My Trainer Said One Thing, My Pedals Another — a 10% ERG Gap and Power Match in WattLog.pro
In ERG mode a smart trainer holds power by its own measurement. WattLog.pro records the ride from your power meter pedals when they are paired. All through October those two meters told me different things: depending on the day and the minute of the ride, from 4% less to 10% more. Power match is the feature that sets the trainer so that the pedals hit the target.
Two meters, two numbers
I ride a KICKR CORE with Favero Assioma PRO pedals. In ERG mode the trainer sets the resistance itself to hold the target power. The catch is that it measures that power its own way: it works it out from flywheel speed and the resistance it applies. The pedals measure the force I put into the cranks. Two different measurements at two different points in the drivetrain.
When a power meter is paired in WattLog.pro, the ride is recorded from it, not from the trainer. That is on purpose. I use the same pedals on the road, so indoor and outdoor numbers can be compared. My ramp test, which gave me an FTP of 193 W, was recorded from the pedals too. Zones, workout targets, TSS and the whole Form tab are worked out from that FTP. Everything rests on the pedal reading.
The trainer doesn't know any of this. It is told "122 W", holds its own 122 W and thinks it is doing a fine job. But the cockpit shows me the pedal power. If the pedals say 130 W, I see a number outside the zone for the whole block, and in ERG there is nothing I can do about it, because I'm not the one setting the power. That is exactly what my workouts looked like for the past few weeks.
Where the gap comes from
- A different point of measurement. Pedals measure power before the chain, a direct-drive trainer after the cassette. In between there are chain and derailleur losses. In lab tests, chain drive efficiency ranged from about 81% to almost 99%, depending on gear, chain tension and power [1]. A few percent "extra" on the pedals is therefore normal.
- The accuracy of both devices. Wahoo quotes ±2% for the KICKR CORE, Favero ±1% for the Assioma. At the extremes that is already 3% with nothing broken. I covered how to read such a mismatch on a chart in "Trainer vs. Pedals — Which to Trust When Reading the Chart?". When power meters were compared against an independent reference, some of them were off by a few percent [2].
- Temperature and warm-up. A trainer estimates power from a model, and the model shifts as bearings, belt and magnets warm up. In my data the gap was always largest in the first minutes of the ride and shrank over the first half hour.
- Calibration. A trainer spindown and a pedal zero offset reduce the gap, but don't remove it. On 30 October I did both, and in a short test ride the KICKR still held 106 W while the Assioma read 9.7% more.
How big it was for me
I went through my October ERG rides with the pedals paired. I rode the same plan ("Endurance & Fat-Max Flush", 70 minutes in Z2) three times, so I also have a like-for-like comparison.
| Ride | Plan | Pedals vs ERG target |
|---|---|---|
| 9 Oct | Endurance & Fat-Max Flush, Z2 | +1.1% average; first 5 min +4.8% |
| 25 Oct | Endurance & Fat-Max Flush, Z2 | +2.4% average; first 5 min +5.2% |
| 30 Oct | short test ride, after a spindown and zero offset | +9.7% |
| 31 Oct | long ride with three tempo blocks | first 2 min +8%; tempo at 160 W: −3.9%, −1.3%, −0.4%; Z2 at 127 W: −0.2%, +2.3%, +2.5% |
| 1 Nov | Endurance & Fat-Max Flush, with power match | −0.2% average over the main block |
The conclusion is not "the trainer reads 10% low". The gap changes from day to day, during a ride and with power. On 31 October, within a single workout, it went from +8% in the warm-up through −4% in the first tempo block to +2.5% in the last Z2 stretch. A fixed correction such as "subtract 5%" would help in some minutes and hurt in others. The correction has to learn as you ride.
How power match works
Power match compares what the pedals show with what the server told the trainer to hold, and works out the ratio. It then sends the trainer the target divided by that ratio. If the pedals read 5% more than the trainer, the trainer is told 5% less, and the pedals land on the target.
- It only learns from steady riding. After every change of target it waits 20 seconds for the trainer and the legs to settle. Samples more than 25% off the command are ignored, because that is a stop, a sprint or a drop in cadence, not a calibration difference. Nothing below 50 W counts either.
- It works from the last minute. The ratio comes from the last 60 samples, and the first correction goes out only after 30.
- It moves in steps, not every second. A new target goes to the trainer only once the difference exceeds 2%. Without that, ordinary noise would send the trainer a new target every few moments.
- It is capped at ±15%. A bigger gap is a fault: a flat battery, a loose pedal, no calibration. That needs fixing, not hiding.
The target in the cockpit doesn't change: the plan says 122 W, the cockpit shows 122 W and that target stays in the saved ride. A small note under the target reads "Power match: trainer set to 116 W", so you can see what the trainer was actually told.
Where it applies and where it doesn't
Power match corrects the ERG target. So it only works where the trainer is given a target in watts. Where there is no target, there is nothing to match.
| Mode | Power match | Why |
|---|---|---|
| Workout from a plan or the library, ERG on | yes | every step has a target in watts that the trainer holds by itself |
| Route converted to an ERG workout | yes | once converted it is an ordinary workout with power targets |
| Ramp test | no | towards the end of the ramp you fade and the pedals read below the target. Power match would take that for a measurement gap and raise the resistance at the worst possible moment. The FTP comes from the pedals anyway (75% of the best minute), so the correction would add little here |
| Critical power (CP) test | partly | only the warm-up, recovery and cool-down. You ride the two all-out efforts (3 and 12 min) without ERG, on your own, so the result does not depend on the trainer |
| Route ride (gradient simulation) | no | the trainer sets resistance from the gradient, and the power is up to you |
| Free Ride | no | no power target, you choose how hard to ride |
| "Free" step in a workout, pause, ERG off | no | at that moment the trainer holds no target |
In modes without a target the trainer doesn't decide the power, so its measurement doesn't affect the workout. The ride is recorded from the pedals anyway.
All the work happens on the server; the trainer simply receives different watts. That is why power match works over every connection: in the browser, in the desktop app and on the phone, with no app update. It is on by default and kicks in by itself when a power meter is paired. You can turn it off under Settings → Trainer → Power match.
The first ride with power match
On 1 November I rode the same plan for the third time: a 10-minute warm-up from 46 to 63% FTP, 70 minutes at 63% FTP (122 W) and a 10-minute cool-down.
The first minute shows why the feature exists. Before power match had enough data, the trainer held 90 W while the pedals averaged 109 W, 22% more. The server sent the first correction after a minute and a half. From then on, the trainer's commands sat below the plan.
At the start of the main block the trainer was told 112 W, 8% below the target. Early on, the trainer, the drivetrain and I were all warming up: my cadence rose from 83 rpm in the first minutes to 96 rpm and then stayed there. The gap shrank and the correction with it: for the rest of the block the commands moved between 114 and 119 W. The pedals barely moved at all. The 70-minute average was 121.8 W, and every 5-minute slice fell between 120.7 and 122.7 W. Heart rate in 5-minute slices stayed between 135 and 137 bpm, with no drift to the end of the block.
How it felt. Earlier workouts felt harder than they should have, and the power in the cockpit often refused to sit in the zone. Today was calmer: the number on the screen stayed where the target was.
What this doesn't prove
- We don't know which meter is right. Power match doesn't make the power "true". It makes the trainer agree with the meter I record rides from and calculate FTP from. If the pedals are off, they are off consistently, and the zones are calculated from them anyway, so the workout lands on the intended intensity.
- It is one ride. I rode the same plan before with a different main-block target (63→68% and 63→70% FTP, and on 9 October at an FTP of 184), so heart rate from those rides can't be compared directly with today's.
- The measurement gap doesn't explain everything. In two rides of the same plan the pedals read 1–2.5% above the target on average. That is too little to blame the whole "too hard" feeling on the trainer. Bigger gaps showed up at the start of rides and on single days.
- It can't tell fatigue from a measurement gap. If you fade and the pedals read below the target, power match takes it for a difference between the meters and raises the resistance. In a Z2 workout that practically never happens, but in a ramp test, which by design ends in exhaustion, it does. I ran my October ramp test through the estimator: on the last step it would have lifted the trainer from 252 to 256 W 15 seconds before the end. That is why power match is off in the ramp test.
- The first minute is uncorrected. Power match needs steady riding to learn anything. It starts from zero on every ride, because the gap is different every day.
- The correction still jumps a little. In the main block the trainer got a new target 14 times in 70 minutes, usually 2–3 W up or down. It doesn't show on the pedals, but the threshold could be tuned.
What I take away from it
All through October ERG held power by the trainer's reading, while I watched the pedal power and wondered why it wouldn't sit in the zone. The gap was small on some days and big on others, always largest at the start. Power match doesn't guess which meter is right. It sets the trainer so that the meter I take my FTP and zones from holds the target. On the first ride, 70 minutes at 122 W came out as 121.8 W on the pedals.
If you train with power pedals or a power meter crank on an ERG trainer, it is worth checking whether both devices say the same thing. My data says they don't. More on where zones come from in the article on the ramp test and the 3+1 plan and in the guide to power zones.
Sources
- Spicer JB, Richardson CJK, Ehrlich MJ, Bernstein JR, Fukuda M, Terada M. Effects of frictional loss on bicycle chain drive efficiency. J Mech Des. 2001;123(4):598–605.
- Maier T, Schmid L, Müller B, Steiner T, Wehrlin JP. Accuracy of cycling power meters against a mathematical model of treadmill cycling. Int J Sports Med. 2017;38(6):456–461.
Train smarter with WattLog.pro
WattLog.pro collects data from your trainer and shows what's really happening with your fitness.
Try WattLog.pro for free →