Key takeaways
A dyno never measures power directly. It measures torque and rotational speed, and power is calculated from the two. Different dyno types measure at different points in the driveline, apply different test methods and use different correction standards, so two honest dynos can give two different figures for the same car. A figure is only meaningful when you know how it was measured.
What a dyno actually measures#
Every dyno measures two quantities: torque, the twisting force at a rotating shaft, and rotational speed. Power is then calculated, because power is torque multiplied by rotational speed. In metric units, power in kW is torque in newton metres multiplied by rpm, divided by about 9,549. Everything else on the screen, including the horsepower figure, is derived from that. A hub dyno is coupled directly to the car's wheel hubs. The wheels and tyres are removed and a drive adaptor is bolted to each hub, so the dyno's absorber units resist the hubs and measure the torque passing through them. Because no tyre contacts a roller, there is no tyre slip and no rolling resistance from the tyre in the measurement. The reading includes the losses in the gearbox, differential and driveshafts, because the power has already passed through them.
Where power is measured, and how load is applied#
The first reason figures differ is the measuring point. Power drops at each stage between the crankshaft and the road, so the same engine reads progressively lower the further down the driveline you measure. Driveline loss is not a fixed percentage. It varies with the gearbox, the differential type, whether the car is front, rear or four-wheel drive, the gear used and the oil temperature. Quoting a single "typical" loss to convert a hub or roller figure back to a crank figure is a rough estimate at best.
| Dyno type | Measured at | Includes driveline losses | Main caveats |
|---|---|---|---|
| Engine dyno | Crankshaft or flywheel | No | Engine must be removed or run on a test bed; not the same as in-car conditions |
| Hub dyno | Wheel hubs, tyres removed | Yes | Needs hub adaptors fitted; gearbox, diff and driveshaft losses still included |
| Roller dyno | Tyre contact patch on a roller | Yes, plus tyre effects | Tyre slip, tyre temperature, pressure and strapping all influence the result |
The second difference is how the machine applies load. An inertia dyno uses a heavy drum of known inertia. The car accelerates the drum, and power is calculated from how quickly the drum's speed rises. It is simple and quick, but the sweep rate is set by the drum's inertia and the car's own output, so the operator cannot choose it. A load-controlled dyno has an absorber, commonly an eddy-current or water-brake unit, that applies a controlled resistance. That allows two test styles. In a sweep (or ramp) test the engine accelerates through its rev range at a chosen rate while the absorber holds a defined load. In a steady-state test the dyno holds a fixed rpm and load for several seconds at each point, which lets the system stabilise before logging. Steady-state testing is slower and puts more heat into the car, but it is how a calibrator checks fuelling and ignition at specific points on the map. Sweep rate matters. A very fast sweep can under-read a turbocharged engine because boost and temperatures have not stabilised, while a very slow sweep adds heat soak. A result is only comparable with another if the sweep rate and method are comparable.
Correction factors and repeatability#
Engines make power from the oxygen they ingest, and the amount of oxygen in a litre of air changes with temperature, pressure and humidity. As a rough guide, near normal ambient temperatures, air density falls by about 1% for every 3 degrees Celsius of temperature rise. A car tested on a cold, high-pressure morning will read higher than the same car on a hot, humid afternoon. To make results comparable, dyno software applies a correction factor that scales the measured figure to a reference set of atmospheric conditions. Several standards exist, including SAE and DIN procedures, and they do not use identical reference conditions or formulae. Some correct for humidity and some do not. A corrected figure from one standard cannot be compared directly with a corrected figure from another, and an uncorrected figure is simply what happened on the day. Note that bhp and PS are different units. Converting between them without noticing is an easy way to gain or lose roughly 1.4% on paper. As a reference point, the International Standard Atmosphere is defined as 15 °C and 1013.25 hPa at sea level, and one mechanical horsepower equals about 745.7 W, against about 735.5 W for a metric horsepower (PS).
For tuning work, the most valuable property of a dyno is not that it reads "high" or "low" but that it reads the same way every time. A repeatable setup allows a before-and-after comparison on the same day, which shows what a calibration change actually did. Repeatability depends on controlling the variables that move the result:
- Intake air, coolant and oil temperature at the start of each run, because heat soak reduces power.
- Consistent cooling airflow over the radiator and intercooler, since road speed airflow is absent on a dyno.
- The same gear, sweep rate and start and end rpm for every run.
- The same tyre pressures and strapping on a roller, or secure hub adaptors on a hub dyno.
- The same fuel, and the same correction standard and smoothing settings in the software.
Tip
Questions worth asking about any figure
Ask where it was measured (crank, hub or roller), whether it is corrected and to which standard, which gear and sweep method were used, and whether it is a peak or an average of several runs. If the answers are not available, treat the number as a rough indication only.
What this means in practice#
Do not compare your figure with someone else's unless the measuring point, correction standard and test method match. Use a single dyno, under controlled conditions, to compare your own car before and after a change, and judge the whole curve instead of the peak. A controlled test environment, such as a ventilated dyno test cell, is useful for exactly this reason: it keeps the conditions steady so that the comparison is fair.
