Three Way Power Meter Comparison

My neighbor and cycling mentor Les and I were chatting about indoor cycling with power. He wondered aloud if it was possible to know if his Vector 3 power meter pedals were accurate. I shared that I was in possession of a Drivo smart trainer to complement my Stages left side crank arm power meter. Before the conversation ended we had a basic plan to assemble all three power meters onto my Zwift training platform and see how they stacked up.

For a proper test we would need to eliminate or account for as many variables as possible. I started with some notes that were eventually translated into a testing protocol. The protocol was then translated into a step by step checklist. There were a lot of moving pieces and we hoped to hit our marks with no forgotten steps.

The checklist is appended to the end of this post. It is not an entertaining read but does show our process. I'll review a few of the protocol details before revealing our findings.

Research indicated that the Drivo should see about 5% less power due to drivetrain frictional losses as compared to the pedals and crank arm. In the days before the test I took a few steps to maximize my setup's efficiency. I installed a new chain, cleaned and lubed the jockey pulleys in the rear deraileur, and gave the chain at least 30 minutes of break in before the test.

We did not know how temperature would affect the power meters so we planned to zero at the beginning of the test and again after a warmup phase.

The Stages left arm only power meter was connected to the Zwift app while the other two were paired to Garmin head units. All three recordings would be uploaded to Strava for comparison. The Stages would be considered the baseline as I have the most experience using that unit.

My legs provided the power to be measured. We prescribed five blocks in the protocol. In each block I ramped up to the target wattage and held that wattage for the prescribed number of minutes. Most of our analysis, however, would be against the final 250 watt five minute block.

In addition to a straight power accuracy comparison at 250 watts I also wanted to answer a couple other questions. Do I have a left to right power imbalance? Does efficiency change depending on wattage? How efficient is my particular drive train?

The checklist helped us get through the test without missing any steps. But the test was not hiccup free. There was a mixup between old and new Look style cleats that cost us some time. My checklist included a re-zero after the warmup block but neither of us knew how to re-zero the Vector 3 so they were not re-zeroed. The Vector 3 pedals flashed a low battery warning to the Garmin screen mid test. This is to indicate 20% battery life remaining. We soldiered on without stopping to refresh the batteries.

Strava's 'Analyze' tool in action.


Files were uploaded to Strava and we took our first look at the numbers.

I used the 'analyze' tab in Strava to highlight just the five minute block at 250 watts to gather the following averages:

Stages    Drivo    Vector
248        218        219

Expressed as percentage difference from baseline:
Stages    Drivo    Vector
0        -12.1%    -11.7%

For power meters that are advertised as accurate to within one or two percent this initial result looked pretty poor. But we had more digging to do.

Using the Garmin Connect web application we learned that I do have a significant power imbalance. For the entire ride I recorded a 52%/48% left to right imbalance average. The imbalance grew to 53%/47% for the 250 watt block under analysis.

Single sided crank arm power meters calculate overall power by measuring one leg and doubling the result. This works fine if we assume both legs are sharing the load. Mine legs were not. I corrected the Stages power average mathematically using the 53/47 imbalance reported by the Vector 3 pedals and updated the tables:

Stages
corrected    Drivo    Vector
234            218        219

Expressed as percentage difference from corrected baseline:
Stages
corrected    Drivo    Vector
0            -6.9%    -6.5%

The Drivo, where we expected about 5% loss due to drive train friction, was now in the ball park. But the Vector 3 number was still worrisome.

A couple days passed. My subconscious processed the numbers. My subconscious processed the protocol. A light bulb lit within my mind. We failed to configure the Vector 3 pedals for my longer 175mm cranks. Doh!

My cranks are 175 and I verified the Vectors were configured for 170. I made another mathematical correction and another update to the data:

Stages                Vector
corrected    Drivo    corrected
234            218        225

Expressed as percentage difference from corrected baseline:
Stages                Vector
corrected    Drivo    corrected
0            -6.9%    -3.8%

The final analysis was to investigate efficiency differences at higher wattages. The 400 watt data follows and includes both corrections:

Stages                Vector
corrected    Drivo    corrected
383            363        373

Expressed as percentage difference from corrected baseline:
Stages                Vector
corrected    Drivo    corrected
0            -5.3%    -2.6%

At higher watts the Drivo showed slightly less frictional loss as a percentage. The Vector also shows a small percentage difference compared to 250 watts that could not be related to drive train efficiency changes.

One caveat needs to be mentioned. The Garmin Connect software reports leg imbalance as whole percentages. The 53/47 reported by the software may have been anywhere between an actual 52.5/47.5 and 53.3/46.5. When we are hoping to see power meter accuracies near one percent the rounding error might be more than trivial.

Where do I go from here?

I might choose to improve the protocol based on what we learned and rerun the test. But that's not on my schedule yet. The Vector 3 pedals have returned to their home.

I'm also looking at a choice between two power meters that do not agree for my indoor cycling. The Stages appears to read high due to my leg imbalance. I have years of training data from this device. The Drivo appears to read low due to normal drive train losses. But rear wheel measured power is more applicable to how much power propels me in the real world.

I'd like to thank Les for his inspiration and assistance. Without his help and his pedals this test could never have left the ground.


 

Power meter comparison protocol

The following steps are presented in a more or less chronological order. This document to be printed and used as roadmap during actual testing.


Clean drivo cassette, chain rings, lube jockey wheels, remove jockey wheel seals, install new 9 speed chain at full length, do not augment factory lube, install drivo.

Ride new drive train for at least 30 minutes to break it in. Pair crank as PM in Zwift. Pair Drivo as controllable in Zwift. ANT only. Never BlueTooth. Always use Zwift Fuego Flats with trainer difficulty set to zero. Start Zwift ride.

Pair Drivo as PM to Dwaine's Garmin. Choose ride type "Indoor". Start ride in Garmin.

Record which cog is used for 150W and which cog is used for 250 watts per crank arm power meter. Aim for 90 to 95 cadence.

150w cog: ___
250w cog: ___

Try to zero Stages via phone app while Stages is still paired to Zwift. Record findings: ____________________

Remove pedals from Stages crank arms.

Invite Les, Vector 3 pedals, and Les' Garmin head unit. Install Vector 3 pedals. Record ambient garage temperature.

Garage temperature: ___

Pair and Zero Vector.
Zero message: __________________________

Zero Stages via anonomous phone app.
Zero message: __________________________

Pair Drivo to my Garmin 520.

Start Zwift. Pair Stages to Zwift as power and cadence source. Pair Drivo as controllable. ANT only! Zwift waits at 'ride' screen.

Start ride on both Garmin devices. Choose ride type "Indoor".

Choose route 'Tempus Fugit' in Watopia. Set trainer difficulty to zero. Start Zwift ride.

Start hardware warmup phase.

Get paper and pencil ready to jot down start and end times for each of the following blocks:

Ride at least two minutes manually holding wattage at 150. Use cog ___.

Reduce wattage for at least 30 seconds.

Ride at least five minutes manually holding wattage at 250. Use cog ___.

Stop pedalling.

Zero Vector.
Zero message: __________________________

Zero Stages via anonomous phone app.
Zero message: __________________________

Ride at least two minutes manually holding wattage at 150. Use cog ___.

Reduce wattage for at least 30 seconds.

Ride at least five minutes manually holding wattage at 250. Use cog ___.

Reduce wattage for at least 30 seconds.

Ride at least one minute manually holding wattage at 400.

Stop pedalling.

Stop and save ride on Zwift and both Garmin devices.

Upload rides from Garmin to Dwaine's Strava account. Choose ride type "Workout".

Use 'analyze' tab to find average watts for each of the four power blocks:
Stages:Drivo:Vector
cold 150 watts: ___:___:___
cold 250 watts: ___:___:___
warm and zeroed 150 watts: ___:___:___
warm and zeroed 250 watts: ___:___:___
warm and zeroed 400 watts: ___:___:___

Calculate % high/low relative to Stages for Drivo and Vector for each of the four blocks:
Drivo:Vector
cold 150 watts: ___:___
cold 250 watts: ___:___
warm and zeroed 150 watts: ___:___
warm and zeroed 250 watts: ___:___
warm and zeroed 400 watts: ___:___