When Specialized set out to create their S-Works Carbon Power Cranks, they started with a lofty list of objectives. Topping the list was accuracy. Although they wanted their meter to be light, durable, reliable, and easy to use, accuracy in all conditions was the most important box they had to tick. To achieve that end, they had to call their scientific ace—Rodger Kram, Ph.D.

Kram has spent a lifetime studying physiology and the biomechanics of terrestrial locomotion. He’s the guy to call when you want to know the exact metabolic energy expenditure of an ant—or an elephant. Yes, he has already done the study. When the Olympic Committee wanted to know if sprinter Oscar Pristorius had an unfair advantage running with the aid of his lower-leg prosthetics, it was Kram who did the computations. If you have some time to kill, you can read Kram’s 200+ scientific studies listed just on Google alone. In order for Specialized to claim their meter was the most accurate in the world, they needed Kram to prove it.

Specialized recruited Ph.D Rodger Kram to help test the accuracy of their new power meter cranks.

Whereas most power meters are tested in labs using a basic lever and strain gauge, Kram knew he needed to create a more accurate, real-world testing platform. A longtime advocate of treadmills, Kram’s tests distill the variables to the basic metrics of time, mass, and length. Using his lab at the University of Colorado, Kram and his international tribe of fellow doctorates built a full-size treadmill capable of measuring the exact energy expenditure of a rider on a bike. Unlike other test scenarios whereby the rider is perched atop a bike fixed to a trainer, the treadmill provides the most realistic readings. The level of exertion could then be correlated to watts recorded at the crank.

To attain the most accurate measurements, Kram and his team had to fold in a host of computations and controls. As Kram walked our team of journalists through the testing process you could smell the fug of frying brain synapses. It was a lot to pretend to understand.

In the most general overview, the treadmill tests were rather simple. A motor turned the treadmill at a consistent rate with a specific watt output. The treadmill was put on an incline to force the rider to pedal with enough force to stay in position on the belt. The rider’s output had to match the treadmill motor’s output. That’s the simpleton’s overview. But, nothing is that simple.

Specialized recruited Ph.D Rodger Kram to help test the accuracy of their new power meter cranks.

In order to calculate the wattage perfectly, several variables had to be addressed. The belt generates heat, and that heat had to be factored into the test. While there is no wind drag, there is spoke churn. That resistance had to be considered. The weight of the rider and the slope of the treadmill had to be in perfect balance. Every time the rider took a sip of water, it was included in the test calculations. If they sweated too much, that too had to be offset.

When asked if there were formulas used to negate drivetrain drag, Kram seemed a bit nonplussed. Drivetrain loss appears to be the Loch Ness Monster of locomotive testing. Although not perfectly quantified, multiple experts have estimated the mechanical drag of a bicycle’s drivetrain at around 2%. While they can’t isolate drivetrain drag and mechanically measure it, they can arrive at that number value by other means. So, 2% it is.

The most mind-bending calculation Kram and his team considered in their testing was the precise impact of the earth’s gravity as it pulled the rider down the slope of the treadmill. Gravity is not the same wherever you go. Ask Kram what the specific pull of gravity is for Boulder, Colorado and he’ll have the answer.

Specialized recruited Ph.D Rodger Kram to help test the accuracy of their new power meter cranks.

As the testing progressed, more criteria were introduced. The S-Works power cranks were tossed in a freezer then heated to such a high temperature they had to be handled with oven mitts. They were calibrated when hot, then tested cold. They even scrutinized the meter’s readings with the drivetrain heavily cross-chained. In that test, they found cross-chaining had minimal effect on crank-based meters but did impact spider-based units considerably.

After weeks of testing with the Specialized engineering team, their new cranks appeared to live up to their claims as the most accurate power meter on the market. They were ready for launch but needed real-world road testing.

They took the cranks outside — and they didn’t work.

Specialized recruited Ph.D Rodger Kram to help test the accuracy of their new power meter cranks.

When Scientists Become Detectives

While talking with Chris Yu of Specialized, he mentioned the accuracy setback that befuddled the engineering team and delayed the product launch. During the initial road tests, a couple of riders had reported random accuracy discrepancies from right to left crank. Because they’re incredibly smart, the Specialized team recreated the riding scenarios and made a critical observation. As riders pedaled north or south, east or west, the sun was imparting unexpected thermal imbalances from crank to crank. In short, the sun was warming one arm and not the other. Although it didn’t effect aluminum crankarms, it was something they had to address with carbon arms. Fortunately all it took was a simple tweak of the software and the fix was made. After more testing, the Specialized confirmed the accuracy of the meter at +/- 1.5%.

Specialized recruited Ph.D Rodger Kram to help test the accuracy of their new power meter cranks.

More than just a power meter tester

The more time we spent with Kram’s treadmill, the more we all began to see its possibilities. Just as Specialized uses their Win Tunnel to test aerodynamic drag, their treadmill (Specialized now has their own in Morgan Hill, CA…called the Win Mill) can be used to test just about anything not impacted by airspeed. They can test tire drag in real-world conditions with a minimum of variables. They can measure the drag of a straight-line sprinter and a rider with a wiggly front wheel. It’s possible to fold in additional tools like infrared cameras and Retul measurement systems to better evaluate locomotive efficiency.

Which means all of the development time, effort and resources that went into creating the new Power Cranks is just the beginning. And with Kram’s brain involved, the sky is the limit.



  1. > Whereas most power meters are tested in labs using a basic lever and strain gauge

    It can’t possibly be done that naively at other power meter manufacturers.

    • Outside of SRM, you’ll find most low end power meters are that bad. Some try to compare say a pedal or hub or whatever and use an SRM as the control then meausre difference without taking into account an SRM is +/- 2% accurate to start so your numbers are also ttje off by that much. Add in crann arm flex or frame flex the ability to make a cheap, worthless powermeter is pretty easy.

    • All,
      As a rule, we don’t publish sponsored stories without full disclosure. Christophe attended Specialized’s Power Crank launch event for us and wrote this story because we thought it was really interesting, not because we were paid or otherwise influenced to do so. If our stories sometimes seem overly positive on a product or brand, it’s because we’re truly excited about it. Specialized gets a bad rap for some of their less than stellar PR moves in the past, and it’s easy to hate the big brands, but having spent a lot of time with them over the years, they genuinely do care about the rider, their products and advancing the sport and technology. Not to go overboard in praising them, but how many other brands do you know that have invested millions into testing equipment then also use it to produce entertaining videos like their Win Tunnel series? Or pour hundreds of thousands of dollars worth of cash and equipment into NICA?

      While I appreciate that our readers are skeptical of puff pieces, I assure you we don’t write them just because we’re asked to. And if any content is sponsored, we make it very clear within the article.

      Thanks for reading,


      • Tyler,
        Sorry to disagree with you but you can’t have such an affirmative title “most accurate in the world” when you have no proof and not the engineering knowledge to challenge this claim!
        if you title “history behind Specialized power meter, building on 4iii legacy” that would have been fair.

        • I agree with Brent. Having such a title should only be appropriate in a setting where some sort of unbiased statistical analysis of competitors is provided. In this instance, there is no data provided to support or deny the title. To me this only devalues Bike Rumor’s articles.

          I’m not saying that I don’t believe the title or think that Specialized couldn’t make a product deserving of this title, I am only saying that this article has NO data, or a link to any data of any kind.

    • Love it…Myke2241, The dam best response out there ” The most accurate PM out there, is the one you own” Just Brilliant!!!!

  2. For the record, you don’t need a treadmill to accurately and precisely calibrate or test a power meter. There are a number of ways to do this. All that’s really required is a device whose power output is well characterized across a desired power range and a method for measuring changes in power readings as the result of temperature changes. I will give Specialized and Dr. Kram credit: they did devise one of the most complicated ways to calibrate/test a power meter.

  3. If you’re using it as a team or manufacturer for equipment testing across multiple bikes, accuracy has some value. But for an individual who is only using one power meter for training purposes, consistency (thermal, temporal) & precision >>>>> accuracy.
    I can’t see in the article where they tested a statistically significant sample size of different power meters to validate their ‘most accurate’ claim, but perhaps they did. And I do wonder that if the thermal differences were significant enough to matter, but have been controlled for with a pure _software_ fix, how reliable the (+/-?) 1.5% accuracy claim is.

  4. ha-ha! “most accurate”? they adapted 4iiii power meter. changed form factor, improved water resistance, and slightly tweaked strain gauge. and after this stated +/- 1.5%. meanwhile garmin vector 3 +/- 1.0%. this article looks like sponsored bulshit. sorry.

  5. This test protocol seems overly complicated. I trust the lever and strain gauge method more. The temperature susceptibility of precision electronics is well understood. Specialized doesn’t need a bio mechanics expert to confirm accuracy, it needs good engineers that understand high precision analog/digital conversions, voltage references, EMI noise immunity, strain gauge designs, proper signal filtering, PVT variations and how precision electronics drift with age. This is the kind of needless experimental confirmation I’d expect out of an otherwise idle QA group of engineers that didn’t quite make the cut to be involved in the actual product design.

    Also, 1.5% is not even close to what a good design could achieve. Its what a good cheap design can achieve. For another $10 in electronic bill of materials costs, premium voltage references and components could bring that number down an order of magnitude, but no bike application would ever need that.

    • This testing came from their marketing budget, not their QA budget. Also, precision electronics does not eliminate carbon stiffness variation over time.

      • Sorry Cheese but contrary to many cyclists belief carbon stiffness DOES NOT vary over time. I worked in a lab and it was one of our research subject. Unless you load it above 95percent of failure load there is almost no fatigue or stiffness load across millions of loads.

          • he’s still correct though. I don’t know what this stages thing you’re referencing is, but if they had a poor resin system it would have problems from day one. epoxy doesn’t change over time. if it did the layers of composites would shear and crack.

  6. Accuracy is great, but all you really need is precision since its rare you’ll compare your data to another rider’s.

    Example, if my power meter always reads 20% high, I’ll train to an FTP 20% high.

    The only time accuracy is an issue is when switching from one bike to another and needed power data.

  7. Hey walk over this bridge in Miami, an engineer signed it off. He has degrees and licenses. It was designed by people with academic resumes as long as your arm!

    I say this as an engineer who know himself and other engineers as well as people in marketing. Case in point, somehow Specialized gifted engineering department somehow forgot that all strain gauges are afected by temperatures, which the sun will induce via radiation alone.

  8. Great! But if you’re a normal human (or even a world tour pro) this will make no real world difference over pretty much every other direct force power meter in the world. I’ve been training and racing with PM’s for close to 7 years and have gone between PowerTap, SRM, and Stages (sometimes all on the same bike) literally makes zero difference. Now obviously If you’re heavily weighted on one side (tiny % of population) than you’ll need dual sided but if Chris Froome can win the Tour on Bread, Water, Ventolin, and a mere stages than you don’t need anything more for your weekly proffamateur Masters 50+ crit.

    • ‘I’ve been training and racing with PM’s for close to 7 years and have gone between PowerTap, SRM, and Stages (sometimes all on the same bike) literally makes zero difference.’

      Then you ain’t doin’ right.

      • Ohh well progressed through multiple categories, won several road races and State TT titles and had a lot of fun riding my bike. Sounds like I am doing it right.

  9. They should have spent that money and brain power on making sure Allez forks didn’t have to be recalled for several months with only a $75 voucher to console folks that bought those bikes and had stationary art pieces for several months

    • You could’ve asked for a refund from your retailer (the two around where I live were giving them, if asked), and obtained another non-recalled bike, even a non-Specialized.

      But, hey, nice zing.

  10. This is amusing-
    A buddy of mine got his hands on one of these units last fall (2017)- It has proven to be highly inaccurate and pretty much unreliable.
    But much like a broken clock- It is still occasionally correct.

  11. This is what hate about Specialized….hyperbole.

    So tell us about why you hooked up with 4iii?……..isn’t this product really theirs, with a marketing speel wrapped around it?


  12. There was a comparison study with other power meters on cyclingnews – specialized showed power on both sides when pedaling one legged. Pure PR puff piece.

    • Not trying to disagree with you as I think this is mostly marketing but don’t 4iiii meters estimate L/R split based on crank position? I know my old PowerTap C1 did it this way and occasionally you could trick it but again L/R balance is pretty much useless to 99.99999% of riders.

        • I know that, I’m talking about the balance e.g. 48%Left 52%Right. Some have strain gauges in each side so it’s a precise measurement of what each leg is actually doing whereas my PowerTap didn’t so it just estimated based on the assumption that 100% of the torque is being pushed on the downstroke.

  13. Infocrank. What ever you might think of it as a product there are some pretty high end sports science professionals, not least British cycling, adopting it as the demonstrably most accurate power meter available.

  14. A good amount of self-declared power-meter manufacturing experts in the comments section here, lol. Well done to Specialized and all other brands investing and striving for excellence in science to validate and improve the products they make.

  15. This is the Rodger Kram that was part of the study that showed Pistorius had no advantage when simple time split analysis showed that he didn’t slow down in the later half of a 400m when all the runners with lower muscles fatigue and slow down markedly.

    Additionally, frictional losses in the drivetrain are known to scale with power, so using a baseline (and very low) single value for losses rather invalidates all the other effort.

    As others have said – I’d be much more impressed by them building a gold standard dynamic test rig and comparing a bunch of PMs on that.

  16. Ironic that a company that is so concerned with trademark names has named their product identically to another cycling product on the market for years.

COMMENT HERE: (For best results, log in through Wordpress or your social media account. Anonymous/fake email comments may be unapproved or deleted. ALL first-time commenter's posts are held for moderation. Check our Comment Policy for full details.)

This site uses Akismet to reduce spam. Learn how your comment data is processed.