Strava with two power meters
#1
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
Strava with two power meters
this is a variant on “a man with one watch knows the time….”
i really like strava’s power curve feature, primarily for comparing different types of efforts over long periods of time. for the first year of data, i used only one bike. it’s PM was probably a bit high, but very consistent. now i have a second bike which i ride 90% of the time for the last 3 months, and the PM reads around 10% lower. to add to that, for a while it was functioning one sided, reading another 10% lower since my left leg is weaker. there are a also a couple rides before I zeroed it where it was almost 15% HIGHER.
so now my power curve is showing historical points/peaks that are quite high in some places, and I have quite a few rides which are low. I truly don’t care which is right, I assume the brand new dual sided 4iiii is right once zeroed, but it really doesn’t matter. is there a way to adjust future rides with the +10% bike to keep everything more comparable? is there an easy way to do the same to some of the +15% outliers, or should I just download the GPX, strip the power channel, and reupload?
the PM on the original bike can’t be adjusted, it’s a fundamentally different type than the crank or pedal based ones.
i really like strava’s power curve feature, primarily for comparing different types of efforts over long periods of time. for the first year of data, i used only one bike. it’s PM was probably a bit high, but very consistent. now i have a second bike which i ride 90% of the time for the last 3 months, and the PM reads around 10% lower. to add to that, for a while it was functioning one sided, reading another 10% lower since my left leg is weaker. there are a also a couple rides before I zeroed it where it was almost 15% HIGHER.
so now my power curve is showing historical points/peaks that are quite high in some places, and I have quite a few rides which are low. I truly don’t care which is right, I assume the brand new dual sided 4iiii is right once zeroed, but it really doesn’t matter. is there a way to adjust future rides with the +10% bike to keep everything more comparable? is there an easy way to do the same to some of the +15% outliers, or should I just download the GPX, strip the power channel, and reupload?
the PM on the original bike can’t be adjusted, it’s a fundamentally different type than the crank or pedal based ones.
#2
Senior Member
Join Date: Mar 2010
Location: Seattle, WA
Posts: 23,208
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 18883 Post(s)
Liked 10,646 Times
in
6,054 Posts
I think Garmin units have a scale factor that will do exactly what to want. If you're using a Garmin go into the menu, find your PM under sensors and accessories, and see if you have that option?
Far from ideal, but you can use software to change the power data in your files too.
Far from ideal, but you can use software to change the power data in your files too.
#3
Junior Member
A difference of 10% between two different type power meters is very good. I have a dual sided 4iiii Precision Pro 105 and I use it as a standard to test my power meter app. I wouldn’t worry about normalizing the data in Strava. Just keep adding data and let Strava do its thing.
Chuck
#4
Senior Member
Join Date: Apr 2021
Posts: 4,083
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 2333 Post(s)
Liked 2,097 Times
in
1,314 Posts
I have five different power meters and would estimate the range to be 4%. 10% error is rather large.
I am pretty sure you cannot scale the power values with my Garmin 1030.
I can scale them in GoldenCheetah, it makes my coach feel good. I cannot scale the power in Strava in my account. I think you have to go to Zwift, there is a lot of expertise on tweaking power realtime over there.
I am pretty sure you cannot scale the power values with my Garmin 1030.
I can scale them in GoldenCheetah, it makes my coach feel good. I cannot scale the power in Strava in my account. I think you have to go to Zwift, there is a lot of expertise on tweaking power realtime over there.
#5
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
I think Garmin units have a scale factor that will do exactly what to want. If you're using a Garmin go into the menu, find your PM under sensors and accessories, and see if you have that option?
Far from ideal, but you can use software to change the power data in your files too.
Far from ideal, but you can use software to change the power data in your files too.
#6
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
#7
Senior Member
Join Date: Apr 2021
Posts: 4,083
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 2333 Post(s)
Liked 2,097 Times
in
1,314 Posts
Whether a 10% lower indicated power level is significant depends on how you train or how you use your PM.
Being off by 10% is basically being off by one zone. If I think I am riding in the endurance sone (zone 2), I could actually be doing tempo in zone 3. Worse, a sweet spot session becomes a threshold affair. I do not use a PM this way, so, it would not affect me.
Being off by 10% is basically being off by one zone. If I think I am riding in the endurance sone (zone 2), I could actually be doing tempo in zone 3. Worse, a sweet spot session becomes a threshold affair. I do not use a PM this way, so, it would not affect me.
#8
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
Whether a 10% lower indicated power level is significant depends on how you train or how you use your PM.
Being off by 10% is basically being off by one zone. If I think I am riding in the endurance sone (zone 2), I could actually be doing tempo in zone 3. Worse, a sweet spot session becomes a threshold affair. I do not use a PM this way, so, it would not affect me.
Being off by 10% is basically being off by one zone. If I think I am riding in the endurance sone (zone 2), I could actually be doing tempo in zone 3. Worse, a sweet spot session becomes a threshold affair. I do not use a PM this way, so, it would not affect me.
when i look at the power curve, i’m really just trying to understand if, for example, what felt like a very hard 5 minute push during a generally easy three hour ride was in fact max-output for 5 min, or did it just feel that way because the 2 hours before were at my sustained max, etc. or, was a 30 minute ride after work that felt super strong actually strong or just because of a big tailwind 😂😂
having the historical peak of the curve based on rides that might be 10% high (or vice versa) definitely messes it up, since my range for anything last a few minutes is really narrow.
i *think* i’m getting faster, since i get PRs on segments i’ve done dozens of times pretty often, but that could also be lucky wind…
Likes For mschwett:
#9
Senior Member
Join Date: Apr 2021
Posts: 4,083
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 2333 Post(s)
Liked 2,097 Times
in
1,314 Posts
i like to understand my overall effort for a ride, which is pretty much impossible to gauge by any other method. wind varies so much that even on the same route speed isn’t a great indicator. plus different routes, traffic, and the fact that my HR is +/- 10% depending on time of day for the same effort.
when i look at the power curve, i’m really just trying to understand if, for example, what felt like a very hard 5 minute push during a generally easy three hour ride was in fact max-output for 5 min, or did it just feel that way because the 2 hours before were at my sustained max, etc. or, was a 30 minute ride after work that felt super strong actually strong or just because of a big tailwind 😂😂
having the historical peak of the curve based on rides that might be 10% high (or vice versa) definitely messes it up, since my range for anything last a few minutes is really narrow.
i *think* i’m getting faster, since i get PRs on segments i’ve done dozens of times pretty often, but that could also be lucky wind…
when i look at the power curve, i’m really just trying to understand if, for example, what felt like a very hard 5 minute push during a generally easy three hour ride was in fact max-output for 5 min, or did it just feel that way because the 2 hours before were at my sustained max, etc. or, was a 30 minute ride after work that felt super strong actually strong or just because of a big tailwind 😂😂
having the historical peak of the curve based on rides that might be 10% high (or vice versa) definitely messes it up, since my range for anything last a few minutes is really narrow.
i *think* i’m getting faster, since i get PRs on segments i’ve done dozens of times pretty often, but that could also be lucky wind…
1. Power average and time on a standard 4-5 minute hill. This is sort of a proxy for VO2 max power. I go as hard as I can but only once every 10-14 days.
2. Power at 120 beats per minute, this is approximately my first ventilatory VT1 point. For me, this is the most important metric.
10% off in power has a big effect on TSS and TSB because it is based on IF squared. 0.8 vs 0.7IF is about 15 TSS points per hour. That is huge.
#10
I'm good to go!
Join Date: Jul 2017
Location: Mississippi
Posts: 14,992
Bikes: Tarmac Disc Comp Di2 - 2020
Mentioned: 51 Post(s)
Tagged: 0 Thread(s)
Quoted: 6196 Post(s)
Liked 4,811 Times
in
3,319 Posts
Also, if the bikes give you two different riding positions, then you might just not be getting the best transfer of power out of one bike.
#11
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
Everyone measures improvement differently. I have two basic metrics.
1. Power average and time on a standard 4-5 minute hill. This is sort of a proxy for VO2 max power. I go as hard as I can but only once every 10-14 days.
2. Power at 120 beats per minute, this is approximately my first ventilatory VT1 point. For me, this is the most important metric.
10% off in power has a big effect on TSS and TSB because it is based on IF squared. 0.8 vs 0.7IF is about 15 TSS points per hour. That is huge.
1. Power average and time on a standard 4-5 minute hill. This is sort of a proxy for VO2 max power. I go as hard as I can but only once every 10-14 days.
2. Power at 120 beats per minute, this is approximately my first ventilatory VT1 point. For me, this is the most important metric.
10% off in power has a big effect on TSS and TSB because it is based on IF squared. 0.8 vs 0.7IF is about 15 TSS points per hour. That is huge.
maybe i should get power pedals for the creo, they’d probably be a closer match. but a pain.
#12
Senior Member
Join Date: Mar 2010
Location: Seattle, WA
Posts: 23,208
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 18883 Post(s)
Liked 10,646 Times
in
6,054 Posts
What file formats can you work with?
#13
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
one app exports TCX; the other GPX and FIT. In the few cases where I’ve had data problems before (split rides, mostly) I’ve used the tools on gotoes dot org to fix them. Looked around there a bit but don’t see anything like a “power scaler,” although i could export to CSV, edit the data in excel/sheets, and then go back. But that would be a serious PITA.
#14
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
possible. the geometry is pretty close between the two bikes, and generally i get much faster segments on the newer, lighter bike with the lower power readings. I’m pretty sure the power meter in the creo is high or the 4iii unit is low (seems less likely.)
#15
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
here's an illustration of the issue, although with a little closer inspection a slightly different picture emerges. the chart represents 22 segment attempts of the first part of a nearby hill. although there can be occasional gusts, it's quite sheltered from the wind by a high ridge on the windward side. the grade is relatively steady and averages 10.6%, per google earth. strava thinks it's a slightly different length and quite a bit steeper, but i believe the google earth data to be more reliable. the discrepancy observed with the power meters would actually be far more if strava was right. my speeds are low enough (6mph ish!) here that rolling resistance and aerodynamics should not be a large factor at all. i always climb this in my lowest gear, without stopping, rarely out of the saddle. there's no traffic and the surface is pristine, should be a pretty good test case assuming the map/elevation data is in the ballpark.
my weight per my digital scale has been within a few pounds this whole time - 185 - and the 190 is including shoes, shorts, shirt, keys, helmet. the bike weights are within a pound or so. the maximum variation here might be a couple pounds either way, perhaps 1.5 to 2 percent.
the yellow rows are with the turbo creo power meter, which is generally believed to be high and not totally consistent with more traditional strain based meters.
the reddish row is the first ride on a new bike, half the weight, with a 4iiii precision pro dual meter that i had not yet zero'd. when riding, it was immediately obvious that the numbers were quite a bit higher than the creo, which is not what i would have expected.
the blue rows are the same bike and meter, but zero'd. unfortunately i also mistakenly split the left and right halves (believing "pair" meant pairing the meter to my phone haha) so these are left side only. my left/right balance averages 47/53, so you'd expect these numbers to be lower. but not that much lower. note the power drops from 282 to 245 for a ride with a higher average heart rate that was 5 seconds faster.
the green rows are since i got my head unstuck, repaired the left and right, and zero'd the meter before most every ride.
so what's the punchline? if you plug all the same data into bikecalc (or any of the other online physics based calculators) the only result which matches the observed time to complete the segment is the uncalibrated 4iiii, which is pretty much dead on! the current "correct" configuration deviates from the physics by almost 15% on average, as does the creo. the left only 4iiii is closer to 20%, but that's not surprising if it's off by 15% when considering both legs.
for reference, the segment:
setting aside all the power and junk, for the same heart rate at the same time of day, the lighter bike is around 10 seconds faster (7%) up the hill for the same HR, which happens to be exactly the proportional difference in total weight!
my weight per my digital scale has been within a few pounds this whole time - 185 - and the 190 is including shoes, shorts, shirt, keys, helmet. the bike weights are within a pound or so. the maximum variation here might be a couple pounds either way, perhaps 1.5 to 2 percent.
the yellow rows are with the turbo creo power meter, which is generally believed to be high and not totally consistent with more traditional strain based meters.
the reddish row is the first ride on a new bike, half the weight, with a 4iiii precision pro dual meter that i had not yet zero'd. when riding, it was immediately obvious that the numbers were quite a bit higher than the creo, which is not what i would have expected.
the blue rows are the same bike and meter, but zero'd. unfortunately i also mistakenly split the left and right halves (believing "pair" meant pairing the meter to my phone haha) so these are left side only. my left/right balance averages 47/53, so you'd expect these numbers to be lower. but not that much lower. note the power drops from 282 to 245 for a ride with a higher average heart rate that was 5 seconds faster.
the green rows are since i got my head unstuck, repaired the left and right, and zero'd the meter before most every ride.
so what's the punchline? if you plug all the same data into bikecalc (or any of the other online physics based calculators) the only result which matches the observed time to complete the segment is the uncalibrated 4iiii, which is pretty much dead on! the current "correct" configuration deviates from the physics by almost 15% on average, as does the creo. the left only 4iiii is closer to 20%, but that's not surprising if it's off by 15% when considering both legs.
for reference, the segment:
setting aside all the power and junk, for the same heart rate at the same time of day, the lighter bike is around 10 seconds faster (7%) up the hill for the same HR, which happens to be exactly the proportional difference in total weight!
#16
Senior Member
Join Date: Mar 2010
Location: Seattle, WA
Posts: 23,208
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 18883 Post(s)
Liked 10,646 Times
in
6,054 Posts
What operating system do you use? If there's a way to batch update a bunch of tcx files?
#18
Senior Member
Join Date: Mar 2010
Location: Seattle, WA
Posts: 23,208
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 18883 Post(s)
Liked 10,646 Times
in
6,054 Posts
It will scale the power field for every TCX file in a folder by some % that you choose.
#19
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
I know there are some cumbersome ways to go about it, like using Golden Cheetah or Excel. I spent five minutes on Google and didn't find anything definite. It's a very easy technical problem, so I spent another 15 minutes writing a TCX file power scaler. I can either post the source code for you if you're able to build it (C#) or I can email you an exe file.
It will scale the power field for every TCX file in a folder by some % that you choose.
It will scale the power field for every TCX file in a folder by some % that you choose.
Likes For mschwett:
#20
Randomhead
Join Date: Aug 2008
Location: Happy Valley, Pennsylvania
Posts: 24,399
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 4 Post(s)
Liked 3,699 Times
in
2,519 Posts
Dual sided shimano power meters are known to be inaccurate. The shape of the drive side arm is too complex for power meters, even the one that shimano sells. Every engineer thinks they can design a power meter, and so I think the entire industry is failing to try hard enough. But this is a known issue
#21
Senior Member
Thread Starter
Join Date: May 2021
Location: San Francisco
Posts: 2,039
Bikes: addict, aethos, creo, vanmoof, sirrus, public ...
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 1279 Post(s)
Liked 1,393 Times
in
711 Posts
Dual sided shimano power meters are known to be inaccurate. The shape of the drive side arm is too complex for power meters, even the one that shimano sells. Every engineer thinks they can design a power meter, and so I think the entire industry is failing to try hard enough. But this is a known issue
https://youtu.be/M0tWTUwpt1k
https://youtu.be/M0tWTUwpt1k
it doesn’t appear any more complex in shape than any other crank arm now?
Last edited by mschwett; 05-31-22 at 04:47 PM.
#22
Senior Member
Join Date: Mar 2010
Location: Seattle, WA
Posts: 23,208
Mentioned: 89 Post(s)
Tagged: 0 Thread(s)
Quoted: 18883 Post(s)
Liked 10,646 Times
in
6,054 Posts