Pedals, cranks or spider: which power meter to buy

Every power meter answers the same question — how hard are you pushing — but they do not all listen in the same place. A pedal measures your foot. A crank arm measures one leg's arm bending. A spider measures both legs together as the torque arrives at the chainrings. A hub measures what survived the chain.

Those are four different quantities, and the differences between them are small but systematic. They explain most of the arguments about which meter "reads high", why your single-sided unit disagrees with a friend's on the same climb, and why a number that changes by 3% when you change the hardware is not necessarily telling you anything about your fitness.

Here is where each type measures, what the published validation studies found when they were compared against laboratory equipment, and which one is actually the right buy for how you ride.

Where each type measures

Power is torque multiplied by angular velocity, and every design is measuring a deflection somewhere in the chain of parts between your foot and the road. The further down that chain you measure, the less of the total you see — and the more of the drivetrain's own losses have already been taken out.

Single-sided doubling, and how wrong it gets

A left-arm crank meter measures one leg and multiplies by two. That is exact only if your legs contribute equally, which they do not. The question is how unequal, and whether the imbalance is stable.

The bilateral asymmetry literature is clear on two things. Asymmetry in cycling is real and common — reviews of leg preference and pedalling asymmetry find it in most riders. And it is not a fixed number: it varies with intensity, with cadence, with fatigue within a ride, and with whether you are seated or standing. Work comparing an instrumented crank system with instrumented pedals found that the two do not always agree about the split, which matters if you are using the figure to make decisions.

That variability is the real problem with doubling, more than the size of the imbalance. If you were 52/48 at every intensity, a single-sided meter would read about 4% high all the time, and 4% of a constant offset is harmless — your zones simply calibrate around it. What actually happens is that the offset changes with how hard you are riding, which is why a comparison of PowerTap, Stages and Garmin Vector against SRM found the single-sided crank showing the largest deviation, with the error growing as intensity rose.

What the validation studies actually found

Manufacturers quote a tolerance — plus or minus 1%, plus or minus 2% — and those numbers are near-useless because they are the manufacturer's own claim under the manufacturer's own conditions. What matters is independent comparison against a laboratory reference. Here is the short version of that literature.

Two honest caveats before you read too much into that table. First, the studies are small — typically one or two units of each system, sometimes a single rider protocol. A finding about one PowerTap in 2004 is not a law of nature about all hub meters. Second, the reference itself is not perfect: several of these studies use SRM as the criterion rather than a dynamometer, which measures agreement with a good power meter rather than agreement with truth.

The defensible summary is that the type of meter matters less than whether it is two-sided, and whether it is zeroed properly. Modern pedals and spiders from established makers all land within a few per cent of a laboratory reference. Single-sided crank meters are the outlier, and the reason is arithmetic rather than engineering.

Temperature, drift and the zero offset

Strain gauges respond to temperature as well as to force. Every power meter compensates for this in firmware, and every power meter compensates imperfectly. The 2004 comparison of SRM and PowerTap made the point plainly: both systems drifted as temperature rose during testing, and the extent of the drift differed between them.

This is the single most common cause of a power meter that "goes wrong". You bring the bike out of a 20 °C house into a 3 °C morning, ride off immediately, and the first twenty minutes read 10 W high or low. Nothing is broken. The gauges have not settled and the zero is stale.

The fix is dull and it works: let the meter reach ambient temperature, then do a manual zero before every ride, unweighted, with the crank still. Do it again if the temperature changes a lot mid-ride — after a long descent, or when the sun comes out on a spring morning. Pedals are more exposed to air temperature than a spider tucked behind a chainring, so if anything they need it more. Our guide on how to calibrate your power meter has the procedure for each system.

Transferability: the question people forget to ask

How many bikes do you own, and do you want power on all of them?

• Pedals move in five minutes with an 8 mm hex key and no measuring. If you have a road bike, a gravel bike and a turbo bike, this is the answer and it is not close. The caveat is that you are committing to one cleat system across all of them.
• Crank arm units move if the bikes share a crank standard, and otherwise do not. In practice most people fit one and leave it.
• Spiders require swapping a chainset, which means the right bottom bracket, the right chainline, and a workshop half-hour. Realistically it lives on one bike.
• Hubs move with a wheel, which is convenient until the day you want the light wheels for a race and the power meter is in the heavy ones.

There is a related question about the indoor trainer. If you already own a smart trainer that measures power, a second power source outdoors is not a duplicate — it is a different measurement in a different environment, and the two will disagree. We cover why in power meter vs smart trainer and, in more detail on the disagreement itself, how accurate is your smart trainer's power.

Left/right, pedal smoothness and the metrics you can ignore

Two-sided systems give you a balance figure. Some also give torque effectiveness and pedal smoothness. These are measured quantities, and they are real, but the evidence that acting on them makes you faster is thin to absent.

The honest position: a persistent, large balance shift — say from a stable 51/49 to 57/43 over a few weeks — is worth investigating, because it can accompany an injury or a bike-fit change. A balance figure that wanders between 49/51 and 52/48 within a ride is normal variation and means nothing. And deliberately trying to pedal "more smoothly" has not been shown to improve efficiency or performance; we go into that in your pedal stroke.

Whatever meter you choose, the number that matters is the one built from your own best efforts over time rather than a single test. Moveee's power profile fits your curve from ride history, which also makes a hardware change visible — swap from single-sided to pedals and you should expect the curve to step, and it is better to see the step than to mistake it for form.

So which should you buy?

• One bike, tightest budget, training for yourself: a single-sided crank arm. Accept that the absolute number is an estimate and that it will not match anyone else's. It will still let you pace, hold intervals and track trends.
• More than one bike, or you want the number to travel: dual-sided pedals. The validation record for the current generation is good and the convenience is genuine.
• One bike, you want the most stable long-term reference: a spider. Protected, temperature-stable, and it measures both legs without assuming anything.
• You race on different wheels: not a hub.

Whichever you pick, change it as rarely as you can. Every hardware change resets the meaning of your training history, and a new FTP established on a new meter is a new baseline rather than a change in fitness. The best power meter is the one you stop thinking about, zero before every ride, and keep for five years.