Why does my power vary between devices?
You ride with pedal power meters outside, then jump on a smart trainer indoors and see 10–25 watts difference. Which number is right? In most cases, both are. Power varies because devices measure at different points in the drivetrain and because setup details matter. Here is how to make sense of the numbers and keep your FTP and training zones consistent.
Key idea: pick one primary source of truth for testing and pacing, and understand how your other device reads relative to it.
Where power is measured changes the number
Power measured closer to your feet is higher than power measured after drivetrain losses. Clean, well-aligned drivetrains typically lose around 2–5% from pedals to rear hub or trainer. Poor lubrication, contamination, cross-chaining, and high torque can increase that loss.
| Sensor location | What it measures | Typical difference vs pedals | Notes |
|---|---|---|---|
| Pedals / crank / spider | Torque at the crank | Baseline (highest) | Reads before drivetrain loss; good for pacing outdoors |
| Rear hub | Power at the wheel | −2% to −4% | Includes chain and gear losses |
| Direct-drive smart trainer | Power at the cassette | −2% to −5% | Typically a bit lower than pedals, depends on drivetrain condition |
| Wheel-on trainer | Power after tire/roller | −3% to −8%+ | Tire pressure, roller tension, and slip increase variance |
So if your pedals average 250 W and your trainer shows 238 W during the same steady effort, that 5% gap can simply be drivetrain loss plus measurement tolerances.
Why two accurate devices still disagree
- Drivetrain loss changes ride to ride. Dirty chain, dry lube, cross-chaining, worn pulleys, or a misaligned derailleur can add several watts of loss. Clean and lube can recover 5–15 W at the wheel in extreme cases.
- Single-sided meters and leg balance. Left-only meters double your left leg. If your balance is 47/53, the meter under-reads by ~6%. Balance also shifts with intensity, cadence, and fatigue, so error is not constant.
- Zero-offset and temperature. Strain gauges drift as temperature changes. Do a warm-up (5–10 minutes), then perform a zero-offset/ calibration for each device, especially after large temperature swings.
- Crank length setting. Pedal meters need the correct crank length in the app/head unit. The wrong length skews power, particularly at lower cadences.
- Installation and torque. Pedal meters must be installed to spec torque with the correct washers; spider/crank meters need proper bolt torque. Loose chainring bolts or preload issues can affect readings.
- Firmware, smoothing, and recording. Some trainers apply ERG power smoothing or report a smoothed value; head units differ in averaging, rounding, and recording interval. Set 1-second recording and disable smoothing when comparing.
- Battery and signal quality. Low batteries, ANT+/BLE dropouts, or recording on two different head units can introduce gaps and bias averages. Keep batteries fresh and record both sources on the same head unit if possible.
- Cadence source differences. Power meters and trainers may derive cadence differently. Spikes or stalls in cadence data can affect power calculation and smoothing.
How to compare devices correctly
If you want to know the true difference between your devices, use a simple protocol. The goal is to find a consistent ratio (or offset) you can trust.
- Warm up and calibrate. Ride 10–15 minutes. Zero-offset each device. Set both to 1-second recording. Disable ERG power smoothing.
- Record simultaneously. Pair both devices to the same head unit if possible, or to two head units started at the same time. Ensure averages include zeros.
- Do steady steps, seated. Three steps of 6 minutes each at roughly 150 W, 220 W, and 280–300 W, with 3 minutes easy between. Keep cadence steady.
- Compare step averages. For each step, compute the ratio:
ratio = pedal_power / trainer_power
If the ratio is stable (e.g., 1.03–1.05 across steps), you have a consistent percentage difference. If the ratio grows with power, you may have a slope error (contact support). - Optional high-torque check. Do 2–3 x 20–30 seconds at 50–60 rpm around threshold. Some devices diverge more at high torque; note the ratio.
Use the steady-state ratio to interpret your numbers. Example: ratio = 1.04 means your trainer reads ~4% lower than your pedals under steady load.
What to trust for FTP and training zones
- Pick a primary device. Use the same power meter for your FTP tests, outdoor pacing, and key workouts. Consistency beats absolute accuracy.
- Use power match if available. Many platforms can control the trainer based on your bike power meter. This aligns indoor watts to your outdoor source and simplifies training zones.
- Keep separate FTPs if you must. If you cannot power match, maintain two FTP values in your notes or training app: one for pedals (e.g., 280 W), one for trainer (e.g., 268 W). Set workout targets accordingly so the effort feels the same.
- Track trends per device. Compare like with like. Improvements in watts and time-to-exhaustion should be measured on the same device you used previously.
- Race with what you pace with. For events, base pacing on the device you will use on the day.
Quick troubleshooting checklist
- Clean and lube your chain; check chain wear and pulley condition.
- Set correct crank length in pedal power meter and head unit.
- Install pedals to spec torque with required washers; snug chainring and crank bolts.
- Warm up, then zero-offset both devices; repeat after big temperature changes.
- Disable power smoothing; use 1-second recording and include zeros in averages.
- Update firmware; replace or charge batteries.
- Use the same gearing and cadence when comparing; avoid big cross-chains.
- If differences exceed ~7–8% after all checks or the ratio changes with power, contact the manufacturer.
Bottom line: power differences between devices are normal and explainable. Control the variables you can, pick a primary source for FTP and training zones, and use a simple comparison to understand the gap. Once you do, your watts will make sense everywhere you ride.
