Understanding power accuracy in smart trainers

Where the watts come from

Power is torque multiplied by angular velocity. A trainer knows the flywheel speed precisely; the question is how it knows the torque.

• Direct measurement uses strain gauges or an optical torque sensor in the drive path. The trainer measures the force the chain applies.
• Estimation uses a model: the trainer knows how much braking it commands and how fast the flywheel spins, and it infers your input from the electrical and mechanical losses it has characterised at the factory. Many well-regarded direct-drive trainers work this way and are accurate once calibrated.
• Wheel-on estimation adds the tyre: the roller speed is known, but tyre pressure and clamp tension change how much power is lost between the pedals and the roller. This is why wheel-on units ask for a spin-down calibration and quote wider tolerances.

Manufacturers rarely publish which method a model uses; the accuracy claim is what you get.

What ±2 % means

A ±2 % claim says that, under the manufacturer’s test conditions, the reported power is within 2 % of the true value: 196 W to 204 W when you ride at 200 W. It does not say:

• how accuracy changes with temperature during a 90-minute session,
• how it behaves at very low or very high cadence,
• how fast the reading responds to a change in effort,
• whether the claim is tested at 150 W or at 1,000 W.

Two trainers with the same claim can behave quite differently in practice, and independent reviewers compare them against a reference power meter precisely for that reason. A Cadexa measurement, when we publish one, always sits in its own column next to the manufacturer claim, never merged with it.

Accuracy versus consistency

For training, consistency often matters more. If your trainer always reads 3 % low, every interval, every week, your FTP and your zones are internally consistent and progress is still visible. A trainer that drifts 4 % between a cold start and minute forty makes the same workout feel different for reasons that have nothing to do with you.

Practical checks:

• Warm the trainer up for ten minutes before a test or a calibration.
• Calibrate wheel-on trainers every few rides and after changing tyre pressure.
• If you own a power meter, ride a steady 20 minutes recording both; a stable offset is fine, a wandering one is not.

Power matching

Many apps can read power from a crank or pedal power meter while still controlling the trainer in ERG mode. The trainer then adjusts resistance until the power meter reads the target. This removes trainer accuracy from the equation and keeps indoor and outdoor numbers on the same scale. The protocol detail: the app subscribes to the Cycling Power Service of the meter and writes targets to the FTMS control point of the trainer.

What to look for when buying

• A claim of ±1 % to ±2 % from a direct-drive unit is the current norm; treat ±2.5 % to ±3 % wheel-on claims as honest rather than inferior.
• Automatic or temperature-compensated calibration reduces the human factor.
• If accuracy is central to your training, the independent measurements linked from a product’s specification table are worth more than the claim.