# Inside Stages Cycling

Nov 1, 2016

Unless you slept through high school physics, the equation Power = Torque x Cadence should look vaguely familiar, even if you haven't needed to use it since those awkward adolescent years. For Stages Cycling, that equation is much more than a foggy memory; it's the cornerstone of the company, the calculation that their power meters perform thousands of times per ride.

Stages began by designing power meters for stationary bikes, the type used in commercial gyms. Taking the lessons learned from that product, they then launched their own crank arm mounted power meter in 2012. The device's price, light weight, and weatherproof, wireless design helped it turn heads when it was launched, and over the last four years Stages' athletes have brought home multiple World Championship medals in everything from road to enduro racing. Initially, the power meters were only available on aluminum crankarms, but Stages recently released their own carbon crankarm and power meter combination.

The basic principle behind a Stages power meter is fairly simple – a strain gauge is mounted onto a non-driveside crankarm, where it can detect even the smallest amount of flex. An accelerometer is used to determine a rider's cadence, providing the second key part of the equation. Since the power meter is only mounted on one side, the number is multiplied by two, and the result is the amount of power (measured in watts) that a rider is putting out at any given moment. It sounds simple, but the execution is a good deal more complicated, due to the steps required to ensure that the device produces accurate, reliable data.

What exactly is that data used for? Well, by monitoring their power output, athletes are able to keep track of just how much effort they're putting out rather than relying solely on gut instinct. Sure, that four hour ride might have felt hard, but how hard was it really? For the recreational cyclist, the weekend warrior that's happy enough just to be out of the house and in the woods, power isn't going to mean as much, but for elite (or aspiring elite) athletes it can be a very useful tool for training and racing.

In order to learn more, we visited Stages' manufacturing and research and development facility in Boulder, Colorado, the location where every single power meter is designed and assembled.

Stage's power meter assembly process traces a circular path around the room, beginning behind the back wall, where boxes of left crank arms reside, and then continuing on a counterclockwise path to each station until the finished product is ready to be packed and shipped.

Floor to ceiling shelving holds left crank arms for road and mountain bikes waiting to have a power meter installed.

Stages designed their own custom fixture that can easily be adapted to work with different crank arm lengths as they go through the etching process.

The laser etcher prepares the aluminum crank arms by removing any paint or anodization off the backside of the arm that could interfere with the application of the strain gauge.

With their carbon cranks, Stages has arranged for the vendor to deliver them without any clearcoat on the portion of the arm where the strain gauge will be placed.

Prepped cranks arms ready to move onto the next step.

A strain gauge is carefully glued and clamped onto each crank arm. For quality control, employees use an app to monitor how much time passes between applying the adhesive and the gauge.

Once the gauge is affixed and clamped the cranks are put into an oven to cure the adhesive.

The wires coming from the strain gauge are then soldered to a circuit board. The circuit board is able to determine how much the crank deflects under a rider's efforts, information that's used to calculate torque. ANT+ and Bluetooth capabilities, along with an accelerometer are all part of the power meter's electronics. The power meters are also equipped with what Stages call Active Temperature Compensation (ATC) in order to ensure the gauges remain accurate no matter the outside air temperature.

A custom gluing robot is used to attach the outer portion of the meter with an adhesive similar to what's used for car windshields. This helps to ensure that the finished product is waterproof, and able to withstand rides in the worst conditions. The robot is a newer addition to the assembly floor, where it has sped up what used to be a very tedious and time-consuming process.

Next comes a simple but effective test to ensure the power meter is waterproof.
An air tube is attached to the battery compartment and the arm is placed in a container of water. If there aren't any bubbles the device passes and heads onto the next step.

Before it can be shipped out, each unit needs to be calibrated. Using a Bluetooth app, the device is set up first without any weight and then with a 90kg weight, basically 'teaching' the meter to understand an incoming force. A static test for accuracy is also performed on 20% of the alloy units, and on 100% of the carbon arms.

Why the difference in numbers? The carbon crank mounted power meters are still relatively new, and Stages wants to monitor them closely to ensure they're as close to perfect as possible. As time goes on that number will drop - originally 100% of the aluminum units were tested as well.

With the assembly process complete, it's time to package and ship the finished product.
How many power meters does Stages produce each year? They were reluctant to reveal an exact number, but they did say they sell the most power meters in the world.

Research and Development

In addition to the main assembly area, Stages' Boulder location also houses a R&D lab full of metal working machinery, bins of small parts, and stationary bikes connected to computers in order to analyze a product's performance.

Several of Stages' founders used to work at Schwinn, and were able to purchase these massive machines at bargain basement pricing.

Do not enter.

That's a wrap. Keep an eye out for an Ask Us Anything with Stages in the coming months where you'll be able to learn even more about training and racing with power.

• + 69
They can torque the torque, but can they walk the walk?
• - 22
Garpur44 (Nov 1, 2016 at 1:57) (Below Threshold)
Props from me, top pun!
• + 9
power = talk + walk?
• + 36
I doubt that a mere bit of metal could handle my watts tbh. I'm an absolute monster.
• + 36
says theurbanfreeride you get all your watts from hucks to flat
• + 18
I love how this is made to sound technical... employees use an app to measure the time between application and drying... you mean like a watch, or a timer...

Equation at the start should also replace cadence for angular velocity. Stages make assumptions of constant angular velocity based upon cadence I believe
• - 1
Which to be fair is a reasonable assumption when using round chainrings (and the easy way to prove that is ride with a loud hub and notice how often it ratchets while pedalling: never)
• + 3
Well, maybe it is an app with a timer. I can imagine they work on several cranks simultaneously and they have a single timer application to monitor different cranks. Would be pretty horrible to be working on a single crank at a time with 3 minutes waiting time in between. 20 cranks in an hour, boxes full of these waiting on the shelf. Sun is out, buddies are riding...

I prefer that picture above it, right hand side.

I don't think it matters much whether they use cadence or angular velocity. A cadence sensor measures how long it took for a full rotation, an (angular) velocity sensor measures it over a shorter time or distance interval. At the end of the day it is an average measured within a finite interval. Cadence is much more convenient in case of a crank, unless they went fancy and have acceleration sensors in that unit.
• + 3
" Stages make assumptions of constant angular velocity based upon cadence I believe"

Why? Most off-the-shelf accelerometers can give them instantaneous crankarm position and instantaneous angular velocity.
• + 2
@grgsmith: one extreme case is cycling at low cadence (say 70rpm) out of the saddle and up a hill. Probably not constant angular velocity and definitely not constant torque.

No ratcheting only proves that the angle between the wheel and the cassette never exceeds the engagement angle (say 15 degrees for DT Swiss or 5 degrees for Chris King).
• + 2
@kilpatrick: Good example. The constant angular velocity assumption also creates a different power output in round vs oval rings which is an artefact of the assumption rather than an actual difference. True it is a fair assumption in most cases but it is an assumption nonetheless and there are other power meters where this isn't the case.

My issue with using "cadence" in the original equation is a) It simply isn't correct and assumptions in anything should be transparent b) It just generally reduces Engineers to tears to read something like

@SJP: What sampling frequency do you deem "instantaneous"? IIRC Rotor have one of the highest sampling frequencies out there which is ~500HZ??? I forget what Stages use but it is nothing near that. "Instantaneous" in other devices is more in the GHz territory. I guess too, even if you can measure the angular acceleration frequently, you still have to be able to process, transmit and display all that data. A super accurate PM with a Garmin is fairly useless due to the low number of packets of information that get sent whereas another head unit with other PM's could potentially use that same information differently. If it was straight forward they would have done it I'm sure

P.S. I use Stages and love it
• + 1
Hi Beardless.
This
6-axis MEMS accelerometer/gyro outputs data at 4 Hz to 8 kHz, user-programmable, and is cheap. Certainly, 1000 points per revolution is close enough to instantaneous. Many MEMS accelerometers have mechanical resonance freqs in the hundreds of kHz. You don't need to transmit much data, you can calculate the power per revolution onboard with the simplest microprocessor (probably with the one build into that example MEMS accel/gyro), so you are transmitting of order bytes/second.
• + 2
@grgsmith: Your hub doesn't ratchet because the rider keeps applying torque to the chainring, which in turn keeps the pawls engaged. Measuring cadence and then backing out angular velocity is probably a reasonable assumption for road cyclists where the incline angle doesn't change much per stroke (and your cadence doesn't change much within a cycle), but it's less well suited to mountain biking where cadence can change appreciably in a single cycle. Given the sampling rate available with even pretty basic chips, I'm guessing they're sampling both the strain gauge and the accelerometer at (at least) 1kHz. At a cadence of 90, that sampling rate means that you're sampling approximately every degree of crank rotation, which gives you a far more accurate power measurement than looking at cadence because it gives you a far more accurate view of how both angular velocity and torque change within a cycle.

.... all that said, I kinda doubt that you can look at the power output within a cycle, and I really doubt that the higher sampling rate has much of an effect on the overall energy output in a ride..
• + 8
Am I the only one who doesnt need electronics and a load of maths to tell me that if I trained a bit, I could be more powerful? haha
• + 1
I personally won't be investing in a power meter until I'm racing Cat 2, if I ever make it there!
• + 1
@kilpatrick: is that xc? we can all be more powerful if we train more, dont see the need for something electronic to tell me that
• + 1
@PeterT: Road racing at a much higher level than I currently do. I agree with you, I don't need electronics at this point. I just need to ride more.
• + 9
The fact that it appears they only have 4 employees seems staged.
• + 6
I feel like I'd just be depressed and de-motivated to ride if I knew how little power I generate. I prefer blissful ignorance.
• + 2
I don't need a power meter, but I'd have one if money was no object. I think they are less useful for a general Mtb ride, but for turbo training they are great. Not having the option to ride trails as much as I would like means stuff like this is very useful, but I'll never get one until the price comes down - I'll stick to estimated power...
• + 1
I found the opposite. Estimated power based upon rear wheel speed is great on the turbo but the real use of the PM is to see how that work on the turbo relates to the real world
• + 2
"How many power meters does Stages produce each year? They were reluctant to reveal an exact number, but they did say they sell the most power meters in the world."

They're the only ones not telling or do they have spies?
• + 1
seems like they are doing great work (based on feedback from riders using the arm) but wow that is the dirtiest assembly space I have seen in a long time?

would have assumed they'd at least clean it for a press visit

clean work space = superior manufacturing and QC outcome (physical interaction and also mental state of technicians)
• + 2
There only gluing on a strain gauge. this is done in much dirtier environments with no issues as long as you clean the mating surface. I wouldn't worry too much. They only had four staff who was going to clean ????
• + 3
@darransandwich: Am I the only one bothered by the cover picture with the power meter glued on super crooked. Thought that a comment about it would have been made.
• + 1
@Helm72: um. they're not glued.
• + 1
@obstrep:

"A strain gauge is carefully glued and clamped onto each crank arm. For quality control, employees use an app to monitor how much time passes between applying the adhesive and the gauge."
• + 1
@Helm72: I don't see a crooked one. You mean the wires sticking out being all crooked? They still need to get hooked up to the electronics.
• + 1
@dthomp325: the original poster was referring to the first picture. The power meters do not appeared to be glued in that photo.
• + 1
@TucsonDon: First picture in article, look at the row in the centre of the shot. From the bottom, cranks arms 1 and 2 have lovely looking centrally mounted power meters. However crank 3 has a power meter so far off centre it's overhanging the bevelled edge of the crank. Looks like an employee is just wanging them on willy nilly without much care or attention. That would be fine if the product was cheap...
• - 1
@esucky: still not glued.
• + 1
Pinkbike should do a calibrated test and also compare to other types of power meters for consistency AND accuracy across a range of readings. If I remember correctly I think some of the original pinkbike team are hardware/firmware engineers, so they should no doubt have a good idea on how to test.

I'd be curious to see the wave forms for different types of pedals. I've always suspected that certain clipless pedals could allow slightly longer pedaling force duration. It would be interesting to find out.
• + 2
"are you using protection" then "don't use too much..." well one could probably work with only one glove, or only one earplug, or only one shoe...
• + 2
How can it be accurate when it only measures the left crank (x2), most people right legs are stronger so surly just doubling the left leg power leaves it under rated?
• + 3
Pro athletes like Gwin use left cranks both left and right.
• + 1
@vinay: lol
• + 1
Duh ! Just put both cranks on the same side then
• + 4
It's accurate for training use as you will get the same figures for a given effort. When stages first came to the market everyone knocked it because it wasn't "accurate" by only being left side. However if it's the only type power meter you have it will work just as well as anything else on the market for achieving effective training or targeting set power figures during an event
• + 2
@alexm71 it doesn't have to be "accurate" it has to be consistent to be a valid training tool.

Also, @vinay has it right that some pros do get to use double sided Stages (the pro stuff has light blue markings). I've been out of the PM market for a while but isn't there a double sided version coming to the public too (maybe it is already out?!)
• + 1
@BeardlessMarinRider: Correct on the consistent part.

Stages has said in the past they're not going to come out with the double sided system they test with SKY. They say that SKY didn't fine the extra information useful as they R/L ratio was very similar on the riders. From what I've seen most of the team was back to L only setups.
• + 3
I appreciate the flannel coordination.
• + 2
...and not one comment from @WAKIdesigns regarding all this pysics... did he die?
• + 3
Give him an hour, he's googling.
• + 5
Hired by Trump. Member of Troll Force One. Trolling and meme'ing on cycling forums on Florida.
• + 3
Of course they're in Boulder.
• + 1
Shimano lets these guys glue parts to thiere cranks?
those old milling machines and lathes used to be built in America and built to last.
I miss that era.
• + 1
Maybe it can be useful in testing tires compons,geometry ,position on the bike ,etc. what fits better in a Rider ,other than that for the rest of us just another gimmick.
• - 2
You may have missed an important issue, power isn't torque x cadence. Power is torque x angular speed meaning that here is proportionallity between the power output and crank length. I.e longer cranks will gove you a greater power output for a given torque at the same rpm. That's why they will have the jigs set up to ensure the strain gushes are equally spaced from the bottom bracket.
• + 3
Are you sure? I use the term curcuit frequency (omega) for the Power formula. And that comes down to (Pi x n)/30. So no influence of the crank length at a given torque.
I agree that the angular speed of your pedal attachment point is higher on longer cranks, but isn't it just logical that only a full revolution, meaning the rpms, count? They drive the rear wheel in the end, but how fast your feet move for a given rpm should have no influence on Power at all.
• + 1
@ArturoBandini: I think thats how it works for an electric motor putting out a constant torque, but you wont do that on a bike. Power will loosely follow a sin wave with flats, clips will add a bit more as you can pull up, so you need to derive the power from the sum of flex for the rotation, regardless of how long that rotation takes. Also its the power at the BB (aka what you are putting out) not the wheel (after the transmission).
• + 16
Looks like some have slept through high school physics, indeed.

And cadence is an angular speed : 120 (rpm) equals to 120x2π/60 (rad/s).

Yes, power output and crank arm length are proportionnal, because Torque is N.m.
100kg cyclist putting all his weight on a horizontal crank arm gives 980N, and the distance is 0,175m, so torque (at maximum) is 171.5 (N.m). Average torque in one cycle is somewhere between this and 0 (N.m).

Let's say 120 (rpm) at 50 (N.m) on average will give Power = 120x2π/60 x 50 = 200π = 628 N.m/s = 628W.

The only thing they missed is the one-sided measure : if they measure 200W in the left crank arm but the rider puts in 210W in the right crank arm, it will give a result of 400W instead of 410W. Probably negligible in most cases.
• + 6
Doesn't torque factor in crank length already? Torque is measured in Newton meters so wouldn't the equation be something like power=(cadence)(force*crank length)
• + 1
@natemeyer: correct
• + 3
I agree the substitution of cycling vernacular into the pure mathematics that govern the physical world is somewhat displeasing, however, the angular speed is the same all along the crank arm. As was stated by natemeyer, the torque has already accounted for the crank length, and accounting for it again is not only unnecessary but will also yield a false calculation of power.
• + 17
@Uuno wrote the most, so I'm with him.
• + 1
@simooo: You know me? I can write more .

Does the system also measure the torque put into backpedaling? If you pedal slowly while standing up, you'll distribute your weight between both pedals so the torque from both cranks is fighting each other. No problem of course, otherwise (if you sit down) you'd simply have that load under your ass and you need a 200USD saddle to keep your voice proper low. But if Stages doesn't subtract the torque of one foot from the other, it is going to measure a much higher torque than you actually put into the system. Also the system could never measure the power the muscles put into fighting each other before the foot even moves, depending on your riding technique. That is probably the most important thing to measure, to become efficient at pedaling.

At the end of the day it is also never going to measure the power put into pumping the bike which, depending on the terrain you're riding, makes for a very important part of the power put into propelling your bike.

@simooo: Welcome!
• + 2
@Uuno: while Left/Right data is relatively new in terms of mass availability, the current data suggests that a majority of individuals fall within a +5/-5 from 50/50 split, and the SD drops off fairly rapidly after +10/-10.

if you spend much time on the wattage google forums there's also a lot of discussion around whether 50/50 is even optimal or whether 55/45 is a negative. Again, early days in the research.

What all of the above means is single side measurement while not perfect, having both available may not discern a lot either especially if you are only using a single power meter for measurement. I liken it to a scale that is consistently wrong by 2lbs. While not ideal, so long as you are always using that same scale and the purpose is to measure change/improvement over time, then does it really matter?
• + 6
I like carrots
• + 1
@freerabbit: I like them orange, yellow and purple.
• + 2
I'd like to see tests done for torque with an oval ring
• + 2
Google it and prepare for you mind to be blown
• + 1
So the Athertons were on Schwinn in the early days? I think that's the big take-away here.
• + 2
• + 1
I am a third year engineering student. This summer I want to intern at a company like Stages. Anyone have any advise??
• + 2
That Hardinge is sweet...
• + 1
that one looks a little dirty, but those hardinge tool room lathes are sweet! i have one and retro fitted it into a omni turn cnc lathe.
• + 1
like a pimp of lefty crank
• - 3
So they measure the flex of the arm in a what looks to be less than 1cm area? Hard to believe it can be accurate. I wish there were more details on how it worked.

And isn't carbon much stronger than AL, thus there is less deflection to measure, so it would be less accurate?
• + 4
Its pretty likely that the strain gauge works by measuring the change in electric resistance through the gauge. (The change being caused by the slightest deflection) The technology for measuring variations in electronic systems is extremely accurate, and sensitive.

I would imagine the making sure the measurements are accurate over time / abuse is a huge challenge. Maybe that's the secret sauce but i'd love to hear more on these topics too.
• + 1
@JungleT: Ah of course, that makes sense. I wonder how much deflection there is, nanometers?
• + 2