Introducing Arofly, “the world’s smallest power meter” + video

janusz0 wrote:

OK George, please show us graphs of simultaneous readings from an Arofly and a strain gauge powermeter on the same bicycle over a hilly course. (I wouldn’t  expect precise agreement, but would expect the integrated readings over the entire course to be close.)

Like the ones in this review -- https://road.cc/content/review/227916-arofly-power-meter -- which appear to indicate that the basic concept doesn't work at all?

rkemb wrote:

janusz0 wrote:

OK George, please show us graphs of simultaneous readings from an Arofly and a strain gauge powermeter on the same bicycle over a hilly course. (I wouldn’t  expect precise agreement, but would expect the integrated readings over the entire course to be close.)

Like the ones in this review -- https://road.cc/content/review/227916-arofly-power-meter -- which appear to indicate that the basic concept doesn't work at all?

Thanks for the reminder rkemb.  I'd forgotten about that article!  It doesn't look good, but maybe George can point us at an up to date comparison.

asdfqwerty wrote:

Anyway, proof is in the pudding, maybe DC Rainmaker will get his hands on one...

Apparently it is only being released to the market in February 2017. This can go one of two ways: If it works as advertised, we will very soon see these things everywhere. If it doesn't, we will never hear of it again. Personally, I will watch what happens with the greatest of interest, and will happily part with my £100 if it works. But I certainly won't be at the head of the queue.

jerome wrote:

OK not completely away

The pictures in the FCC submission clearly show a hole and a dent to depress the valve (saw that on another forum). So here is my new guess: the device actually measures tire pressure, as a mean to measure barometric pressure. This is the way they found to protect the pressure sensor, that would otherwise be exposed to water as soon as you run in a puddle. When barometric pressure drops, the tire enveloppe slighly swells and the pressure drops into the tire drops. I do not the relation between the pressure inside the tire and the pressure outside, but maybe it is close enough to 1:1 on that rather small range. As Griff500 rightly pointed out, the FBM320 has an absolute max rating of 10 bars, so they probably asked for a custom version with different gain/offset before ADC. They have to average the pressure over maybe 10s to keep only long term variations.

So to summarize: change of cyclist altitude causes change of barometric pressure, reflected by a change in tire pressure. That give a rate of climb/descent, and along with cc+bike weight gives one part of power. The other part of power required to overcome aerodynamic forces is estimated with cc speed, with is derived from the centripetal acceleration.

'hope this is not a too long and speculative post

Inventing things again? Griff500 never mentioned 10 bars anywhere, he said the supplier only rates it as far as 15psi, which is 1 Bar.

Also, as  I have been at pains to point out, the manufacturer clearly states that the device measures "the reaction force from the ground".  You seem to be telling him he is wrong and it works in a totally different way!

You will also have great difficulty getting a value for the elasticity of the tyre/tube combination, which determines the relationship between internal and external pressure.  1:1, that's a joke. How would the tyre hold 100psi if it has a 1:1 pressure relationship to the outside!

Then of course there is the not inconsiderable problem that changes in tyre pressure due to pedal power changes (the bit that Arofly have told us they measure) are much greater than the altitude related changes you are trying to measure, and will totally mnask your altitude effect.

Griff500 wrote:

jerome wrote:

OK not completely away

The pictures in the FCC submission clearly show a hole and a dent to depress the valve (saw that on another forum). So here is my new guess: the device actually measures tire pressure, as a mean to measure barometric pressure. This is the way they found to protect the pressure sensor, that would otherwise be exposed to water as soon as you run in a puddle. When barometric pressure drops, the tire enveloppe slighly swells and the pressure drops into the tire drops. I do not the relation between the pressure inside the tire and the pressure outside, but maybe it is close enough to 1:1 on that rather small range. As Griff500 rightly pointed out, the FBM320 has an absolute max rating of 10 bars, so they probably asked for a custom version with different gain/offset before ADC. They have to average the pressure over maybe 10s to keep only long term variations.

So to summarize: change of cyclist altitude causes change of barometric pressure, reflected by a change in tire pressure. That give a rate of climb/descent, and along with cc+bike weight gives one part of power. The other part of power required to overcome aerodynamic forces is estimated with cc speed, with is derived from the centripetal acceleration.

'hope this is not a too long and speculative post

Inventing things again? Griff500 never mentioned 10 bars anywhere, he said the supplier only rates it as far as 15psi, which is 1 Bar.

Also, as  I have been at pains to point out, the manufacturer clearly states that the device measures "the reaction force from the ground".  You seem to be telling him he is wrong and it works in a totally different way!

You will also have great difficulty getting a value for the elasticity of the tyre/tube combination, which determines the relationship between internal and external pressure.  

 

Griff500 posted this. Has Griff500 been busted using multiple logins, or does Griff500 usually refer to himself in the 3rd person?

davel would like to know.

davel wrote:

Griff500 posted this. Has Griff500 been busted using multiple logins, or does Griff500 usually refer to himself in the 3rd person? davel would like to know.

No, not usually, just this once  

I am thinking of holding a discussion with myself though, because i'd get more sense out of me than out of Jerome

jerome wrote:

I did not mean to be offensive (not native english speaker BTW), just to tell that lenghty speculative posts (not yours in particular) lead to nowhere. There is only one way we can see the device work right now (assuming it works) and this is not by measuring tire pressure, but measuring rate of climb/descent and guesstimating aerodynamic forces.

Sorry, I don't believe the climb / descent theory for one minute. If all you were doing was measuring altitude, why on earth would you mount the sensor on the wheel and have the additional complexity of a 670mm or so cyclic change of height?  Then of course there is the accelerometer. If you were using that to measure acceleration, why have it rotating?  You'd spend a lot of processing power just figuring out which way is up and which way is forward. These points, plus the fact that the Arofly press release I quoted says they are measuring tyre pressure, and the clear description they give of power measurement being a function of reaction against the ground, seem very clear. 

Strava already does exactly what you say based on digital map data, and does it well. Comparing my sector times with guys with power meters give remarkably close results, except when there is wind, which for Scottish cyclists like myself can sap as much power as hills. 

The App set-up sheds no light  the issue. They talk of switching from bike to bike and simply booting up bluetooth, yet there is no mention of providing weight figures or wheel sizes, or calibration. 

jerome wrote:

"What, no retraction of the smartass "got it?"  comment?"

Nop, not till you explain how you measure force (required for power) with only pressure.

No idea. Been wondering myself.

I'm just the guy whose researched what little there is out there on this and tried to correct a misconception. The insults are totally unjustified, but if you want to stand by your post, so be it.

What, no retraction of the smartass "got it?"  comment?

jerome wrote:

Would you trust the marketing blob or the schematic/datasheet?

"It is very common in novel applications [...] to personally test and characterise the component" not when there are components readily available, or when you overshoot the specified range 6 folds.

Neither. I'd trust the test I would do myself, and I've done similar things a number of times, sometimes in high reliability high accuracy applications.

This is often a matter of component manufacturers testing and underwriting against the standard commercial component standards and intended use, rather than going to the expense of testing to a range where the component is not envisaged to be used. If I am selling a barometer chip, why would I test and underwrite its performance to 110psi?

I'm not an expert on silicon pressure sensors, but a bit of basic research comes up with comments about the diaphragms used being not only resistant to extreme overpressure, but remaining perfectly elastic and relatively linear up to the point of fracture. Which kinda suggests if it doesn't break, you will get some sort of sense out of it. (This is very different to overtemperature where at 125 degC or so things can rapidly start to go off a cliff.) There is no point us  guessing if its good enough, because we have neither the data, nor the detailed knowledge of the algorithm or implementation these guys use.

jerome wrote:

They also mention a Pitot but there is no need for a Pitot to measure air pressure in the chamber, where there is no wind AFAIK.

Nowhere does it say there is a pitot tube. They say the sensor is based on F117 pitot tube technology, because pitot tubes are required measure small relative pressure changes!  They presumably mean an integrated pressure sensor, a/d convertor signal processor and serial bus on a chip as opposed to the old days of analogue pressure sensors in pitot tubes. (Your bike manufacturer may well brag about his aerospace grade carbon fibre, but it doesn't mean your bike will fly!)

jerome wrote:

To all those having a degree in anything: the pressure sensor does not measure tire pressure, and does not claim to. First, because it cannot do it (someone checked the datasheet), second because in order to convert it to a force, I would require the area of the tire patch. Someone to measure the tire patch realtime?

As mentionned above by IanMunro and others: the pressure sensor measure barometric pressure and infers a rate of climb together with speed, known with the accelerometer.

Got it?

"Ian Munro and others" clearly didn't read the Arofly blurb before guessing how it works.

Quote Arofly:

Algorithm and calibration technology

Via the tyre valve the Arofly measures tiny tyre pressure variances. According to producer Taiwan’s TBS Group, “the bicycle tyre is the first to know your pedaling power effectiveness by its reaction force from the ground. Through core and patented algorithm and advanced calibration technology the precise pedaling power is registered and with that the cycling performance. The Arofly is a hi-tech, precision ‘Pressure Sensor’ that turns tiny air pressure variances into digital data.”

"

Yes, I agree it sounds like BS, but let's see where they go with this.

It was me who checked the data sheet for the pressure sensor, and although as I said, it is only rated up to 15psi, that does not mean it turns into a pumpkin at higher pressures. It is very common in novel applications (Oil & Gas, process plants, miltary, submarine, polar exploration etc) to use an electronic component outwith the manufacturers rated range. It usually requires the designer to personally test and characterise the component under the conditions he intends to use it under.

tonyleatham wrote:

 

This aerofly device claims a massive operating temperature range and I just find it hard to believe that it's possible to pick out variations in pressure caused by application of force to the pedal from background variations caused by uneven road surfaces, pressure changes caused by heating and cooling of the braking surface etc. etc. I'm also not at all sure that you can reliably translate this into a meaningful power measurement. Potentially possible to get a relative measure (i.e. "you're pedalling harder now than you were"), but I'd not even bet on that.

I'm pretty sure that it's beyond my capabilities to design this device and get it working reliably to a point I'd by happy with.

 

While you were playing with toys, I was doing miltary signal processing.

To be fair, the FBM320 pressure sensor used appears to be noted for it's onboard temperature compensation capabilities. The fact it is only rated up to 15psi does however question how accurate it would be at the 90-100psi region many of us run our tyres at.

As for differentiating between road surface vibrations and pedal stroke?  Most road vibration is in the 10's of Hz range and random in nature. Power variation due to pedalling at a cadence of up to 90, would produce periodic variations of 0-3 Hz.  Any electronics or physics undergraduate could design a digital filter to cope with that in the timeframe of an exam question! Remember also that a conventional crank mounted pressure sensor already needs to be able to filter out a the vibration not filtered out by tyre and frame, (the lowest frequency stuff closest to pedal cycle in frequency) so filtering is still necessary here.

So assuming you could get the pressure reading, the signal processing is trivial.  The Pressure sensor used however doesn't appear to be up to use anywhere near tyre pressures.

Incidentally, for those arguing Ant versus BLE, the schematic also shows Ant2.

Griff500 wrote:

tonyleatham wrote:

 

This aerofly device claims a massive operating temperature range and I just find it hard to believe that it's possible to pick out variations in pressure caused by application of force to the pedal from background variations caused by uneven road surfaces, pressure changes caused by heating and cooling of the braking surface etc. etc. I'm also not at all sure that you can reliably translate this into a meaningful power measurement. Potentially possible to get a relative measure (i.e. "you're pedalling harder now than you were"), but I'd not even bet on that.

I'm pretty sure that it's beyond my capabilities to design this device and get it working reliably to a point I'd by happy with.

 

While you were playing with toys, I was doing miltary signal processing.

To be fair, the FBM320 pressure sensor used appears to be noted for it's onboard temperature compensation capabilities. The fact it is only rated up to 15psi does however question how accurate it would be at the 90-100psi region many of us run our tyres at.

As for differentiating between road surface vibrations and pedal stroke?  Most road vibration is in the 10's of Hz range and random in nature. Power variation due to pedalling at a cadence of up to 90, would produce periodic variations of 0-3 Hz.  Any electronics or physics undergraduate could design a digital filter to cope with that in the timeframe of an exam question! Remember also that a conventional crank mounted pressure sensor already needs to be able to filter out a the vibration not filtered out by tyre and frame, (the lowest frequency stuff closest to pedal cycle in frequency) so filtering is still necessary here.

So assuming you could get the pressure reading, the signal processing is trivial.  The Pressure sensor used however doesn't appear to be up to use anywhere near tyre pressures.

Incidentally, for those arguing Ant versus BLE, the schematic also shows Ant2.

 

 

I'm quite happy admitting that this device is not something I could design - I speak from experience of having worked on something similar. Part of the reason I think it's incredibly complicated is that the amplitude of the pressure variation caused by pedalling I suspect is incredibly small and in the real world, all assumptions about what can and can't be filtered usually don't survive the first test. Add to that the conversion time of the sensor is pretty slow (of the order of normal pedalling cadence), and I just don't think it's possible with this circuit.

You may scoff at my non-military experience, but I've worked both in defence (BAe) and consumer electronics, and by far the best people I worked with were outside of defence. 

If you feel it's something you could make work, do it and I'll buy one off you for $129.

tonyleatham wrote:

Griff500 wrote:

tonyleatham wrote:

 

This aerofly device claims a massive operating temperature range and I just find it hard to believe that it's possible to pick out variations in pressure caused by application of force to the pedal from background variations caused by uneven road surfaces, pressure changes caused by heating and cooling of the braking surface etc. etc. I'm also not at all sure that you can reliably translate this into a meaningful power measurement. Potentially possible to get a relative measure (i.e. "you're pedalling harder now than you were"), but I'd not even bet on that.

I'm pretty sure that it's beyond my capabilities to design this device and get it working reliably to a point I'd by happy with.

 

While you were playing with toys, I was doing miltary signal processing.

To be fair, the FBM320 pressure sensor used appears to be noted for it's onboard temperature compensation capabilities. The fact it is only rated up to 15psi does however question how accurate it would be at the 90-100psi region many of us run our tyres at.

As for differentiating between road surface vibrations and pedal stroke?  Most road vibration is in the 10's of Hz range and random in nature. Power variation due to pedalling at a cadence of up to 90, would produce periodic variations of 0-3 Hz.  Any electronics or physics undergraduate could design a digital filter to cope with that in the timeframe of an exam question! Remember also that a conventional crank mounted pressure sensor already needs to be able to filter out a the vibration not filtered out by tyre and frame, (the lowest frequency stuff closest to pedal cycle in frequency) so filtering is still necessary here.

So assuming you could get the pressure reading, the signal processing is trivial.  The Pressure sensor used however doesn't appear to be up to use anywhere near tyre pressures.

Incidentally, for those arguing Ant versus BLE, the schematic also shows Ant2.

 

 

I'm quite happy admitting that this device is not something I could design - I speak from experience of having worked on something similar. Part of the reason I think it's incredibly complicated is that the amplitude of the pressure variation caused by pedalling I suspect is incredibly small and in the real world, all assumptions about what can and can't be filtered usually don't survive the first test. Add to that the conversion time of the sensor is pretty slow (of the order of normal pedalling cadence), and I just don't think it's possible with this circuit.

You may scoff at my non-military experience, but I've worked both in defence (BAe) and consumer electronics, and by far the best people I worked with were outside of defence. 

If you feel it's something you could make work, do it and I'll buy one off you for $129.

Sorry about the scoff, but when someone opens a post by stating " I have a degree", they are setting themselves up!
I never said I could make it work for the price and size, I said the theory works but it's the wrong sensor. Read my post!

As for the pressure change being 'incredibly small", it is exactly the same change conventional power metres measure at the crank, you are just measuring it at a different point in the system. (Back to Newton 3rd).

Griff500 wrote:

tonyleatham wrote:

Griff500 wrote:

tonyleatham wrote:

 

This aerofly device claims a massive operating temperature range and I just find it hard to believe that it's possible to pick out variations in pressure caused by application of force to the pedal from background variations caused by uneven road surfaces, pressure changes caused by heating and cooling of the braking surface etc. etc. I'm also not at all sure that you can reliably translate this into a meaningful power measurement. Potentially possible to get a relative measure (i.e. "you're pedalling harder now than you were"), but I'd not even bet on that.

I'm pretty sure that it's beyond my capabilities to design this device and get it working reliably to a point I'd by happy with.

 

While you were playing with toys, I was doing miltary signal processing.

To be fair, the FBM320 pressure sensor used appears to be noted for it's onboard temperature compensation capabilities. The fact it is only rated up to 15psi does however question how accurate it would be at the 90-100psi region many of us run our tyres at.

As for differentiating between road surface vibrations and pedal stroke?  Most road vibration is in the 10's of Hz range and random in nature. Power variation due to pedalling at a cadence of up to 90, would produce periodic variations of 0-3 Hz.  Any electronics or physics undergraduate could design a digital filter to cope with that in the timeframe of an exam question! Remember also that a conventional crank mounted pressure sensor already needs to be able to filter out a the vibration not filtered out by tyre and frame, (the lowest frequency stuff closest to pedal cycle in frequency) so filtering is still necessary here.

So assuming you could get the pressure reading, the signal processing is trivial.  The Pressure sensor used however doesn't appear to be up to use anywhere near tyre pressures.

Incidentally, for those arguing Ant versus BLE, the schematic also shows Ant2.

 

 

I'm quite happy admitting that this device is not something I could design - I speak from experience of having worked on something similar. Part of the reason I think it's incredibly complicated is that the amplitude of the pressure variation caused by pedalling I suspect is incredibly small and in the real world, all assumptions about what can and can't be filtered usually don't survive the first test. Add to that the conversion time of the sensor is pretty slow (of the order of normal pedalling cadence), and I just don't think it's possible with this circuit.

You may scoff at my non-military experience, but I've worked both in defence (BAe) and consumer electronics, and by far the best people I worked with were outside of defence. 

If you feel it's something you could make work, do it and I'll buy one off you for $129.

Sorry about the scoff, but when someone opens a post by stating " I have a degree", they are setting themselves up! I never said I could make it work for the price and size, I said the theory works but it's the wrong sensor. Read my post! As for the pressure change being 'incredibly small", it is exactly the same change conventional power metres measure at the crank, you are just measuring it at a different point in the system. (Back to Newton 3rd).

Sorry my degree offends you - what would you rather I do? Pretend not to have one?

In my book, buggering about with minuscule variations in pressure reported by a slow sensor that no matter what the data sheet says will be prone to temperature drift is TOTALLY different to working with a fast strain gauge that will see a much bigger differential. I'm sure if you spend some time thinking about it, you'll get to this conclusion in the end. Or maybe playing with soldiers hasn't given you enough experience.