# Checking vital signs and building alarm channels

> The first job of the data system is to keep the car running. Check vitals before performance, learn their normal patterns, and build alarm logic that warns the driver without crying wolf.

Area: Data Acquisition & Analysis · Starting level: L1 Foundation · Updated 2026-11-16
Source page: https://race-engineer-academy.vercel.app/lesson/daq-vital-signs-alarms

## L1 · Foundation

**There is no performance in a car parked by the track.** Before you look at a single speed trace, check the vital channels: engine oil pressure and temperature, water temperature, gearbox and differential temperature, fuel pressure, battery voltage, maximum RPM (over-revs), and – increasingly – tyre pressures (TPMS), brake temperatures, clutch pressure and knock.

**Two views:**
1. A **statistics table** (min/max/average per run or lap). Fast, but blind to context: minimum oil pressure of 0 bar might just be the engine stopped in the pits.
2. A **whole-run time plot** of the vitals. Temperatures should rise and stabilise; oil pressure is higher when cold and then settles (e.g., ~5–5.5 bar hot at race RPM in Segers' GT example). Look for trends: a voltage that slowly falls over a run, a temperature still climbing at the end of a stint.

Segers shows a qualifying run where an alternator belt failed: battery voltage started at 10 V (already low – the problem began in the previous run), fuel pressure sagged as the pumps lost voltage, and finally the temperature sensors jumped to ECU default values. The car stopped on track. Two failures: the engineer missed it in the previous run, and the driver got no warning.

## L2 · Practitioner

**Alarm channels.** Use conditional math channels so alarms only trigger when they matter:

$$\text{LowOP} = (P_{oil} < 3.0\ \text{bar})\cdot(n_{eng} > 3000\ \text{rpm})$$

On the dash, add time qualifiers (e.g., condition true for roughly 0.2–1 s, chosen from the duration of harmless transients such as surge; RPM above threshold for > 2 s) so a single noisy sample or a pit-lane idle doesn't trigger it.

**Relationship checks** catch problems statistics miss: oil pressure vs RPM (should follow a smooth curve), oil pressure vs lateral g (surge), fuel pressure vs fuel level (pickup starvation late in a stint), water temperature vs speed (airflow blockage from grass or debris after an off).

**Alarm design for the driver:**
- Prioritise: an alarm can mask others on the dash. A fuel-pressure warning shouldn't hide a low-voltage warning that explains it.
- Few, meaningful alarms with clear actions (pit now / short-shift / lift-and-coast / ignore until told).
- The driver must know what each alarm means, its threshold, and how to acknowledge it, because many alarms cover other dash information.
- The engineer must know thresholds too – race-deciding calls ("stay out, it's a sensor") are made on them.

**Typical thresholds** (engine-specific – always use the manufacturer's values): oil pressure warning ~2.5–3.5 bar above 3000–4000 rpm; water temperature warning 105–115 °C on pressurised systems; battery voltage below ~12.5 V running (on a 14 V charging system) means the alternator is not keeping up.

## L3 · Advanced

**Sensor fault or real fault?** The key trackside decision. Indicators of a *sensor* problem: physically impossible step changes, a value frozen at a round number or ECU default, noise inconsistent with the physical time constant (oil temperature can't move 20 °C in 0.1 s), and no corroboration from related channels. Indicators of a *real* problem: consistent drift, corroborating channels (oil temp up *and* oil pressure down; water temp up *and* power down), and repeatability by track location.

**Use redundancy.** Many cars have two sources for the same quantity: ECU and logger battery voltage, ABS and TC wheel speeds, pedal sensor and throttle body position, tank level and calculated fuel used. Cross-plotting them is the quickest plausibility check.

**Trend across runs.** Make run charts of maximum oil temperature, minimum hot oil pressure at a fixed RPM band, blow-by or crankcase pressure, and gearbox temperature per run. A slow trend over a weekend (bearing wear, oil consumption, radiator blockage) is invisible within one run but obvious across ten.

**Endurance operations.** Engineers on long races set tiered alarms on telemetry: amber (watch, inform driver), red (act now). Pre-agree with the driver and the team what each level means, so the radio call is short.

## Key formulas

**Conditional alarm channel**

$$\text{Alarm} = (P_{oil} < P_{min})\cdot(n_{eng} > n_{min})$$

- P_{oil}: oil pressure (bar)
- P_{min}: threshold, e.g. 3 bar
- n_{min}: RPM gate, e.g. 3000 rpm

## Trackside takeaways

- Vitals first, every run – even when the driver is shouting about handling.
- Statistics tables hide context; always look at the whole-run trace too.
- Gate alarms by RPM/time so they trigger only when meaningful.
- Cross-check redundant channels to separate sensor faults from real faults.
- Run charts of vitals across a weekend reveal slow degradation.

## By car category

### GT3

Manufacturer ECUs have built-in protection strategies; know the factory alarm list and what the driver sees on the display.

### Hypercar (LMH/LMDh)

Hybrid systems add battery/MGU temperatures, isolation faults and high-voltage safety states – alarm handling is safety-critical (car must be declared safe before marshals or mechanics touch it).

### LMP2

Spec engine with manufacturer-defined limits; engine data may be partly visible to the supplier and organiser.

### One-make Cup

Sealed engines and limited channels; your job is mostly early detection and reporting to the series engine supplier.

### Formula 4

Simple dash alarms (oil pressure, water temp); teach the driver exactly what to do when each light comes on.

## References

- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §3.1 Check the Car’s Vital Signs
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §2.8.6 Basic Tips
