Resolving large discrepancy between indoor trainer and crank power measurements


I have been using xert for a few months with an old bike on a kickr core trainer. I recently got hold of a 4iiii crank power meter for outdoor rides on another bike. There seems to be a very large discrepancy between the two power measurements. Yesterday I tried the 4iiii meter on a 2hr ride that was not very strenuous and it showed up with an XSS of 559 (plot attached).

Are there any tricks for aligning these power measurements? Do I need put my road bike on the trainer and use different apps to record both sensors at the same time, and then adjust the scaling factor available on the 4iiii meter? I can’t put the crank meter on my trainer bike so this still wouldn’t account for possibly large drive train differences between the bikes.


That looks pretty wild. Appears you were well above threshold for nearly half an hour straight, so is your thought that the new crank power meter is way overstating?

Did you calibrate the power meter before riding?

At least in Zwift, it’s pretty easy to see what both devices are reading during the power selection screen. Both will be listed and show the current output, I’ve used that method before to compare things.

I am not sure whether the new crank power meter is reading high or the kicr core is reading low. Either way I think that I would be happy with getting relatively consistent measurements regardless of the absolute accuracy.

I did calibrate the crank meter just before the ride.

That actually gave me an idea of how to possibly adjust the crank power measurements into the same ballpark as the trainer. Looking at the power curves for this ride, the “MMP for this activity” is pretty much parallel to the fitness signature curve all the way from 8:00 through 40:00, with a scaling factor of about 0.8. This might be a caveman approach but I suspect that applying this 0.8 factor to the crank meter should significantly improve the correlation between readings. Without the hassle of switching cassettes and shuffling bikes on the trainer. And presumably an outdoor breakthrough attempt should give an even better scaling factor right?

When I first got my crank power meter it came up reading significantly higher than my turbo, but after calibrating both the difference became more subtle (~10W). Particularly with my crank I find I need to calibrate after any big temperature change - it was once occasion when I used on a warm spring day (following a very cold period) that it started returning very high values. I now also have some Rally pedals which seem to be similar in a steady state but respond a bit differently to surges in power.

Chart indicates TP is significantly understated assuming 4iiii power is close to actual watts.
You could play with the MPA Analytics tab and raise TP a few pts at a time and click Refresh until the MPA flatline portion disappears and difficulty graph (shaded portion) looks normal for the effort. That will give you a general idea of what’s watt. :smiley:

Next put your power metered bike on the KICKR Core and experiment with sensor settings in EBC connected to A) your trainer for both control and power or B) trainer control to the KICKR and power to 4iiii.
Powermatch is enabled with a B connection so 4iiii watts will be controlling trainer resistance.
Pick a workout with a mix of target intervals ranging from LTP to TP.
After warm-up swipe up and swap between A and B connections and note any difference in effort required to sustain the same target.
You can also experiment in Slope mode ignoring the interval targets and monitoring watts alone.
For example, generate 150 watts with A versus B connections. Then try 250 watts.
Any significant discrepancy should be obvious.
Core accuracy should be +/- 2% when unit is calibrated and firmware is up to date.

Note: Disable Power Smoothing on the KICKR if currently enabled.

Great, thanks for the detailed what’s watt suggestions. Off to do some experiments…

1 Like

Indievelo had dual power recording. Its free atm, so you just do a ride on there and record both your kickr and power meter and compare.

1 Like

Ah, this may be a big cause of inaccuracy - I did the calibration in my garage after a chilly night and the daytime outside temperature rose to almost 30F warmer than that. I see that 4iiii claims to have built in temperature compensation but in the next sentence they say you should be sure to calibrate before each ride after allowing the meter time to reach ambient temperature…

Thanks, I will give that a try. I have ordered a low budget bottom bracket and crankset that will let me replace the square taper bb on my beaten up old trainer bike with hollowtech ii so that I can run both meters at the same time on the trainer bike for aligning the power measurements

maybe in this thread I missed it but are you using single sided 4iiii or dual? This could also cause a thing like you describe if the right left balance is skewed and you are only reading the left side. Also interesting if you can have both the 4iii and the Kickr reading displayed at the same time that also is a good help option. I have 4iiii dual power meters and they align really well with my trainer though it is not a Kickr.
Also one other thing I did not see in your post or the thread, did you do a spin down calibration on the Kickr Core? From what I understand they do advise to do that as well on the Core.

Thanks for everyone’s suggestions.

Yes, I did do a spin down calibration and on the Kickr trainer and a zero offset on the 4iiii (single sided).

After replacing the crankset on my battered old trainer bike I was able to run both meters at the same time. It turns out that the readings matched very well with just a 0.95 scaling factor applied to the 4iiii, which sounds like a reasonable value for the drive train losses from crank to cassette.

So… what caused the apparently huge difference in power measurement in my original post? When I imported my data to xert the TP seemed lower than I expected and I experimented with ‘recalculate progression’ a few times (without reading the docs to realize that the current signature seeds the new progression) and when it changed a few times I gave up and thought it doesn’t matter as it will just converge on the correct value. And then I went ahead and did a bunch of recommended workouts on auto without realizing that I really needed breakthrough attempts to get an accurate signature. Long story short - TP was about 40W too low. Which explains why the lazy side of me was quite pleased with how much easier xert workouts seemed

Now, with a more accurate signature, doing a similar ride I get the attached results. Much better. And indoor workouts are appropriately challenging.

Thanks for the help!


Massive drop in XSS generated between the two rides.

1 Like