# Math channels: a practical toolbox

> Math channels turn raw sensor data into engineering quantities: corrected speed, understeer angle, roll gradient, slip ratio, total gear ratio, combined g and delta time. Learn the core set, their assumptions and where they break.

Area: Data Acquisition & Analysis · Starting level: L2 Practitioner · Updated 2026-11-16
Source page: https://race-engineer-academy.vercel.app/lesson/daq-math-channels-toolbox

## L1 · Foundation

**A math channel is a calculation done on logged channels, plotted like a sensor.** Every analysis package supports arithmetic, conditions, derivatives, integrals, filters and session constants (track width, wheelbase, motion ratio). They let you see things no sensor measures directly.

The starter set every race engineer should have:

| Channel | What it tells you |
|---|---|
| **Corrected / reference speed** | best estimate of true vehicle speed (undriven wheels averaged, or GPS-fused) |
| **Combined g** $\sqrt{G_{lat}^2+G_{long}^2}$ | how much of the tyres' grip is used through the corner phases |
| **Understeer angle** | steering used beyond what the corner geometry needs |
| **Roll angle and roll gradient** | body roll per g – checks roll stiffness |
| **Slip ratio** | wheelspin or lock-up of a wheel relative to the reference speed |
| **Total gear ratio** | engine RPM / wheel speed – gear in use, clutch slip, missed blips |
| **Delta time** | where one lap gains or loses against another |
| **Brake balance** | front pressure / total pressure |
| **Coasting** | neither throttle nor brake |

Rule of thumb: build and test math channels at the factory on old data, not in the garage between sessions.

## L2 · Practitioner

**The core formulas and their practical use.**

**Reference speed.** Average the undriven wheels (front on RWD) to cancel the inside/outside difference in corners. On cars whose front axle is driven – AWD, or LMH hybrids such as the Ferrari 499P that deploy through the front axle above a regulated speed threshold – use GPS-fused or IMU-corrected speed. LMDh cars (Porsche 963) have a rear-axle hybrid, so averaged front wheels remain a valid reference outside braking.

**Combined g:** $G_{comb}=\sqrt{G_{lat}^2+G_{long}^2}$. A dip between peak braking and peak cornering shows the driver is not blending brake release with turn-in (early braking / poor trail braking).

**Understeer angle** (bicycle model):
$$\delta_{US} = \frac{\delta_{sw}}{i_s} - \frac{57.3\, L\, G_{lat}\, g}{V^2}$$
with steering wheel angle $\delta_{sw}$ (deg), overall steering ratio $i_s$, wheelbase $L$ (m), $V$ in m/s. Gate it to $|G_{lat}|>0.3$ g and $V>15$ m/s. At low speed the $1/V^2$ term amplifies lateral-g noise and offsets (0.05 g of noise is ~0.02° at 250 km/h but ~1.2° at 30 km/h); at low lateral g the channel is all offset (toe, alignment, sensor zero) and no signal. True ratio channels such as $\delta_{US}/G_{lat}$ or roll gradient divide by $G_{lat}$ and explode on straights. Typical race car values are roughly 0.5–2° road-wheel.

**Roll angle** from damper potentiometers:
$$\phi_F = \arctan\left(\frac{x_{LF}-x_{RF}}{MR_F\,T_F}\right)$$
with $MR$ = damper travel / wheel travel, the convention used throughout this academy (wheel rate $= k\,MR^2$). Segers defines the inverse ratio (wheel / damper) and therefore *multiplies* damper travel by it – same physics, so always check which convention a formula or a setup sheet uses. **Roll gradient** is $\phi/G_{lat}$ in deg/g; GT3 cars typically sit around 0.3–0.6 deg/g, high-downforce single-seaters 0.1–0.3 deg/g. The front/rear roll-angle ratio is a quick check on roll stiffness distribution.

**Slip ratio:** $SR = (V_{wheel}-V_{ref})/V_{ref}$. Positive under traction, negative under braking. Peak longitudinal grip for slicks typically sits at 6–12 % slip.

**Total gear ratio:**
$$i_{tot} = \frac{2\pi\, r_{roll}\, n_{eng}}{60\, V}$$
($n$ in rpm, $V$ in m/s, ideally the undriven-wheel reference speed). With this sign the channel steps **up** on each downshift (lower gear = larger ratio). Flat steps = gears; on a downshift, a momentary dip *below* the new gear's line (with the rear wheel speeds dropping below the fronts) = insufficient blip – the driveline drags the engine up and briefly slows the rears; an overshoot *above* the line = over-blip; drifting values within a gear = clutch or wheel slip. Segers multiplies the channel by −1 so that higher gears plot higher, which flips all of these: in his plots under-blipping appears as upward spikes. Always check the sign convention before diagnosing.

**Delta time:** running time difference vs distance between two laps (see the lap-comparison lesson).

**Conditional channels** are products of booleans: coasting = (throttle < 5 %)·(brake < 5 bar); low-oil-pressure alarm = (P_oil < 3 bar)·(RPM > 3000).

## L3 · Advanced

**Assumptions you must respect.**

- **Understeer angle** is a steady-state, linear-range, bicycle-model quantity. In transients (turn-in, kerbs) and near the limit it loses strict meaning, but as a *comparative* channel (same car, same corner, setup A vs B) it is very effective. Kinematic and compliance steer, Ackermann and toe all add offsets; compare deltas, not absolutes. A yaw-rate sensor allows a cleaner alternative: compare measured yaw rate with the steady-state yaw rate implied by steering, or compute $r_{ss}=G_{lat}g/V$ and compare to measured yaw rate to estimate body slip angle rate.
- **Roll angle from damper pots** measures *suspension* roll only. Tyre deflection adds roll you can't see without ride-height lasers; with soft-sidewall GT tyres it can be 20–30 % of the total. Motion ratio is geometry-dependent – use a polynomial from kinematics software rather than a constant if travel is large. Kerbs give enormous instantaneous roll-ratio values: gate them out (|lat g| > 0.8 g, steady steering).
- **Slip ratio** depends on rolling radius, which grows with speed (centrifugal) and shrinks with load. Calibrate wheel circumferences per tyre set at low driving torque (pit limiter, clutch-in roll, light part-throttle); otherwise 1–2 % offsets hide real slip. Do not calibrate flat out: near top speed the driven tyres carry the full drag force and genuinely run about 1–3 % slip.
- **Differentiation amplifies noise** in proportion to frequency. A derivative of a channel with 1 mm resolution at 500 Hz has 500 mm/s quantisation steps per count – filter or use a central difference over several samples.
- **Integration drifts.** Integrated long g for speed must be reset or blended (complementary filter) with wheel or GPS speed.

**Building robust channels.** Use session constants for all setup-dependent numbers (static ride height, motion ratio, steering ratio). Use gating rather than zero-replacement when computing averages, otherwise zeros skew the mean. Name channels clearly with units and keep a library under version control so every engineer in the team computes the same "understeer angle".

## L4 · Expert

**From channels to estimators.** At the top level, math channels become state estimators:

- **Body slip angle** $\beta$ cannot be measured cheaply (optical sensors such as Correvit/Kistler or dual-antenna GNSS are needed). Integrating $\dot\beta = G_{lat}g/V - r$ drifts within seconds; teams fuse it with a bicycle-model prediction in a Kalman filter and reset on straights.
- **Tyre loads** combine static corner weights, aero load from ride-height maps, lateral/longitudinal transfer and damper force (from velocity and the damper map). With load cells or pushrod strain gauges you can close the loop and check the aero map on track.
- **Aero coefficients** from coast-down or constant-speed runs: $C_L A$ from ride-height deflection and spring/tyre stiffness, $C_D A$ from a clutch-in (or neutral) coast-down on a flat straight, after removing rolling resistance (which grows with downforce) and the gravity term $g\sin\theta$ from GPS elevation or a track map; on a lifted coast in gear you must also model engine braking and driveline losses, which can reach several hundred newtons.

The trap at this level is false precision: an estimator with five calibrated constants can be wrong by 10 % and still look perfectly smooth. Validate every estimator against a physical check (pit-lane loads, known downforce at a straight-line test, rig data) before the team starts making setup decisions on it.

## Key formulas

**Understeer angle (bicycle model)**

$$\delta_{US} = \frac{\delta_{sw}}{i_s} - \frac{57.3\,L\,G_{lat}\,g}{V^2}$$

- \delta_{US}: understeer angle (deg, road wheel)
- \delta_{sw}: steering wheel angle (deg)
- i_s: overall steering ratio (–)
- L: wheelbase (m)
- G_{lat}: lateral acceleration (g)
- V: speed (m/s)

**Suspension roll angle from damper pots**

$$\phi = \arctan\left(\frac{x_L - x_R}{MR\,T}\right)$$

- x_L, x_R: damper displacement left/right (mm)
- MR: damper travel / wheel travel (–), academy convention
- T: track width (mm)

**Roll gradient**

$$k_\phi = \frac{\phi}{G_{lat}}$$

- k_\phi: roll gradient (deg/g)
- \phi: roll angle (deg)
- G_{lat}: lateral acceleration (g)

**Longitudinal slip ratio**

$$SR = \frac{V_{wheel} - V_{ref}}{V_{ref}}$$

- V_{wheel}: wheel surface speed (m/s)
- V_{ref}: reference (vehicle) speed (m/s)

**Total gear ratio**

$$i_{tot} = \frac{2\pi\, r_{roll}\, n_{eng}}{60\, V}$$

- r_{roll}: rolling radius of driven tyre (m)
- n_{eng}: engine speed (rpm)
- V: reference vehicle speed, e.g. undriven wheels (m/s)

**Combined acceleration**

$$G_{comb} = \sqrt{G_{lat}^2 + G_{long}^2}$$


## Trackside takeaways

- Gate derived channels to meaningful conditions: understeer angle blows up at low speed (1/V²), ratio channels such as roll gradient at low g, and kerbs corrupt both.
- Damper-pot roll is suspension roll only; tyre deflection is invisible without ride-height lasers.
- Calibrate rolling radius per tyre set before trusting slip ratio.
- With i_tot = engine speed / wheel speed, a dip below the new gear line on a downshift means insufficient blip, an overshoot above it means over-blip; drift within a gear means slip. Check the sign convention (Segers plots −i_tot).
- Use session constants for setup-dependent numbers and keep one shared, versioned math library.

## By car category

### GT3

Use ABS and TC activity channels with slip ratio; with ABS the long-g vs brake-pressure relationship saturates, so judge braking on deceleration and ABS intervention, not pressure alone.

### Hypercar (LMH/LMDh)

LMH 4WD hybrids (Ferrari 499P, Toyota) drive the front axle above the regulated deployment speed threshold – use GPS/IMU-fused reference speed for slip work. LMDh cars (Porsche 963) are rear-axle hybrids, so averaged front wheels stay a valid reference outside braking.

### FIA F3

With a fixed sensor set, a strong math library (understeer angle, roll gradient, combined g) is where teams differentiate.

### TCR

Front-wheel drive: use the rear wheels for reference speed; slip ratio on the inside front at exit is the key traction metric.

### IndyCar

On ovals, understeer angle must account for banking and the very small steering angles (a few degrees) – absolute offsets from toe and alignment dominate, so compare deltas only.

### LMP2

Roll gradient and roll-angle ratio from damper pots quickly confirm ARB changes; expect small values (high springs/heave stiffness).

## References

- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §2.4 Mathematical Channels
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §7.5 The Understeer Angle
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §9.1–9.2 Measuring Suspension Roll Angle; The Roll Gradient
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §4.2–4.3 Traction and Longitudinal Slip; TCS and Slip Ratios
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §6.2–6.4 Down-shifting; Total Gear Ratio Channel
- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 5 Simplified Steady-State Stability and Control (understeer gradient definitions)
