I'm working on a little project to measure the power put out by a cyclist while riding. The usual way of doing this is with strain gauges, my current design has them placed on the bike's crank arm. Ideally I'd like this to be as easy to install as possible and as we all know, strain gauges are anything but so I'm trying to think up different ways of doing this.

In theory the power could be measured at any point in the drive train from the shoe through to the wheel. I've got a couple of alternative ideas but am a bit lost of how to go about measuring without strain gauges.

Idea number 1 is to have something in the cyclist's shoe that could measure downward force, as it's in the shoe it'd need to be as flat as possible so not to cause any discomfort.

Idea number 2 is to measure the tension of the chain but again I'm not really too sure where to start doing that.

Any ideas or pointers would be appreciated,

Cheers....

 

Have you surveyed the commercial products that are already out there for this purpose? A lot of smart people have worked on this exact problem for decades, so it's probably safe to assume that the easy ideas have been exhausted. A major improvement will most likely only come from a new sensor technology.

 

I'm working on a little project to measure the power put out by a cyclist while riding. The usual way of doing this is with strain gauges, my current design has them placed on the bike's crank arm. Ideally I'd like this to be as easy to install as possible and as we all know, strain gauges are anything but so I'm trying to think up different ways of doing this.

In theory the power could be measured at any point in the drive train from the shoe through to the wheel. I've got a couple of alternative ideas but am a bit lost of how to go about measuring without strain gauges.

Idea number 1 is to have something in the cyclist's shoe that could measure downward force, as it's in the shoe it'd need to be as flat as possible so not to cause any discomfort.

Idea number 2 is to measure the tension of the chain but again I'm not really too sure where to start doing that.

Any ideas or pointers would be appreciated,

Cheers....

Idea #2 could be implemented by placing a small sprocket in the middle of the chain, and have it press inwards (or outwards, depending on the design) using a spring. The deflection of that spring could be a good indicator of the chain's tension.

 

wayneh, but where's the fun in that?! I'm like building things just to see if I can

cmartinez, I was thinking something like that but the problem I've been coming across is when the gears change and the chain moves. I've been toying with the idea of trying to measure chain deformation and stretch but can't think of a way that's precise enough.

Another idea is to use some sort of QTC sensor built into an innersole, the problem there is that I don't think that QTC will be able to handle the forces involved.

 

far shot
http://www.loadstarsensors.com/why-build-capacitive-load-cells.html

 

wayneh, but where's the fun in that?! I'm like building things just to see if I can

cmartinez, I was thinking something like that but the problem I've been coming across is when the gears change and the chain moves. I've been toying with the idea of trying to measure chain deformation and stretch but can't think of a way that's precise enough.

Another idea is to use some sort of QTC sensor built into an innersole, the problem there is that I don't think that QTC will be able to handle the forces involved.

Isn't there a small tensor idling sprocket already placed on the rear wheel gear change mechanism? Maybe installing a high resolution encoder on it's pivoting arm would do the trick.

 

Isn't there a small tensor idling sprocket already placed on the rear wheel gear change mechanism? Maybe installing a high resolution encoder on it's pivoting arm would do the trick.

Isn't that small sprocket tied to a shift bracket and the shift cable? How would that be used for "sensing" input power?

 

Isn't that small sprocket tied to a shift bracket and the shift cable? How would that be used for "sensing" input power?

I'm not exactly sure, but I'm under the impression that that thing reacts to the chain's tension by oscillating back and forth. But I do not know by how much, and if it's substantial enough for an encoder or not.

​

 

Torque sensors are installed on most pedal assist systems (PAS) on high end ebikes.
You can read all about it here:

http://www.electricbike.com/torque-sensors/

 

It's been a while since I worked on/with one of them but IIRC the "cage plate" in the second picture ties both idlerstogether. And the cable moves the jockey sprocket. If it was free to move on its own it would not keep the chain in one gear ratio.

 

Is this going on a daily rider bicycle where people are wearing sneakers, or on a higher end bicycle where riders are typically attached with specialized toe clips?

Assuming the ladder; maybe you can split the pedal clip into 2, and sandwich an FSR in between the two halves. Or build it into the bottom of the riders riding shoes. You'll probably have to make it battery powered and wireless though... This will give you force. Then maybe you can use something like a hall effect sensor, or the sensors that cars used on the crank shafts, on the bicycle crank to determine revolutions of the crank. Essentially count the teeth on the sprocket as they go by. Force times angular distance traveled will give you the work applied.

 

I'm not exactly sure, but I'm under the impression that that thing reacts to the chain's tension by oscillating back and forth. But I do not know by how much, and if it's substantial enough for an encoder or not.

View attachment 110930

View attachment 110931​

That's for keeping chain tension consistent regardless of gear selection. As the rider shifts gears, the chain jumps to different sprockets, changing the required length of the chain. When the chain jumps to smaller sprockets, suddenly the chain is too long and will be floppy, and vice versa. That part is spring loaded and keeps the chain tension constant as the chain moves between different sized sprockets. Note that this happens on the UNloaded side of the chain, and therefore cannot be used to judge chain tension due to rider power input.

You could put a spring loaded sprocket pushing perpendicular to the chain on the top side of the chain with an FSR in-line with the tensioner to measure chain tension. The sprocket will need to deform the strait line created by the chain. The more force the rider exerts on the pedals, the more force the chain will exert on the sprocket as it attempts to make a strait line. Then count teeth on the sprocket as they go by to measure crank angular velocity.

 

Measuring the jockey wheel is a nice thought but that only acts to take up the slack as the chain moves to different gears. Any chain measurement would need to be taken between the chain rings and the cassette (sprockets at the back) as that's where the chain is under load, once it's passed the cassette and got as far as the jockey wheels, it's already transferred the force to the back wheel.

 

Mr Software, I'll have a look into FSRs, they're not something that I've used before but could in theory do the trick, I just wonder if they can take the sorts of forces that they'd need to for this. And yup, the idea is for this to go on a racing bike with the rider clipped in.

For measuring revolutions, there are already wireless cadence sensors available that transit over ANT+ so was planning on just using one of those.

 

wayneh, but where's the fun in that?! I'm like building things just to see if I can

Well that's fine, but personally I find it more fun if I go in with heads up and know full well what has already been done. It's not fun to reinvent a wheel only to discover later that you're years behind state of the art.

For instance, I'm nearly certain that the pedal-clip idea in #11 is already a thing.

Ever read Bicycling Science? It's a great book.

 

Why not use a couple of accelerometers? This will measure and account for all force and energy vectors. Like the sideways oscillation.

 

@wayneh I've competed at quite a high level in triathlon and have been a competitive cyclist for 18 years so do have good idea of what's already out there (I didn't mean that to be as pretentious as it sounds ). Generally, they use strain gauges somewhere in the drivetrain, usually the hub, cranks, bottom bracket or pedal body. There have been attempts at other things but I've never seen anything that's particularly successful.

@BR-549 accelerometers could get us in the ball park but the problem is that there are other outside factors affecting the force output than just the speed. I could probably calculate speed, gradient and gear ratio which would give a rough idea but other things that we can't measure such as wind resistance and the weight of the bike and rider would skew things out too much.

 

A wireless strain gauge in the hub or crank will probably be the least invasive solution for the rider. Using an FSR with a sprocket on top of the chain will give you more drag due to the extra sprocket engaging the chain, and I'm honestly not sure how much the riders will like that if they're riding for a time.

 

So the options are:

1. Measure torque at either the pedal's or rearwheel's hub.
   
2. Measure applied force on each of the pedals
   
3. Measure chain tension

And you must measure rpm's if you want to translate measurements into actual power generated by the cyclist.

Personally, I think it's simpler to measure torque, but it would probably be more expensive. Perhaps the chain tension option is best for what you want, and a lot easier to hook up and wire, IMHO.

Also, I strongly suggest you take a careful look at this page. Lots of interesting stuff there.

 

Thanks for the thoughts guys, it's good to chuck these things around although it looks like I'm going back to the strain gauges on the cranks.

Measuring RPM shouldn't be a big head ache, there are wireless cadence sensors that do just that for bikes which transmit using ANT+, plan was to just link up with one of those. I'm going to need an ANT+ module to send the power numbers out to a cycling computer anyway so using it for cadence makes sense.

@cmartinez cheers for that link, I've not seen that one so will give it a read