How to manage non-trivial offset in power meters in TR and elsewhere?

Another variation on the ongoing questions of how to manage cases where power/FTP differences are not trivial…

Just got the Assioma DUO upgrade on sale - primary interest was to improve data precision for CdA field testing on my TT bike.

However, I was surprised to see a 54/46 offset very steady over my first ride with it yesterday. Was expecting more like 52/48 based on the one time I rode with a dual meter on a rental bike a couple years ago.

As a result, the target for threshold felt super easy and I ended up hitting way higher than normal reported power outputs. To complicate this further, my road bike for the foreseeable future will remain L-only, and I have years of good perfectly matched data between the road and TT bike using L-only, and a long history of training where I am very well in tune with my power data enough to know when someting is “off.” Previous testing I did showed the stages/assioma matched within 1W from endurance to sprints - my point is that this is not a case of “oh, nothing can be trusted to more than 5%”- that is a common online cop-out answer when following best practices with good modern power meters :slight_smile:

So - assuming this offset persists, which I will need more riding to see, I can’t just dump all my TT bike data into the same database with an 8% difference when I’m riding both bikes regularly. A difference that large is not something you can just handwave away, but it seems like right now, that is the official solution from TR - “it’ll all just average out”- and basically every road bike workout is too hard, TT workout is too easy, AT will chase its tail based on my feedback and performance against these targets, and it’s completely unnecessary noise.

Other options I could see would be -

Use the Assioma app to scale my dual side power data down by 8% - bad for vanity, but good for data consistency :smile:

I’m not aware of a way to scale a Stages L-side meter up by 8% - but if there is, I’d consider that… this would at least let me not confound what feeds into AT going forward, and have data that is closer to an absolute correct value

Do math in my head so all my outdoor TT workouts aim for 8% over target - but once outdoor workouts are analyzed, that will skew things…

Are there any other 3rd party solutions out there that may conveniently allow some scaling offset to help manage this - esp. if can be done prior to Strava so that everything that syncs from there is then consistent?

Overall, I would be a lot happier if TR would take seriously the need to segregate and offset data sources / positions / indoor/outdoor even if we had to manually assign them into buckets once defined. Without this, for some of us, there will continue to be a lot of unnecessary noise injected into AT that is impossible to accurately compensate. In fact, this could be a very nice differentiating feature if TR was able to provide such organization for things like PRs, etc. I’m not aware of any app that does it today, and at the moment, I’m going to start generating a lot of BS PRs on my TT bike unless I artificially scale down the data.

You mean left-right imbalance?
You can use Goldencheetah or to adjust power values.
But, you need to be sure that,
-Pedals are correct
-Your imbalance stays the same throughout your power curve.

For the first one start with a static weight test. I would say test the stages as well but it does not have that capability afaik.

Second one is tricky. Imbalances are not same all the time. Pedaling style, cleats, tiredness, power level, body position…
They all affect the imbalance. So it is very hard to go back and fix that bad data.

Yea, in this case, L/R imbalance - 46L/54R.

I definitely plan to watch this further and put the pedals on the same bike with my stages for a bit as well to get more data and re-confirm stages and Assioma-L are still lined up… for this first 90 min ride, it was pretty stable throughout except during some short soft pedaling recovery where I wasn’t at all focused - but I only did endurance/tempo/threshold all in saddle / aero. Nothing out of saddle or sprints for example so I will look into that too before making too broad an assumption.

Fit file tools was in the back of my mind… I haven’t tried looking at this in golden cheetah yet. Do you know if there is any way to bulk process offsets? I use that less frequently, but often every few weeks for some bigger picture analysis, and I could imagine it would be useful to say tag a bunch of ride files and apply an offset to them.

This is like a rabbit hole I am regretting having looked into vs. just cruise along with well anchored L-only data… but seeing this explains why I was getting some surprisingly low CdA values (absolute power was under-measured). That dual side power boost may also be my only path to ever measuring a 300W FTP so I’m of course intrigued :smile:

I wish there was just an easy way to segregate datasets by bike, esp. as I already tag this in Strava. Or segregate by the reported power source which I think is already in the file. This would be useful for people with significant power differences on road vs. TT position, different meters, etc.

Get DUOs for your TT bike!

In all seriousness, I don’t have a good answer for you. I’m battling a similar imbalance issue due to IT Band Syndrome on one leg and a long history of favoring it. On course of action is the address the imbalance through physical therapy and single leg exercises and drills.

That’s how I made this mess - did the UNO->DUO upgrade sale they offered :slight_smile: Previously just used L-UNO and L-crank on my bikes… Perhaps the most interesting thing to see will be how much that L/R varies, and whether that in turn ever explains why some days feel harder than others when it would not be so obvious on L-only data.

I’ve had a hard time justifying why I should spend $ on a dual side meter because as you say, it’s not something you can practically “fix.” The application that finally convinced me was to get better data when doing aero field testing - there the extra collection frequency and improved accuracy can both be very relevant when you’re trying to suppress all sorts of noise in the data.

1 Like

One thing you may need to check is if and by how much the imbalance changes depending on fatigue, cadence and intensity. My imbalance at threshold and above it appears to even out whereas at low endurance it can be as high as 60/40. I spent ages trying to correlate L only crank PM’s with my Assiomas and in the end just had to settle for the best approximation I could. I had a Stages but sold it as I couldn’t alter the output as you can with 4iiii’s. I didn’t want to mess with the Assiomas as they are meant to be accurate.