AI 速览
- Crank-arm meters are the cheap entry; spider and pedal meters add true dual-sided data at different price points.
- Portability and consistency usually matter more than the headline accuracy spec for everyday training.
- True dual-sided meters reveal L/R balance and pedalling dynamics that single-sided units simply double.
- Hub and bottom-bracket meters have niche appeal but have largely been displaced by pedal and spider options.
/ 01
Why Power Meter Architecture Matters
A power meter is one of the few training tools whose readings feed directly into your training zones, race pacing, and load tracking. Small errors compound across a season. The architecture you pick — where in the drivetrain the strain gauges sit — determines not just accuracy but whether you can swap it between bikes, see left-right balance, and measure pedalling smoothness.
All power meters estimate torque from strain-gauge deflection, multiplied by angular velocity to get watts. The differences are in the location of measurement, the number of gauges, and how temperature drift is compensated. There is no single best answer — only the right trade-off for how you ride and how many bikes you own.
/ 02
The Four Main Architectures
Crank-arm meters (Stages, 4iiii, Shimano dual-sided) attach to the left crank. They are single-sided by default, cheap, and easy to swap. Spider units like power2max and Quarq measure total power at the chainrings — accurate, dual-sided (calculated), but tied to a specific crankset.
Pedal-based meters (Garmin Rally, Favero Assioma) put the gauges in the pedal body. They give you true dual-sided data (each pedal measures independently), portability between bikes in minutes, and the cleanest L/R balance. The trade-off is fragility — pedals hit the ground in crashes — and slightly higher stack height.
Hub-based meters (the older PowerTap G3 pattern) measure torque at the rear hub after drivetrain losses. Robust and weatherproof, but single-sided, tied to a specific wheel build, and they miss the L/R story entirely. Mostly out of favour today.
Bottom-bracket meters (Rotor INpower, older Ergomo) hide inside the axle or BB shell. Clean aesthetic, protected from crashes, but harder to service and often single-sided.
/ 03
Power Meter Architectures Compared
Where each type of power meter sits on the accuracy and portability axes
| Type | Sides | Portability | Typical accuracy |
|---|---|---|---|
| Crank arm (Stages, 4iiii) | Single | Moderate (per crank) | ±1.5–2.0% |
| Spider (Quarq, power2max) | Dual (calculated) | Low (per crankset) | ±1.0–1.5% |
| Pedal (Favero, Rally) | True dual | High (any bike) | ±1.0% |
| Hub (PowerTap) | Single | Very low (per wheel) | ±1.5% |
| Bottom bracket | Single or dual | Low | ±1.5–2.0% |
/ 04
Accuracy vs Portability
A ±1% pedal meter that you trust and move between bikes will beat a ±0.5% spider meter that sits on a winter bike you never ride. Consistency matters more than headline accuracy for training — you set zones from a 20-minute effort, then track load against those zones week after week. A meter that drifts between bikes undetectably is worse than one that drifts but is consistent.
Calibration rituals are non-negotiable regardless of type: zero-offset before every ride, full calibration at the temperature extremes you ride in, and clean torque on the chainring bolts or pedal threads so the strain gauges do not see phantom loads.
/ 05
Dual-Sided Data and Pedalling Dynamics
True dual-sided meters (pedals, some spider-based duals) measure each leg independently and report left/right balance, torque effectiveness, and pedalling smoothness. These are interesting numbers but rarely change training. Most coaches will tell you to watch trends, not absolutes — a 48/52 balance is normal, and a sudden shift to 40/60 may flag a biomechanics issue worth investigating.
Single-sided meters assume both legs produce equal power and double the left side. If you have a real asymmetry, your absolute watts will be off by that asymmetry percentage. For pacing and load this is usually fine; for sprint analysis it is not.
/ 来源