Well actually checking lots of powermeters. I have 4, all left armed stages. I am convinced a couple read lower than the other 2 and as such I want to figure out best way to test them against each other. I have a smart turbo but it a tacx flux and as such accuracy won’t be able to give me definitive results if I compare it against the power meters.

Alternatively, I’m thinking if I clamp the arms in a vice and then hang a set weight of them then the number I get should be the same across all the devices. Is that thinking correct? Only issue I can think of is one of them is slightly shorter (165mm vs 172.5) than the others which may mean some additional calculations?

Will that work?

The power meter you can self fit comes with a bag for filling with water for calibration. Iirc the procedure sounds similar to what you are proposing.

Could you not use a torque wrench.

Differences in crank length should not be an issue as long as you correct for it.
Torque = force x distance

Can the crank arms read out torque only?
Just asking, never checked when I had a Stages powemeter.

Power = torque x angular velocity
The measurement of angular velocity (cadence) will also need to be checked.
Not sure how best to do this.

Can the crank arms read out torque only?

Don’t think so, I was just going to go by the wattage reading..which now I type it sounds like it probably won’t work!!!

Could you not use a torque wrench

How would that work?

To get a watt reading the crank must be rotating, static weight will not work.
Turbo may be as good as it gets, or same speed up same hill in still air.

Alternatively just stick to one powermeter and forget all about the possible differences 😃

I would compare each one, one by one against the reading the turbo gives. Knowing the +/-accuracy of each device and the turbo, you can work out the range of overlap and the confidence you have that some of them read similarly.

One or two percent difference is barely the difference between a good days training session and a bad days session. If it’s more than a few percent at least you can make an educated adjustment depending which one you are using

Does your head unit report torque? If so you can use this.

If you use an exact weight and hang it off the pedal with the crank arm exactly horizontal you can check the reported torque against what you’ve set up. Torque = force x distance.
That’s effectively the Garmin Vector calibration procedure.

Have you carried out a zero offset on each? Also have you entered the crank length into the head unit as this will affect reported power.

Cadence should be either right or very wrong – it shouldn’t drift slightly so if you suspect a slight error on some cranks it’s probably not that.

Another way would be to use a turbo trainer with built in power meter and check the reported power from the cranks and the turbo. Probably the cranks will report very slightly higher power due to drivetrain losses but that should be small and also consistent from one set of cranks to another.

I just re read your OP and see you have a smart turbo. Whilst the accuracy of the turbo might not be as tight as you’d like, it should be quite repeatable from one measurement to the next. If you check each power meter against the turbo one after the other keeping all settings the same (same gear, same speed, ideally same bike etc) compare the difference between crank and turbo for each to see how similar they read.

Don’t worry about matching the crank to turbo absolute numbers – it’s the difference that matters.

Use the the flux as a dumb trainer and evaluate power for speed.

If it is anything like our Tacx Flux then the “smart turbo” will be the least accurate bit of the whole measurement chain (the one the lad uses is around 30-40% out compared to calibrated strain gauge ones and proper smart turbo used by his coach for power tests).

I do a lot with strain gauge calibration at work – dead weights are fine but it will only work if the electronics give you access to the appropriate raw torque data or a calibration app. There will be a lot of maths, cycle counting, down sampling, filtering etc behind the power data you eventually see.

Also have you entered the crank length into the head unit as this will affect reported power.

Eh, do I need to do that?

I have a garmin 810, I think it gives the torque setting at the calibration stage.

I do a lot with strain gauge calibration at work – dead weights are fine but it will only work if the electronics give you access to the appropriate raw torque data or a calibration app. There will be a lot of maths, cycle counting, down sampling, filtering etc behind the power data you eventually see.

No idea what any of that means so im assuming it’s not going to be as simple as hanging a weight off the crank arm

Basically what data does your power meter app / display give you? Just power and revs?

Or are there more options, calibration menus? If you can delve that deep then a dead weight calibration with cranks level force x distance = torque would work. Read the instructions.

Your crank is likely to be strain gauged to measure bending strain. That is then used to calculate torque. With revs that gives power.

Too tired to explain the rest. One gauged crank but two legs. Loads of data. Maths. Approximations. Throw away excess data and noisy bits. Give pretty smoothed out power graph to the dumb cyclist 🙂

Everyone seems to be banging on about torque but it’s not measuring torque is it? It’s basically a strain gauge on the side of the crank arm isn’t it so is it not measuring either bending or tensile load?

The OPs idea isn’t that whacky really you’d just be using a known “calibration load” applied the same way to all four devices, and see if the readings given correlate, it wouldn’t really tell you if the derived power figure is accurate, just establish if all four cranks are consistent with one another…

it wouldn’t really tell you if the derived power figure is accurate, just establish if all four cranks are consistent with one another…

Yep, I don’t care if they are accurate, only that they are consistent with each other.

I’m going to hang a weight off one of my power meters tomorrow and see if I get any number readings..then can decide if it’s worth repeating on the other devices.

Just ride down the same bit of road in the same gear and cadence for a short distance, enough to stabalise everything, back to back so same weight and weather conditions. Should be accurate enough. The numbers jump around so much your only approximating the output anyway so only really getting a reading to within 5 or so watts.

But power needs a rotation component – that’s what power is. My power meter reads zero if I’m not moving the pedals even if I’m standing on them.

Everyone seems to be banging on about torque but it’s not measuring torque is it? It’s basically a strain gauge on the side of the crank arm

Eh? What else would a strain gauge on the crank be measuring other than torque?

These things measure power by measuring torque and multiplying it by speed. I used to have a Powertap, it could display torque or power, and standing on the pedals but not moving would display a torque figure but not a power reading. As expected, because Physics.

Also have you entered the crank length into the head unit as this will affect reported power.

Eh, do I need to do that?

Possibly. You certainly do on pedal based power meters as power = force x distance x speed, and as the power meter stain gauge measures force then distance must be known (distance = crank length). It might be that as the crank arm based meters are a fixed length then that’s already programmed into the signal. Definitely check the manual!

Different power meters but just for comparison this is the Garmin Vector static calibration process that you could sort of replicate to check consistency if you can get a torque reading out of the meter / head unit.

vector 3 calibration

It used a fixed, known weight and calibrates to that.

I also have 4 stages. All 4 of them have been replaced under warranty at some point. One of them was reading high, one was reading low and two of them kept reporting 0w incorrectly.

I now have 4 of the latest gen3 models.

Two of the new replacements have started reporting 0w again and one other replacement that was reading consistently with the others has started to over read again.

By all means validate that they’re consistent and / or accurate today but there’s no guarantee that they’ll be consistent or accurate tomorrow.

I gave up Stages PM’s due to inconsistency, the biggest issue for me was a lack of sensitivity below about 250w by consistently reading low and screwing up training data. Above 300w they were fine.

Had quite a few replaced on warranty (very good customer service), mainly for inconsistency or drop outs but ultimately I only use Quarq now as they really are fit-and-forget.

Eh? What else would a strain gauge on the crank be measuring other than torque?

Bending, if it were measuring torsional loads on the BB axle then that would be torque.
You’re not measuring ‘P’ from these diagrams:

Where your stages arm is attempting to measure something more like the first load case shown here:

The easiest solution would be to borrow a pedal or hub based PM and whack on each of the 4 cranks and look for any offset trends or weird peaks (I’d bet the Stages jump up to 2000W every now and again).

Hanging weights and trying to be clever sounds good but there may be something iffy with the Stages PM after the strain gauge. What you really want to know is PM output vs another power measurement so maybe try and do that?

Or just buy a P2M.

Bending

And what’s the bending proportional to? I can’t make out the formula in the image – it looks like there’s an el squared on top and a el on the bottom which makes little sense so maybe it’s an eye or something?

Ignore, brain fart.

My power meter reads zero if I’m not moving the pedals even if I’m standing on them.

As I’ve just found out, if there is no cadence then there is no reading! So that’s my plan out the window.

I think the only option is to calibrate against another power meter being used at same time. As alluded to the tacx flux power reading is not much better than that of a wheel on trainer, so in absence of anything else im a bit stuffed.

in absence of anything else im a bit stuffed

I think you’ll be fine. Do the same ride on each crank on the same bike one after the other. Swap the cranks without even taking the bike off the trainer. The tacx flow might not be perfect, but it’ll be consistent (enough) with itself for that long.
The zwiftpower analysis feature will graph each crank against the flow for you.