I wonder if the torque could be measured by sensing the rim position verses the hub. This could be done with timing from two sensors One near the inside of the wheel and one near the outside of the wheel.

The torque would be a _big_ function of the tension of the spokes so careful calibration would be needed.

 

Interesting thought, but at the relatively low torque we're talking about with a bicycle I wonder if bumps on the road would distort the rim vs hub alignment as much or more than torque from the pedals? The measurement would have to have such fine resolution. Also torque from the pedals is going to come in pulses, which will vary in relation to rim position.

 

Interesting thought, but at the relatively low torque we're talking about with a bicycle I wonder if bumps on the road would distort the rim vs hub alignment as much or more than torque from the pedals? The measurement would have to have such fine resolution. Also torque from the pedals is going to come in pulses, which will vary in relation to rim position.

Good points. Here are some thoughts...

I see my description was sloppy. I meant the measurement to be of the rotational change of the rim verses the hub. I believe this is sometimes referred to as wrap.

Bumps will tend to be "applied" across the diameter of the wheel whereas the torque will be rotational so I doubt there will be a lot of interaction. These same bumps will be sensed by a pedal strain gauge as well. This "noise" will have to be dealt with in any measurement. Another confusing factor in this scheme is that torque caused by braking. Maybe with careful sensor placement the braking torque would be recognizable and ignored.

Time measurement is easy to do to high resolution in a slow mechanical system such as this. I suspect that there is plenty of time to do multiple measurements and then do digital filtering on the results.

A good bicycle rider pedals real smooth therefore no pulses.

Hmmm. Since the pedals are geared up there will be multiple pulses per revolution of the rim -- and the number of pulses per revolution will vary with the gear selected. Actually, I think that the pulsing is going to be a nuisance no matter what power measurement method is used.

 

Actually, I think that the pulsing is going to be a nuisance no matter what power measurement method is used.

That's why using an RMS algorithm for this application is ideal. Fortunately, there already are chips out there that do just that.

 

@RichardO could be onto something here!

Measuring at the hub and rim could be tricky but a couple magnets placed each end of a spoke and sensors on the chain stays could give us the spoke deformation, although it'd be tricky to calibrate to give true a wattage reading and could be thrown out by a change in spoke tension somewhere but still, a nice idea that is well worth experimenting with.

I'm not sure I'm too worried about pulses, we don't really need to be outputting at sub-second resolution, the data could easily have some smoothing applied to it to even things out. Even without smoothing, the raw data with those pulses could be useful for analysing a cyclists pedalling efficiency.

Braking is another good point but I think can be easily dealt with, we can monitor crank rotation and we know that if crank rotation is 0 then no power is being generated by the rider, regardless of what our wheel sensors are picking up. A rider isn't likely to be pedalling while braking so it can probably be ignored.

Thanks guys, that's really got my mind working in a different direction now

 

I'm wondering why an accelerometer can't tell you everything you need. I mean, if you know a rider's mass, speed and gear setting, any change in acceleration of the bike frame must be due to a predictable pedal pressure. Analysis of the 3-dimensional accelerometer data should even tell you the position of the bike's frame (off of vertical) at every moment.

Changes in the air resistance hitting the rider would distort things. Air sensors might help you compensate for that but maybe it's too large a factor to eliminate. Hmmm...

 

Maybe an unworkable idea, but, would a "clipless" pedal and shoe, with the strain gauge mounted between the sole and the lug give a better reading of the force? Doing this would allow a reading of both the down and up stroke of the pedaling.

 

@wayneh An accelerometer is a great idea in theory. There is actually a company who produces a power meter very similar to what you're describing, www.powerpodsports.com. The problem though I think is that wind resistance is such a huge deal on a racing bike and while grabbing opposing wind is doable other factors affecting aerodynamics are impossible to catch, bike frame, wheel type, rider position, their clothing and even the angle of the head can all make a massive difference and that's before even thinking about rolling resistance. Powerpod have done quite a good job but they've got there with a boffin who really knows his physics, 10 years of development, some seriously complicated calibration algorithms in the meter itself and some even more complicated software supporting in the thing that I'm not even going to try to figure out.

@shortbus Having something on the shoe would be perfect, it's the part of the whole drive train that's going to experience the biggest flex so would be a great place to mount the stain gauges. The problem I'm having it figuring out how to actually mount it in a way that it's going to last. The shoe gets quite a bit of abuse so making something durable enough is going to be tricky.

 

I wonder if the torque could be measured by sensing the rim position verses the hub. This could be done with timing from two sensors One near the inside of the wheel and one near the outside of the wheel.

The torque would be a _big_ function of the tension of the spokes so careful calibration would be needed.

This concept applied to the crank / chain ring seems viable- could be more stable and predictable than using the wheel.

Make a custom chain ring mount that floats the chain ring rotational axis on springs, so there is a slight rotary displacement between the crank and the chain ring when torque is applied. Now add two sensors that read notches on both parts, dynamically measure the phase change using a MCU counter / timer to calculate torque, RPM and of course power.

Some progressive springs used here could give you a bigger dynamic range, with appropriate calibration.

 

My daughter is studying physical therapy, and I recall her telling me about a way they can measure muscle tension as a person exercises. We tend to think measuring forces on the bike hardware will be easier, but maybe there is a way to measure forces in vivo. I'll ask her about it.

 

@shortbus Having something on the shoe would be perfect, it's the part of the whole drive train that's going to experience the biggest flex so would be a great place to mount the stain gauges. The problem I'm having it figuring out how to actually mount it in a way that it's going to last. The shoe gets quite a bit of abuse so making something durable enough is going to be tricky.

One quick idea. A thin semi-ridgid polymer insole with a stud that would extend through a hole in the out sole of the shoe to bolt the clip containing the pressure sensor to the shoe.

 

My daughter is studying physical therapy, and I recall her telling me about a way they can measure muscle tension as a person exercises. We tend to think measuring forces on the bike hardware will be easier, but maybe there is a way to measure forces in vivo. I'll ask her about it.

Wayneh, I'd really like to hear you elaborate on that.

 

Wayneh, I'd really like to hear you elaborate on that.

The technology she had told me about was EMG, electromyography. This can use external pads to measure the number of motor groups firing but unfortunately there is no easy way to correlate this to work applied to the pedals. Useful technology and intersesting data, but a dead end for this particular application. But you're probably thinking of a different application.

 

But you're probably thinking of a different application.

You got that right ... thanks for the info!