Shimano power meter inaccuracy, and assymetry

Discuss light weight issues concerning road bikes & parts.
TobinHatesYou
Posts: 4300
Joined: Mon Jul 24, 2017 12:02 pm

by TobinHatesYou

Yes, I am talking about any PM that is retrofitted onto an asymmetric Shimano DS crankarm. I wouldn’t say the problem rests with Shimano, just that their crankarm/integrated spider does not lend itself to the current state of the art for crankarm based PMs.

Small percentage points matter because we are trying to enumerate performance and other power meters are accurate (and in many cases also cheaper.)

RocketRacing
Posts: 934
Joined: Thu May 10, 2018 2:43 am

by RocketRacing

Sockman wrote:
Thu Oct 10, 2019 3:50 pm
I own a Shimano PM. It reports the same power numbers as my Quarq, my old P2M & my past two smart trainers. Where exactly does the innacuracy come in?
THe last gen shimano PM's were ok, the current gen are the issue (due to the highly assmetric crank). Last gen had only a mild assymetry, so the issues resulting in dynamic calibration problems were only minimal. What PM/crank do yo have? it may very well be fine.

by Weenie


RocketRacing
Posts: 934
Joined: Thu May 10, 2018 2:43 am

by RocketRacing

TheRich wrote:
Thu Oct 10, 2019 9:46 pm
Hexsense wrote:
Thu Oct 10, 2019 8:28 pm
They need dynamic compensation. To do so, they need to change offset values based on crank angle multiple time a circle.
And even so, it'll only accurate if your pedaling circle is "standard" as there can be some left/right interference, left over momentum etc which isn't equalize to situation they measure and collect the data to create dynamic offset values.

If you get accurate value, good then. But why risking accuracy issue when you can buy power meter that doesn't have to do all these assumption to be accurate.
There IS a way to check the accuracy of a power meter, but it doesn't involve using it as a power meter.

Apparently, almost all the crank based meters that GPL tested are just as wrong as Shimano, but somehow the problem is with Shimano.

Kinda baffled why a 1-2% is somehow critical. Oh, you were actually doing 300w and it read 306w, who cares? Worried about your TSS being off by 1 or 2 points? Is there some huge difference in effort that this critical error hides?

Edit: If you want a legitimate gripe, talk about the price.
Numbers can be off far more, it depends on where you are in the stroke.

Here is my understanding of it:
- there is a theoretical "truth" for actual power of a rider.
- power meters are like weigh scales. THey measure strain with strain (force) gauges. they then often have a correction factor to scale the strain with "the truth", and factor in cadence to give you an actual "power" (force over time = power. the "truth").
- In short, pedal based meters measure "high" as there is no drivetrain loss in these measurements. Hub pased PM probably are the most realistic for power at the wheels, as drivetrain losses have already taken effect. But a cruddy chain will lower these numbers seen at a PM hub. THus they would tend to measure low of the "truth" that your legs make. Now, there is no standard, so there is nothign to stop a power meter from having a correction factor to "correct" for such losses. Who cares, as long as the numbers are consistent. Consistency is the key.
- Where it goes to pot is on trainers. Why? BEcause watts seen effects your performance in zwift and others. The Tacx Neo is felt to be a gold standard for power measurement because it uses magnets of standard force/resistance and actually measures the power you produce (and it matches that power with electro-magnetic resistance).
- good power meters will factor in things like temperature, as that can factor into inaccuracies.
- ideally, a power meter should not require calibration.... or rather should self calibrate. ideally frequently. Think cool am ride turning to hot am ride. power numbers could in theory deviate quite a lot.

So a good PM will self calibrate, and compensate for things like temperature. And where power is measured can effect how close a number is to "the truth". PEdal based power meters are the closest thing to the "truth" as there is little to cause drivetrain losses. Hub based meters and direct drive trainers are closest to the "truth" that the road sees, as drivetrain losses have taken place. It is like hp at the crank, or at the wheels of a car. The crank numbers are also higher as it is the power the engine makes without drivetrain losses. That can be an easy 15% in a car. Power at the wheels is all that matters to determine how fast a car will be.

So what does this have to do with shimano?

Well, when you have an assymetric crank, the force that the strain gauge sees at any one point in rotation actually changes. Because there is no structural symmetry across the 360 degrees of rotation, the numbers seen by the strain gauge is always changing vs the "truth". Ok, so how about correction factors? Sure, why not... but that correction factor changes for every 1 degree in crank rotation. Oh, and by the way, you now have a power estimator. So now we have a power estimator with a very complex set of correction factors to estimate power. CLose enough? Not really... because now the symmetric crank on the other side factors in. this would not be so bad, except that it will need it's own set of unique correction factors, and it all goes to pot because you now need correction factors that take into account on symmetric side, and one assymetric side and forces in both directions as they counter one another. MAthmatical assumptions are made. Compromises are made as there can not be a supercomputer in every head unit. Numbers are rounded, simplified formulas are used to create simpler correction factors that are "close enough."

The current gen shimano cranks all have this issue. The last gen shimano cranks were ok (assymetry was minimal). It is not an issue with the strain gauges. They likely meet the accuracy claims that shimano makes The issue is the assymetry resulting in the measurments that the gauges see, do not replicate reality. correction factors are made to fudge the numbers "right". You know have a power estimator. Those correction factors may work for 95% of situations, but the issue is that when looked at closely, and compared to industry standards (magnetic direct drive trainers, hub pms, pedal pms, spider pms...) they showed more error than the others, and in specific scenarios.

The power seen in 5 power meters will all be different. That is ok. where is is measured, makes a difference in the actual number. As do correcton factors. What seperates the good power meter from the bad is reacting to dynamic changes like temperature, cadence, etc. Also, most importantly, they should all respond the same to an input. The power tracings should all mirror one another like a shadow boxer. The outlier is likely the issue. Does it spike? does it dip? Does it lag? when four all trace the same, and one does not... we can only conclude the off man out is the issue. The current gen shimano crank based PMS were the odd man out. You can't argue with good direct drive trainers measuring actual power, or with good pedal based meters that have no assymetry or drivetrain loss.

Chainring PMs work, crank based PM's work, hub work. etc. But because they operate in a symmetrical, or near symmetrical environment, their strain gauges mirror reality. There are no fudge factors, or at least less assumptions are made. Maybe just scaling. the numbers are closer to the "truth." or at least more consistent with the truth.

And lets not confuse accurancy and consistency.

The current Schimano power meters are less accurate than advertised. End of story. If I am going to pay for a power meter, I want it to be a power meter, not a power estimator. Strava is pretty good at estimating power.

OR... accept the limitations. Shimano PMs are fine for most. Even the pros.

Like the current gen shimano PMs, I also have a "power estimator" on my climbing bike (single sided vector 3s). I realize the limitations, and use my trainer (neo 2) for FTP comparison. I only compare my bike PM to itself, and only with a big grain of salt. I only use it to pace efforts. 5% error is fine for that. That is even fine for pros. My garmin defaulted to the wrong crank length and I was getting an error in the 5% range for much of the summer.

I have a quark on order. When I get it, I will run it parallel to my trainer, and firure out the correcon factor. I will use that to compare efforts on the trainer to outside with the quarq. That will measure my true changes in ftp. Do I need that accuracy? Probably not. But FTP is motivating to me. numbers are motivatign to me, even small ones. And my 600$ is going towards an accurate PM, because why would I pay the same money for a less accurate one?

And if you post on the WW forum, you care about the small numbers too.

Sockman
Posts: 32
Joined: Mon Jun 17, 2019 1:58 pm

by Sockman

So a good PM will self calibrate, and compensate for things like temperature. And where power is measured can effect how close a number is to "the truth". PEdal based power meters are the closest thing to the "truth" as there is little to cause drivetrain losses. Hub based meters and direct drive trainers are closest to the "truth" that the road sees, as drivetrain losses have taken place. It is like hp at the crank, or at the wheels of a car. The crank numbers are also higher as it is the power the engine makes without drivetrain losses. That can be an easy 15% in a car. Power at the wheels is all that matters to determine how fast a car will be.
Sure all fair points, but you need to keep in mind that 150w measured at the pedal, which is equal to 145w (or whatever) at the crank, and 140w at the hub (for example) is still equal to the same effort from the rider. If you measure FTP to tell your mates how big your....number is, then ride with garmin pedals. If you want to train, buy any modern power meter.

FWIW I have a shimano pm (r9100-p) on one bike and a quark on another bike.

TobinHatesYou
Posts: 4300
Joined: Mon Jul 24, 2017 12:02 pm

by TobinHatesYou

Sockman wrote:
Fri Oct 11, 2019 3:09 pm

Sure all fair points, but you need to keep in mind that 150w measured at the pedal, which is equal to 145w (or whatever) at the crank, and 140w at the hub (for example) is still equal to the same effort from the rider. If you measure FTP to tell your mates how big your....number is, then ride with garmin pedals. If you want to train, buy any modern power meter.

Everything is fairly rigid between the pedal body and crank except the spindle, and you're not losing 5W at 150W to heat at the spindle/bearing. You're losing a fraction of a what at most. Even at the extremes of pedals vs hub, your drivetrain is ~98% efficient unless crosschained or dirty.

Sockman
Posts: 32
Joined: Mon Jun 17, 2019 1:58 pm

by Sockman

I think you kinda missed my point/It was poorly made. Who cares about a few watts variance between training devices? If you are serious about training, you most likely have one type of powermeter. Consistencty is key, not the absolute highest level of accuracy!

Sockman
Posts: 32
Joined: Mon Jun 17, 2019 1:58 pm

by Sockman

I think you kinda missed my point/It was poorly made. Who cares about a few watts variance between training devices? If you are serious about training, you most likely have one type of powermeter. Consistencty is key, not the absolute highest level of accuracy!

TobinHatesYou
Posts: 4300
Joined: Mon Jul 24, 2017 12:02 pm

by TobinHatesYou

It’s a device you expect to be accurate and there are plenty of other options that are. It’s true there probably are people who won’t care that their Shimano crank based PMs aren’t accurate, but for so many others it’s going to be like *record scratch* and they end up choosing a Quarq, P2M, Assiomas or whatever.

TheRich
Posts: 469
Joined: Tue Jan 01, 2019 1:36 am

by TheRich

TobinHatesYou wrote:
Fri Oct 11, 2019 9:38 pm
It’s a device you expect to be accurate and there are plenty of other options that are. It’s true there probably are people who won’t care that their Shimano crank based PMs aren’t accurate, but for so many others it’s going to be like *record scratch* and they end up choosing a Quarq, P2M, Assiomas or whatever.
It's really not about "accuracy" though.

If I'm Mr. Analretentive and have three power meters, I have three FTPs. It has to be done because they all read slightly different.

But in reality I just want consistency, even if I DID have three PMs, a percent here or there doesn't matter because that collection of PMs is most likely nore consistent than I am. There is obvious value to power meters, but to act as if not meeting some gold standard (that nothing else meets) will make or break your day is just silly.

The Shimano PM is expensive and has a funky charging cable. THOSE are actual gripes. That it doesn't agree with any other PM doesn't really stand out because none of them agree.

TobinHatesYou
Posts: 4300
Joined: Mon Jul 24, 2017 12:02 pm

by TobinHatesYou

But it *isn't* consistent and that's why its accuracy issues can't even be fixed with math/firmware. The measured power will change based on where the power is applied...and that changes with incline, fatigue and other variables. This isn't simply an AV/cadence resolution thing either. Zero isn't zero consistently around the pedal stroke. You will probably see variances of 5-10% at times depending on your ride...compared with itself.

User avatar
gplama
Posts: 10
Joined: Thu Jan 22, 2009 10:21 am
Contact:

by gplama

eins4eins wrote:
Mon Oct 07, 2019 8:00 am
I'm a happy quarq user, but wonder if this whole inaccuracy discussion isn't overrated. Looks like someone wanting some extra clicks for his youtube channel.
Ineos, Bora, FDJ, Jumbo - all using those cranks. If its good enough for them, how big of a problem could there be?
Almost correct. Although the report was put up on my website which isn't YouTube and isn't monetized. The report and stats are better presented on the web, so I forked out $ to finally spin up that site for the report. If I wanted extra clicks this isn't it. I have a good grip on what clickbait is. This isn't it.

More back story: I spent months and months trying to figure out what I was seeing with Shimano based L/R cranks .. which was the right reading low with my testing protocol. None of the power meter companies could explain it, or more accuractly, admit to the issues I eventually discovered. When it was confirmed by a guy with his name on the patent of the very design, I was pretty pissed to be honest. Not at him, at the industry. I have a lot better things to do with my time than mess about with shit power meters.

I typically dismiss what pro teams use. They're either paid to use, paid to promote, under contract never to criticize, they're pro cyclists... not amatures overly obsessed with details such as "why is this power meter a few % wonky".

Back to why pro teams don't care? They'll still win without power meters. They're non essential equipment. (Non essential as in if they stop working mid-race, it doesn't matter... not to be confused with the benefits of training with power).

Anyhow, much like the debate around power meters not matching or being 'accurate', like it or not the result of people obsessing over this eventually leads to better products on the market. InPeak is a perfect example. They sent me a total potato power meter, then 1/2 potato, they now have one of the best single sided meters I've tested. If only Shimano didn't have their heads so far up their own arses, they'd also sort out their power potato. I expect their next power meter won't be such an embarrasement to an otherwise brilliant product line (I love my Di2).

More ramblings...The term 'influencer' is skewed. Every day I'm asked, pressured, and told what to say and post with people or companies with their own agendas. Those are the real influencers. For me, I'd rather influence companies to make better products.

Dismissing people/posts that point out the finer details gives the industry a free pass to keep selling shit. This is the problem.

RedbullFiXX
Posts: 129
Joined: Sat Aug 05, 2017 6:13 am

by RedbullFiXX

Just started using Favero Assioma duo's, my first foray into power data
Thank you for the sweet deal Clever Training
Very nice product :thumbup:

alexroseinnes
Posts: 50
Joined: Tue May 13, 2008 4:36 am

by alexroseinnes

Excellent post Shane, thanks.

gurk700
Posts: 330
Joined: Fri Jan 20, 2017 7:40 pm

by gurk700

gplama wrote:
Mon Oct 28, 2019 9:54 pm
eins4eins wrote:
Mon Oct 07, 2019 8:00 am
I'm a happy quarq user, but wonder if this whole inaccuracy discussion isn't overrated. Looks like someone wanting some extra clicks for his youtube channel.
Ineos, Bora, FDJ, Jumbo - all using those cranks. If its good enough for them, how big of a problem could there be?
Almost correct. Although the report was put up on my website which isn't YouTube and isn't monetized. The report and stats are better presented on the web, so I forked out $ to finally spin up that site for the report. If I wanted extra clicks this isn't it. I have a good grip on what clickbait is. This isn't it.

More back story: I spent months and months trying to figure out what I was seeing with Shimano based L/R cranks .. which was the right reading low with my testing protocol. None of the power meter companies could explain it, or more accuractly, admit to the issues I eventually discovered. When it was confirmed by a guy with his name on the patent of the very design, I was pretty pissed to be honest. Not at him, at the industry. I have a lot better things to do with my time than mess about with shit power meters.

I typically dismiss what pro teams use. They're either paid to use, paid to promote, under contract never to criticize, they're pro cyclists... not amatures overly obsessed with details such as "why is this power meter a few % wonky".

Back to why pro teams don't care? They'll still win without power meters. They're non essential equipment. (Non essential as in if they stop working mid-race, it doesn't matter... not to be confused with the benefits of training with power).

Anyhow, much like the debate around power meters not matching or being 'accurate', like it or not the result of people obsessing over this eventually leads to better products on the market. InPeak is a perfect example. They sent me a total potato power meter, then 1/2 potato, they now have one of the best single sided meters I've tested. If only Shimano didn't have their heads so far up their own arses, they'd also sort out their power potato. I expect their next power meter won't be such an embarrasement to an otherwise brilliant product line (I love my Di2).

More ramblings...The term 'influencer' is skewed. Every day I'm asked, pressured, and told what to say and post with people or companies with their own agendas. Those are the real influencers. For me, I'd rather influence companies to make better products.

Dismissing people/posts that point out the finer details gives the industry a free pass to keep selling shit. This is the problem.
You are pure class with your attitude, approach and content. Which is the reason I have since unsubbed from many channels on youtube but not yours and a few others. Thanks for all you do and all the info you provide!

Post Reply
  • Similar Topics
    Replies
    Views
    Last post