 chassis and suspension points relative to a level surface plate. If your shop process is simpler, write down exactly what was measured and what was estimated. The success criterion for pass one is that the model inputs are traceable to the car rather than to memory.
Pass two is the wheel-face sweep. Bolt a flat plate to the wheel or hub flange. Put the dial gauges on a rigid stand and contact the wheel face about a diameter apart. Sweep the wheel through the travel range you intend to evaluate. Repeat the sweep. The success criterion for pass two is repeatability: the second sweep returns the same basic camber and steer shape, and any slack or hysteresis is recorded rather than hidden.
Pass three is the comparison. Plot or tabulate camber and steer against travel, compare left and right sides, and compare the physical curves to the 3-D kinematic model. If the car has independent suspension, analyze the pair for roll-center behavior instead of treating each side as an isolated mechanism. The success criterion for pass three is a clear decision sheet with three categories: accepted as matching the model, corrected because measurement or assembly was wrong, or carried forward as a documented compromise for track validation.
Do not finish the drill with a vague note that the suspension looks fine. Finish with evidence. The evidence can be simple, but it should answer the questions this lesson teaches: where the wheel went, whether toe and camber behaved acceptably, whether left and right made sense as a pair, whether slack or hysteresis appeared, and whether the model now represents the car.
When to leave the rig and go to the track
You leave the rig when the basic wheel path is known, not when the whole race car is solved. A good rig result says the geometry is measured, the physical sweep repeats, the model is corrected, the left and right pair makes sense, and no unacceptable steer or camber surprise remains in the travel range you plan to use.
After that, the car still needs dynamic validation. The source material supports validating vehicle dynamic analysis with data logging and using logged observations to make intelligent changes. That is the next layer. Track data can show how the car negotiates maneuvers, how setup changes affect behavior, and how the driver experiences the compromise. It can also expose problems the pseudo-static rig was not designed to reproduce.
The boundary is important. If the car has a known rear toe-out bump curve, do not call that a track-development item and hope the driver drives around it. Fix or document the geometry first. If the wheel path is clean but the car still understeers in a real corner, then track development has a proper job to do. That separation keeps each test honest.
Author Review
No quiz questions are attached to this lesson.
Sources
| # | Document | Chunk | Pages | Score | Collection |
|---|---|---|---|---|---|
| 1 | Racing Chassis and Suspension Design Carroll Smith | 0325ae22-f074-b2da-251c-c8e478b29e9a | 130 | 1 | uio_books_raw_v1 |
| 2 | Race Car Engineering Mechanics Paul Van Valkenburgh | fc51642b-940c-c435-b404-6b029c85542e | 33 | 1 | uio_books_raw_v1 |
| 3 | Racing Chassis and Suspension Design Carroll Smith | 0ec16bd1-2d0f-bc14-b9a3-08bc411dfe18 | 75 | 1 | uio_books_raw_v1 |
| 4 | Racing Chassis and Suspension Design Carroll Smith | b4bfe891-d8ab-74b3-6b93-3e070d953e21 | 187 | 1 | uio_books_raw_v1 |
| 5 | Racing Chassis and Suspension Design Carroll Smith | cb98cc00-5481-d32d-c43d-18838cb107e5 | 184 | 1 | uio_books_raw_v1 |
| 6 | Racing Chassis and Suspension Design Carroll Smith | 52047a73-bbbf-e4e8-51ff-bb6cdbc0101b | 134 | 1 | uio_books_raw_v1 |
| 7 | Racing Chassis and Suspension Design Carroll Smith | 31ea3f7c-e652-53e4-3a96-f2ef6454fe84 | 184 | 1 | uio_books_raw_v1 |