# Roll gradient & roll ratio as setup tools

> Measure suspension roll from damper pots, regress it against lateral g, and you get front and rear roll gradients and their ratio – objective, lap-by-lap metrics of roll stiffness and its distribution that also flag broken bars and deflating tyres.

Area: Springs, Anti-roll Bars & Heave · Starting level: L2 Practitioner · Updated 2026-11-16
Source page: https://race-engineer-academy.vercel.app/lesson/springs-roll-gradient-roll-ratio

## L1 · Foundation

### Roll per g
The **roll gradient** is body roll angle per g of lateral acceleration:

$$RG = \frac{d\phi}{d a_y}\quad [\mathrm{deg}/g]$$

It is a normalised measure of roll stiffness: a stiff car rolls little per g. Road cars sit around 3–7 °/g; race cars with downforce below ~1.5 °/g; high-downforce formula cars only ~0.05–0.2 °/g of *suspension* roll.

From four damper potentiometers you can calculate **front and rear suspension roll** separately. Left-minus-right damper displacement, converted to wheel travel with the motion ratio and divided by the track, gives the angle. Plot roll vs lateral g on an X–Y graph – the slope of the straight-line fit is the gradient. Segers shows a Viper GT car with 0.48 °/g overall, split as 0.31 °/g front and 0.65 °/g rear.

The **roll ratio** is rear suspension roll divided by front:

$$\zeta_\phi = \frac{\phi_r}{\phi_f} \approx \frac{RG_r}{RG_f}$$

In practice Segers takes it as the **slope of rear suspension roll regressed on front suspension roll** (an X–Y plot of the two channels). That slope and the quotient of the two gradients differ by a few % because of scatter and offsets, so pick one estimator and use it consistently.

A higher ratio means the front is relatively stiffer in roll (it rolls less) – a more front-biased roll-stiffness distribution, i.e. towards understeer.

**Why it is useful:** driver feel varies; these numbers don't. If the car was good at this track last year with $RG_f$ = 0.30 and $RG_r$ = 0.60 °/g, you can steer the setup back towards those numbers.

Typical suspension-only gradients (Segers, measured from pots): 2011 Audi GT3 0.30–0.50 °/g; GT1 Corvette 0.25–0.35; IndyCar 0.10–0.20; F1 (2002) 0.03–0.10.

## L2 · Practitioner

### Building the channels and using them
**Math channels** (convention: MR = damper travel / wheel travel; positive = compression):

$$\phi_f = \frac{180}{\pi}\arctan\left(\frac{x_{LF}-x_{RF}}{\mathrm{MR}_f\,t_f}\right),\qquad \phi_r = \frac{180}{\pi}\arctan\left(\frac{x_{LR}-x_{RR}}{\mathrm{MR}_r\,t_r}\right)$$

(Segers writes the same with MR = wheel/damper, i.e. multiplying instead of dividing.) Zero the pots at static ride height on the setup pad so roll is zero when stationary.

**Getting clean gradients**
- Use **quasi-steady cornering** data: gate on low roll *velocity* (damper speeds |v| < ~15 mm/s), exclude kerbs and big bumps (high-pass energy), exclude low speed (< 60 km/h) and pit lane.
- Fit **left and right corners separately** if asymmetric: a difference signals cross-weight, bar preload, asymmetric bump-stop engagement or a different left/right setting.
- Expect scatter from pot accuracy, damping in transients, chassis torsion over kerbs and suspension friction.
- Compute gradients as **lap statistics** and put them in **run charts** across the weekend: each setup change should step the trend; tyre pressure build-up shows as a drift.

**Using them for setup**
- **Bar change check:** Segers' Viper example – front bar soft → hard: $RG_f$ 0.275 → 0.225 °/g, $RG_r$ 0.539 → 0.514, roll ratio 1.84 → 2.18. (The gradient quotient gives 1.96 → 2.28; Segers' 1.84 → 2.18 are regression slopes of rear vs front roll. Use one estimator consistently in your run charts.) If the numbers don't move when the driver says "no change", the change probably didn't happen (wrong blade, link off).
- **Restoring a known balance:** reproduce a reference gradient set rather than reproducing spring numbers – the tyres may have changed.
- **Fault detection:** a linear car must give a straight line. A kink (e.g. rear roll stays zero until front reaches 0.1°) indicates free play – Segers found a broken rear ARB blade this way. A steadily **rising roll ratio** over laps with stable setup points to a **front tyre losing pressure** (front tyre roll stiffness falling) – useful in telemetry-limited series because it's a single number per lap.

## L3 · Advanced

### What the roll ratio really measures
Treat the chassis as rigid. Then **total** roll is the same at both axles; only its split between suspension and tyres differs:

$$\phi_{susp,f} + \phi_{tyre,f} = \phi_{susp,r} + \phi_{tyre,r}$$

Hence (Segers eq. 9.20)

$$\zeta_\phi = \frac{\phi_{susp,r}}{\phi_{susp,f}} = 1 + \frac{\phi_{tyre,f} - \phi_{tyre,r}}{\phi_{susp,f}}$$

The front suspension rolls *less* than the rear not because of some separate "front roll", but because a front-biased roll stiffness sends more load transfer to the front tyres, which therefore deflect more. The roll ratio is a **tyre-deflection-difference** indicator. Consequences (Segers Table 9.6):
- Stiffer front suspension ↑ ratio; stiffer rear suspension ↓ ratio.
- Stiffer **front tyres** (more pressure) ↓ ratio; stiffer rear tyres ↑ ratio.
- Segers' worked example: +10 % rear tyre rate raised the ratio ~5 %; restoring it needed **+25 % rear wheel rate**.

**Converting to stiffness.** With roll moment per g $M_\phi = m_s g\,h_\phi$ (sprung mass × CG height above roll axis), total suspension roll stiffness is $K_\phi = M_\phi/RG$. Segers estimates the distribution as $K_{\phi f}/K_{\phi} \approx RG_r/(RG_f+RG_r)$ and then subtracts the calculated spring contribution ($K_w t^2/2$) to back out the **actual installed bar rate** – a good check of the bar rate tables. Treat this split as an engineering approximation: it ignores the rigid-chassis coupling above and the geometric (roll-centre) path, so use it to *compare* configurations of the same car, not as an absolute LLTD.

**Non-linearity is information.** Bump-stop contact on the outside wheel, ARB preload, droop limiters and aero changes produce kinks. Fit piecewise, or plot gradient vs speed bands – a gradient that falls at high speed often means the outside spring has gone onto its bump rubber.

## L4 · Expert

### Advanced use and pitfalls
- **Separate tyre roll.** Combining pots with ride-height lasers (or load cells/pushrod strain gauges) gives the tyre roll component directly, so you can estimate *dynamic* tyre rate and track it through a stint (pressure build-up, wear). This closes the loop between tyre-pressure strategy and mechanical balance.
- **Chassis torsion.** With finite torsional stiffness, $\phi_{f,total} \ne \phi_{r,total}$; the measured ratio then mixes chassis twist into the "suspension" numbers – Segers notes this is acceptable when comparing setups of one car (chassis is constant) but not across cars. Kerbs excite torsion, another reason to exclude them.
- **Transient contamination.** During roll motion dampers carry part of the roll moment, so instantaneous roll lags lateral g and the X–Y plot forms loops (hysteresis). Filtering both channels identically and gating on low roll rate prevents damping from biasing the slope. The loop area itself is a useful metric of transient roll damping.
- **Aero and speed.** Downforce compresses springs and may bring bump stops into play, so gradients are speed-dependent on aero cars. Bin by speed (e.g. < 120, 120–180, > 180 km/h) – a front gradient that drops above 180 km/h but rear stays constant tells you the front outside bump stop engages and LLTD is moving forward with speed (high-speed understeer).
- **Telemetry economy.** In endurance racing, sending $RG_f$, $RG_r$ and $\zeta_\phi$ as lap statistics costs almost no bandwidth and lets the pit wall cross-check TPMS alarms (real puncture vs sensor fault) – Segers' LMP2 example.

## Key formulas

**Front suspension roll from pots**

$$\phi_f = \frac{180}{\pi}\arctan\left(\frac{x_{LF}-x_{RF}}{\mathrm{MR}_f\,t_f}\right)$$

- x: damper displacement, positive in bump (mm)
- MR_f: damper/wheel motion ratio
- t_f: front track (mm)

**Roll gradient**

$$RG = \frac{d\phi}{d a_y}\;[\mathrm{deg}/g]$$


**Roll ratio**

$$\zeta_\phi = \frac{\phi_r}{\phi_f} \approx \frac{RG_r}{RG_f}$$

- \zeta_\phi: in practice the regression slope of rear on front suspension roll

**Roll ratio vs tyre roll (rigid chassis)**

$$\zeta_\phi = 1 + \frac{\phi_{tyre,f}-\phi_{tyre,r}}{\phi_{susp,f}}$$


**Roll stiffness from gradient (Segers)**

$$K_\phi = \frac{m_s g\,h_\phi}{RG},\qquad \frac{K_{\phi f}}{K_\phi} \approx \frac{RG_r}{RG_f + RG_r}$$

- h_\phi: sprung CG height above roll axis (m)

## Trackside takeaways

- Roll gradient = slope of suspension roll vs lateral g; compute front and rear separately from damper pots.
- Roll ratio = rear/front suspension roll (regression slope, ≈ RG_rear / RG_front); higher = more front-biased roll stiffness. Use one estimator consistently.
- Use steady-state, kerb-free data; fit left and right corners separately; track as lap stats in run charts.
- Tyre pressure changes the suspension roll ratio – compare at equal hot pressures.
- Kinks in roll vs g or front vs rear roll = free play, broken blade or bump-stop contact; a rising roll ratio can mean a deflating front tyre.

## By car category

### GT3

Suspension RG typically 0.3–0.5 °/g. Watch for gradient drop at high speed when outside bump stops engage. Compare left vs right – fuel tank side and driver offset create asymmetry.

### GT4

Larger gradients (~0.5–1.0 °/g) make roll measurements clean and very sensitive to bar changes.

### TCR

Inside rear wheel often lifts – rear roll angle then saturates (droop limited) and the X–Y plot flattens; fit only below lift-off.

### LMP2

Roll gradients and roll ratio are ideal telemetry lap stats for long stints: cross-check with TPMS for slow punctures.

### FIA F3

Very small suspension roll (~0.1–0.2 °/g): pot resolution and calibration matter. Zero pots carefully on the pad.

### IndyCar

Segers' Table 9.4 lists ~0.10–0.20 °/g of suspension roll for a 2001 IndyCar (setup type not stated); on ovals asymmetric setups give deliberately different left/right numbers.

### One-make Cup

Identical cars: roll gradients let you compare your car directly to teammate data and spot damaged parts (bent blades) quickly.

## References

- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §9.1 Measuring Suspension Roll Angle
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §9.2 The Roll Gradient
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §9.3 Using Roll Gradients as a Setup Tool
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §9.4 Front to Rear Roll Angle Ratio
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §9.5 Using the Roll Ratio as a Setup Tool
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §9.6 Suspension Troubleshooting
- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 16 Ride and Roll Rates
