# Frequency-domain suspension analysis & modal decomposition

> Transfer functions, power spectral densities and modal decomposition (heave, pitch, roll, warp) characterise the suspension independent of a particular bump. They reveal body and wheel-hop resonances, quantify grip (tyre load variation) vs platform (body motion), and are the basis of 4/7-post rig and on-track damper optimisation.

Area: Dampers · Starting level: L3 Advanced · Updated 2026-11-16
Source page: https://race-engineer-academy.vercel.app/lesson/dampers-frequency-domain-modal

## L1 · Foundation

### Why look at frequencies?
Any road input – bumps, ripples, kerbs – can be broken down into many sine waves of different frequencies (Fourier). A suspension responds very differently to each:
- **Below the body frequency** (< ~2 Hz): the body follows the road (long undulations, crests).
- **Around the body frequency** (~2–5 Hz on race cars): the body **resonates** – it moves *more* than the road input unless damped.
- **Between body and wheel-hop** (~5–12 Hz): the body is isolated; the wheel follows the road.
- **Around wheel hop** (~12–20 Hz): the **wheel resonates** on the tyre – tyre load fluctuates strongly, grip suffers.
- **Above**: the tyre itself absorbs input.

So a race car has two key resonances per corner: the **body mode** (platform) and the **wheel-hop mode** (grip). Damping is mainly about controlling these two peaks.

The road input frequency depends on car speed: a surface wavelength λ excites $f = V/\lambda$. At 250 km/h (69 m/s), a 20 m undulation excites 3.5 Hz – right on the body mode; a 4 m ripple excites 17 Hz – wheel hop. The same track bump can be harmless at 100 km/h and a problem at 250 km/h.

**Modes of the whole car.** Four corner motions combine into four modes: **heave** (all together), **pitch** (front vs rear), **roll** (left vs right) and **warp** (diagonals opposite). Each has its own stiffness, inertia and damping – and each matters differently for aero and grip.

## L2 · Practitioner

### Practical frequency analysis with car data
**PSD of damper or ride-height signals.** Export damper positions (or laser ride heights) for a stable section of the lap (or the whole lap), at ≥ 200 Hz (500 Hz better for wheel hop). Compute the **power spectral density** with Welch's method (segments of 2–4 s, Hann window, 50 % overlap). Plot on log axes. Look for:
- Peak at **2–5 Hz** = body mode; its height relative to the low-frequency level tells you how well damped it is.
- Peak at **12–20 Hz** = wheel hop; tall/narrow = under-damped wheel.
- Narrow spikes at fixed frequency across speeds = mechanical (driveline, engine order) – not suspension.
- Peaks that move with speed = road wavelength excitation ($f = V/\lambda$).

Segment by speed (e.g. 150–200 and 200–250 km/h) so the road input spectrum is comparable between runs.

**Modal channels** (Segers ch12, displacements positive in bump):
$$H = \tfrac{1}{4}(z_{LF}+z_{RF}+z_{LR}+z_{RR}),\quad P = \tfrac{1}{4}(z_{LF}+z_{RF}-z_{LR}-z_{RR})$$
$$R = \tfrac{1}{4}(z_{LF}-z_{RF}+z_{LR}-z_{RR}),\quad W = \tfrac{1}{4}(z_{LF}-z_{RF}-z_{LR}+z_{RR})$$
PSDs of $H$ and $P$ describe the aero platform; $R$ describes roll dynamics; $W$ (warp) is pure road input (track twist, kerbs) and produces diagonal load changes without body motion.

**Using it**
- Compare runs before/after a damper change: did the body peak drop? did the wheel-hop peak drop or grow?
- On aero cars, minimise heave/pitch PSD energy in 1–6 Hz at high speed (platform).
- On mechanical cars, minimise wheel-hop peaks and 2–10 Hz tyre-load content (grip).

**Four-post rig**: shakes each wheel with sine sweeps or track replays and measures contact-patch load directly – the most controlled place to *measure* tyre load variation (on track you need wheel force transducers, or pushrod load plus hub acceleration – see L4). Limits: non-rolling tyre, no real aero (7-post adds aero actuators), car on rig may differ from car on track.

## L3 · Advanced

### Transfer functions and the quarter car
**Single-DOF transmissibility** (body displacement / road displacement) for frequency ratio $r = f/f_n$:

$$T(r) = \sqrt{\frac{1+(2\zeta r)^2}{(1-r^2)^2+(2\zeta r)^2}}$$

$T ≈ 1$ at low $r$, peak near $r ≈ 1$ (≈ $1/(2\zeta)$ for small ζ; ≈1.4–1.5 at ζ = 0.5, ≈1.2–1.3 at ζ = 0.7), crossover $T = 1$ at $r = \sqrt 2$ regardless of ζ, and isolation above. Note the trade-off: more damping lowers the resonance but **raises transmissibility above** $\sqrt 2 f_n$ – Segers fig. 12.12 shows over-damped settings increasing response in the 1–10 Hz band, hurting grip and comfort.

**Quarter car** (sprung $m_s$, unsprung $m_u$, $K_w$, $c$, $K_t$) gives three useful FRFs (Segers fig. 12.10): road → body displacement (platform/stability), road → tyre deflection (∝ contact-patch load variation, i.e. grip), road → suspension travel. Optimising damping on the **peak** of each gives *different* optima: grip prefers less damping than body control. Heavier unsprung mass raises the tyre-load FRF around wheel hop – the reason regulations and engineers chase low unsprung mass.

**Modal stiffness matrix** (Segers §12.5): with front/rear corner rates $k_f, k_r$ and ARB rates, the modal elasticities are e.g. heave $K_H = \tfrac12(k_f+k_r)$, roll from springs and ARBs, and off-diagonal terms $\propto (k_f-k_r)$ and $(k_{arb,f}-k_{arb,r})$ coupling heave–pitch and roll–warp. The roll–warp coupling term is what sets **mechanical balance** (LLTD): front/rear roll stiffness difference shows up as warp coupling. Tyres and chassis enter in series, softening especially warp.

**Effective modal masses.** Race cars concentrate mass centrally, so the equivalent mass per wheel in roll is only ~20–40 % and in pitch ~60–80 % of the quarter mass (Segers). Roll therefore has high frequency, high stiffness and low inertia; with the tyre comparable in stiffness to the suspension in roll, the roll mode gets "trapped" with the tyre's low damping – race-car roll transfer functions show poorly damped peaks that dampers can only partly control.

## L4 · Expert

### FRFs from on-track data and their limits
**Building FRFs from car sensors** (Segers §12.8): with body vertical acceleration $G_{vert}$, suspension displacement (→ suspension acceleration), wheel loads (pushrod) and known $m_u$, $K_t$:
- Hub acceleration $a_{hub} = G_{vert} + a_{susp}$ (or measured with an upright accelerometer).
- Contact-patch load $F_{cp} = F_{susp} + m_u(a_{hub} + g)$ (Segers' Eq. 12.35 omits $g$ because it works with dynamic components; the $g$ term gives absolute load. Up positive; $F_{susp}$ = total spring + damper + bump-stop force at the wheel; a raw accelerometer reading already contains $g$); contact-patch (road) acceleration $a_{cp} = \ddot F_{cp}/K_t + a_{hub}$ assuming a linear tyre spring.
- FRFs as ratios of spectra: body/contact-patch and hub/contact-patch. Use **cross-spectral estimators** ($H_1 = S_{xy}/S_{xx}$) and check **coherence** – low coherence (< 0.7) means nonlinearity, noise or uncorrelated inputs, and the FRF there is unreliable.

**Practical requirements**: ≥ 500 Hz for hub content; anti-alias filtering; position sensor resolution limits high-frequency acceleration derived by double differentiation (Segers notes this impairs high-frequency data); speed-segmented averaging; correction for anti-geometry load paths in braking.

**Non-linearity.** Dampers are neither linear nor symmetric, bump stops and packers are piecewise, friction locks the suspension at small amplitudes, aero stiffness varies with speed and ride height. FRFs are therefore *amplitude- and speed-dependent* – quote them with the operating condition. Asymmetric damping additionally produces mean offsets (jacking) that no linear FRF shows.

**Aero-coupled dynamics.** At speed, $\partial F_z/\partial h$ and $\partial F_z/\partial \theta$ add (usually destabilising) stiffness and, through flow lag, effective damping to heave and pitch. A 7-post rig with aero actuators or on-track FRFs are the only way to see the real platform modes; porpoising appears as a speed-dependent, lightly damped (or growing) heave/pitch peak in the 3–8 Hz region.

**From analysis to decision.** Define a cost: e.g. weighted sum of contact-patch load RMS (grip, per axle, weighted by aero balance/tyre load sensitivity) and ride-height RMS at the aero reference points (platform). Optimise damper clicks on a rig or a validated vertical-dynamics model driven by measured road profiles, then confirm on track with PSD/histogram metrics and the driver. That closes the loop between Segers ch11 (time/statistical) and ch12 (frequency) methods.

## Key formulas

**Excitation frequency from road wavelength**

$$f = \frac{V}{\lambda}$$

- V: car speed (m/s)
- \lambda: road wavelength (m)

**SDOF transmissibility**

$$T(r) = \sqrt{\frac{1+(2\zeta r)^2}{(1-r^2)^2+(2\zeta r)^2}},\quad r = \frac{f}{f_n}$$


**Modal decomposition**

$$\begin{bmatrix}H\\P\\R\\W\end{bmatrix} = \frac{1}{4}\begin{bmatrix}1&1&1&1\\1&1&-1&-1\\1&-1&1&-1\\1&-1&-1&1\end{bmatrix}\begin{bmatrix}z_{LF}\\z_{RF}\\z_{LR}\\z_{RR}\end{bmatrix}$$

- H, P, R, W: heave, pitch, roll, warp
- z: corner wheel displacement

**Contact patch load from hub dynamics**

$$F_{cp} = F_{susp} + m_u\,(a_{hub} + g)$$

- F_{susp}: total suspension force on the hub, referred to the wheel (N)
- a_{hub}: hub vertical acceleration excluding gravity, up positive (m/s²)

**FRF estimator**

$$H_1(f) = \frac{S_{xy}(f)}{S_{xx}(f)},\qquad \gamma^2(f) = \frac{|S_{xy}|^2}{S_{xx}S_{yy}}$$

- S_{xy}: cross-spectral density
- \gamma^2: coherence (0–1)

## Trackside takeaways

- Two key resonances per corner: body (~2–5 Hz, platform) and wheel hop (~12–20 Hz, grip).
- Road wavelength excites f = V/λ – the same bump matters differently at different speeds.
- More damping lowers the resonance peak but raises transmission above √2·f_n: grip optimum < platform optimum.
- Modal channels (heave, pitch, roll, warp) separate platform, balance and road-twist effects.
- On-track FRFs need high sampling, cross-spectral estimators and coherence checks; results are amplitude- and speed-dependent.

## By car category

### GT3

Body ~3 Hz, wheel hop ~14–17 Hz. Ride-height PSD at high speed shows bump-stop-stiffened body peaks; good for checking platform at Spa/Monza.

### LMP2

Teams use 7-post rigs and on-track FRFs; heave/pitch modal analysis is the core of platform work.

### Hypercar (LMH/LMDh)

Large 18-inch wheels and tyres dominate unsprung mass (hybrid motors are chassis-mounted; carbon brakes are light) – measure the hop frequency rather than assume it. Aero platform modes at speed are critical for BoP-limited downforce.

### FIA F3

Spec parts – frequency analysis helps choose clicks and heave settings; watch tyre-dominated heave peaks.

### FIA F2

Ground-effect-like floor sensitivity: monitor heave/pitch PSD at high speed for aero-induced bouncing.

### Super Formula

High aero, light car: platform modes and their damping strongly linked to lap time.

### IndyCar

Street-course bumps excite wide frequency bands; FRF-based damper development is a major team differentiator.

### Formula 4

Mechanical-grip car: focus on wheel-hop and 2–10 Hz tyre-load content rather than platform.

## References

- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §12.1 Introducing Frequency Analysis
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §12.3 Theoretical Analysis (transfer functions, quarter-car, unsprung mass and damping effects)
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §12.4 Suspension Optimization Using Frequency Analysis
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §12.5 Modal Analysis
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §12.8 Frequency Analysis from Sensor Data
- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 16 Ride and Roll Rates
- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 22 Dampers
