Pedal Assist Sensors

[adrian_sm]

Since most of Europe has laws requiring the rider be pedalling for any assistance to be provided, I thought I would try a PAS out for integration with my friction drive. This is the type of PAS I tried.


Now you would think that these very common off the shelf PAS sensors would be fairly universal in being able to mount to most bikes cranks. Wrong. They did not fit my Avanti mountain bike, Avanti road bike, Giant full suspension MTB, due to no axial clearance for the sensor ring and sensor, and not having an exposed region of the bottom bracket axle. It looks like it may fit an Electra Townie, and a real old road bike. Basically they are incompatible with most modern bike without significant modifications to the bike, which is against the whole ethos of my friction drive so I can not use this type of sensor.

Here are some installation photos I found on the web somewhere for reference.

PAS_sensor_ring_and_magnet_with_crank_removed.jpg (17.68 KiB) Viewed 687 times

PAS_Sensor_ring_and_magnet.jpg (14.1 KiB) Viewed 687 times

PAS_Sensor_behind_crank.jpg (28.36 KiB) Viewed 687 times

Since I am not interested in having the PAS sensor act as the throttle, just as a throttle enabler, I am tempted to use a different sensor setup. Basically another wheel sensor mounted on the chain stay and a magnet on the crank arm, like a cadence sensor. It will only be able to get a signal for every revolution of the cranks, so it would take two revolutions before the throttle can engage. Something more like this:


From looking at my little collection of bikes it appears that it would fit them all.

What do most people use for PAS sensors? Surely there are off-the-shelf retrofit options that work with modern bikes. What am I missing?

- Adrian

[d8veh]

There'e loads of options for your pedal sensor:
1. Get a longer BB set - about $12 for a cheap cartridge type.
2. Fit it in the non-chainwheel side if there's room there. You have to flip the magnet otherwise it works backwards.
3. T ake the magnets out and super-glue them directly to your chainwheel.

If your controller doesn't have a pedal sensor input, you can use a Speedict to process the signal with or without your normal throttle. and then you get loads of other useful functions and information.

[adrian_sm]

I was looking for an off the shelf PAS that actually fits modern bikes without needing to modify them, as my design goal is to not have to swap out or modify any of the standard bike parts when installing my friction drive. So
1) A different bottom bracket won't work.
2) It doesn't fit on the non-chainwheel side either, typically their is even less space there.
3) I am designing my own controller, so Speedict is nice but would double up on most features of my controller. Plus I believe they don't have a different PAS sensor than the one I showed. Page 7 of their user guide shows the same type of unit, and I can find no further details with my limitted search.

So still no solution.

[full-throttle]

Now you would think that these very common off the shelf PAS sensors would be fairly universal in being able to mount to most bikes cranks. Wrong.

The ones in the photos above will only fit a square taper bottom bracket, which are outdated these days. Well, at least on decent bikes

[d8veh]

I was looking for an off the shelf PAS that actually fits modern bikes without needing to modify them, as my design goal is to not have to swap out or modify any of the standard bike parts when installing my friction drive. So
1) A different bottom bracket won't work.
2) It doesn't fit on the non-chainwheel side either, typically their is even less space there.
3) I am designing my own controller, so Speedict is nice but would double up on most features of my controller. Plus I believe they don't have a different PAS sensor than the one I showed. Page 7 of their user guide shows the same type of unit, and I can find no further details with my limitted search.

So still no solution.

Sorry I didn't read properly, I thought that they were your bikes in the photos.
Speedict is a simple microprocessor with three inputs and one output. You can use any microcprocessor like an Arduino, PIC chip or the more user-friendly derivatives like the PICaxe or Genie chips, which can run off any 5v supply the same as a PAS sensor.

The setup that you proposed in the photo above will give a pulse forward and backwards (assuming that it's a reed switch) so that you'll take off when you're getting your pedals ready for start. You'd get a better result by using the PAS sensor or a hall sensor which are both uni-directional.

Another solution would be to glue the magnets to the chainwheel and make your own bracket for the sensor so that you can position it in the right place. This is the solution that people with hollow-tech cranks sometimes use when the standard stuff doesn't fit.

If you're making your own controller, why don't you make just the power side and use a microprocessor to do all the signal processing, which will give more flexibility?

[Lebowski]

These simple PAS 's don't detect torque, right ? It's just a simple magnets / reed switch combo ?

[adrian_sm]

@d8veh.
As I said previously/ I am looking for an off-the-shelf solution that does not require modifications or replacement of existing components.

So....
The PAS sensor above only suits old outdated square taper bottom brackets.
I don't want to glue on magents.
I don't want to have to replace the BB with a Thun BB.

Is there any other off-the-shelf PAS sensors that suit modern cranks? Anyone?

P.S. My controller is simply an RC style ESC, arduino uC, voltage/current sensor, wheel sensor, button & analog throttle inputs. So I am flexible in what the PAS sensor output would be.

[adrian_sm]

These simple PAS 's don't detect torque, right ? It's just a simple magnets / reed switch combo ?

Yep. The ones I showed above, simple tell you the cranks are spinning, and the direction. But not how hard you are pedaling.

I just want to be able to meet the EU laws that require the rider to be pedalling to get the electric assist. At the moment I am probably just going to put a magnet n the crank arm, and a sensor on the chain stay, then just measure cadence. That will be relatively universal, and easy to install, but I just assumed there would be a commercial solution already available.

P.S. Great work on getting the bike together. Now stop posting, and give it a spin around the block.

[justin_le]

@d8veh.
As I said previously/ I am looking for an off-the-shelf solution that does not require modifications or replacement of existing components.
So....
The PAS sensor above only suits old outdated square taper bottom brackets.
I don't want to glue on magents.
I don't want to have to replace the BB with a Thun BB.

Is there any other off-the-shelf PAS sensors that suit modern cranks? Anyone?

Hey Adrian, we've naturally run into this dilemma many times as well. Off the shelf PAS sensors are made for china bikes with square taper spindles and a single chainring. You can dril/file/hack them as necessary to fit on splined cranks or other setups, but even when you do have a suitable square taper spindle you still need to remove the crank and the bottom bracket nut in order to fit in place the sensor pickup, which means specialty bike tools most people don't have. The solution of a single magnet on the crank like you suggest can work, but is of pretty limited resolution when you are just starting off.

What we need is a magnet PAS ring that attaches inside the inner chainring, and is split in 2 semi-circular pieces so that you can insert it in place without even removing the crank arm. The pickup should attach to the downtube with a flexible mount that accommodates different tube diameters and that lets you adjust the sensor's relative position to the magnet ring. Ideally the ring should have 2 sets of magnets in quadrature, so that you can isolate forwards and reverse pedaling, and/or include a simple circuit that only transmits pulses with forward motion.

If somebody makes it, it will sell.

-Justin

[adrian_sm]

Thanks for the reply Justin. Looks like a cheap off-the-shelf solution is not available yet. Sigh.

What we need is a magnet PAS ring that attaches inside the inner chainring, and is split in 2 semi-circular pieces so that you can insert it in place without even removing the crank arm.

But what does it mount to? Modern cranks don't have any exposed axle, so it would have to be the chainring.
Hang on why make it two pieces, just make it one piece with a cutout. Then you need some way to fix it to the chain ring.
Obvious would be to use the chain ring mounts, as they are standardised. But that would still probably require using longer chainring nut/bolt things.
Hang-on, most of those nuts/bolts are magnetic (at least the ones I have). How about just get some 6mm diameter rare earth magnet, stick them on the inside of the chain ring bolts, then mount the sensor. That could work. Wonder if the magnets would stay on?

The pickup should attach to the downtube

Down tube size and shape vary drastically, I was thinking the seat tube would be better, ie. standardised in size and round most of the time. But some front derailuers brackets sit really low (like on my Gaint MTB dura ace) and would get in the way, so it may not be a universal solution. The sensor mount would need some degree of adjustment if an existing bracket is in the way. Maybe back to mounting it off the bottom bracket nut, not sure.

If somebody makes it, it will sell.

Not me. Got enough on my plate at the moment. Maybe we need to accidentally drops some plans at Wuxing, and see what happens.

[Lebowski]

I once thought about a homemade torque sensor, one which detects the actual torque you're providing.
What I came up with is a thick steel plate (or other material) which attached between the crank and the chainrings.
Normally the crank and chainrings are connected together using 4 or 5 bolts (let's assume 4 for the sake
of discussion)

DSC00778.jpg (97.22 KiB) Viewed 570 times

The cranks attach to the red holes, the chain rings to the green holes. The plate has slots in it
such that pressure on the cranks will bend the plate (in the planar direction) and change the
angle between the red and green holes. The plate has magnets (little blue squares) glued to it
such that the torque can be measured with hall sensors (torque changes the angle between the
magnets, a sensitive microprocessors can measure this based on speed and timing etc)

what do you guys think, should be simple, no ?

[adrian_sm]

Looks like it would make the chain ring floppy (side to side), hall sensor accuracy in that setup doesn't look great. I think a strain gauge based setup would be easier.

Plus the whole sensor on the spinning thing issue if getting the wires/signal back.

[Lebowski]

Looks like it would make the chain ring floppy (side to side), hall sensor accuracy in that setup doesn't look great. I think a strain gauge based setup would be easier.

Plus the whole sensor on the spinning thing issue if getting the wires/signal back.

no no the magnets are on the plate, the hall sensors are attached to the frame. About the being floppy, you can
detect extremely small timing differences with a processor. You could use a thick plate where the misallignement
due to the torque is hardly visible to the naked eye and still detect this... Or you could sandwich the bendy
plate between two sturdy plates with the sturdy ones only connected to the cranks

They don't sell strain gauges in the shop here so they're hard to get, plus then you get the wire issue.

[adrian_sm]

Ahhh. I get it now. Sounds like it would work. But a lot of hassle and parts. I think I would end up just buying a Thun BB.

[amberwolf]

They don't sell strain gauges in the shop here so they're hard to get

What about old digital scales?

[Warren]

Adrian,

You are discovering one of the huge frustrations of working on new bikes. Try mounting a "universal" bike rack on the back of almost any new bike.

Up until the 1990's it was pretty simple. In the last ten years, bottom bracket design has gone crazy. The last good idea was the cartridge bottom bracket to replace the old adjustable bearing design. Actually, this was done by the French decades ago, but didn't catch on, until the Japanese started doing it. Many fine quality bikes are still done this way. Then someone got the bright idea of getting rid of the square taper, crank attachment. This was a huge mess. Trendy, bike-shop, bikes offer a new system about ever two years now. No shop can stock all the different ones. All are inferior to the square taper, cartridge bottom bracket in one way or another. But they keep coming out with them. "New and improved" sells new bikes.

Warren, in the trenches at your local bike shop

[full-throttle]

Just a thought - could you say glue strong 4mm magnets inside the chainring bolts and use an off the shelf pickup, mounted using a couple of zip ties around one of the tubes perhaps..

[justin_le]

What we need is a magnet PAS ring that attaches inside the inner chainring, and is split in 2 semi-circular pieces so that you can insert it in place without even removing the crank arm.

But what does it mount to? Modern cranks don't have any exposed axle, so it would have to be the chainring.

Yes, I was thinking something that would pinch the inner chainring. So long as the pinching device was on an arm of sorts that could slide in and out to accommodate different chainring ID's then this would be fairly universal. The only annoyance is that it seems like there is equal distribution of 4 and 5 bolt chainrings, so you'd have to consider the spacings between these pinching arms to make it work on both.

Hang on why make it two pieces, just make it one piece with a cutout.

I thought of that too. I guess it depends on how tight of a resolution you want with the magnetic poles, since some of the splined spindles have large OD's and would require a decently wide cutout. Riding with the THUN sensor currently using 16 pulses per revolution, it really helps to facilitate an instant response from the PAS control that you can't get from say a 5 magnet ring.

Obvious would be to use the chain ring mounts, as they are standardised. But that would still probably require using longer chainring nut/bolt things.

Not to mention that even with the chainring mounts there are still quite a few "standards"
http://sheldonbrown.com/cribsheet-bcd.html

Hang-on, most of those nuts/bolts are magnetic (at least the ones I have). How about just get some 6mm diameter rare earth magnet, stick them on the inside of the chain ring bolts, then mount the sensor. That could work. Wonder if the magnets would stay on?

Most likely they would. Another approach would be to use magnets on the chainring bolts as a method of holding the PAS ring itself in place, and then use a smaller diameter but higher resolution set of magnets for the PAS pickup. With this approach you could have your plastic PAS piece include several sets of mounting magnet holes to account for the different 4 and 5 bolt hole standards, and then the user would simply pop and glue a set of magnets into the appropriate holes for their crankarm.

Down tube size and shape vary drastically, I was thinking the seat tube would be better, ie. standardised in size and round most of the time. But some front derailuers brackets sit really low (like on my Gaint MTB dura ace) and would get in the way, so it may not be a universal solution. The sensor mount would need some degree of adjustment if an existing bracket is in the way. Maybe back to mounting it off the bottom bracket nut, not sure.

Hmm, if there is one thing that is largely standardized it's the length and diameter of a bottom bracket shell. So if you had a pickup that uses two zip ties or narrow hose clamps going around either end of the bottom bracket tube, I think that would do it. It also forces a consistent radial distance from the axis so that there won't be alignment issues if you have a double set of magnets for quadrature encoding. Just need make sure there is a channel and space for the cable routing that often passes under the BB device.

-Justin

[Warren]

Justin,

"if there is one thing that is largely standardized it's the length and diameter of a bottom bracket shell."

You are a couple years behind. This is what is showing up on all the new bikes.

http://www.bb30standard.com/

Manufacturers love it because they don't have to thread the two sides of the shell. A huge cost savings. Most don't even have it bored for a bearing press fit. They use a plastic sleeve between the bearing and the bottom bracket shell. Cheap, simple, and prone to working loose in days.

Nothing to attach to. Nothing to put a ring under. Sorry.

Warren

[kfong]

That got me thinking, why not put a magnet on the spindle as a reference and a magnet on the extreme end of the cranks. You could weaken the cranks so they deflect and measure the difference in angles as torque if the hall sensors can detect such a shift. I'm not sure they can though, but it would solve the wiring issue. Resolution would be poor though.

I once thought about a homemade torque sensor, one which detects the actual torque you're providing.
What I came up with is a thick steel plate (or other material) which attached between the crank and the chainrings.
Normally the crank and chainrings are connected together using 4 or 5 bolts (let's assume 4 for the sake
of discussion)

DSC00778.jpg

The cranks attach to the red holes, the chain rings to the green holes. The plate has slots in it
such that pressure on the cranks will bend the plate (in the planar direction) and change the
angle between the red and green holes. The plate has magnets (little blue squares) glued to it
such that the torque can be measured with hall sensors (torque changes the angle between the
magnets, a sensitive microprocessors can measure this based on speed and timing etc)

what do you guys think, should be simple, no ?

[MattyCiii]

I wonder how hard it would be to put wireless strain gage into pedals? No modification to crank arm or Bb required. Some room in the pedal platform. Bolt on. Nice.

[kfong]

With bluetooth it would be very possible to do. Torque will only be seen on the down stroke unless you wear cleats. Milling out a custom crank would be easy to do. Should be room enough for a single lithium cell as well. Maybe both cranks to get a more consistent torque reading. I'll have to look into this when I have the time.

[justin_le]

I wonder how hard it would be to put wireless strain gage into pedals? No modification to crank arm or Bb required. Some room in the pedal platform. Bolt on. Nice.

Coming soon:
http://sites.garmin.com/vector/#power

I was going to do this as my senior design project at university a few years ago and it's good to this approach finally show up in the market. It make the most sense for modular fit with with any bicycle, and the relative motion between the pedal and crank can easily spin a micro generator that keeps the sensor circuit powered up and transmits the load data wirelessly to a computer on the bike.

Not quite so easy as a DIY project though.

[rkosiorek]

I wonder how hard it would be to put wireless strain gage into pedals? No modification to crank arm or Bb required. Some room in the pedal platform. Bolt on. Nice.

Coming soon:
http://sites.garmin.com/vector/#power

I was going to do this as my senior design project at university a few years ago and it's good to this approach finally show up in the market. It make the most sense for modular fit with with any bicycle, and the relative motion between the pedal and crank can easily spin a micro generator that keeps the sensor circuit powered up and transmits the load data wirelessly to a computer on the bike.

Not quite so easy as a DIY project though.

check out the price though, only $1499USD. but it is a neat idea.

not very difficult to install the strain inside the pedal. issue is how to transfer the measurement wirelessly and not have it look like those big chunky "Block" pedals on kids first trikes.

rick

[Lebowski]

you can also seek the solution in a totally different direction.
When you're doing normal light cycling your cadence and rotation
of the crank is very smooth, the rotation speed is nearly constant.
When you do heavy cycling the speed of the crank varies greatly
over one rotation of the cranks, when you push down with one foot
crank speed goes up, while you transition force from one leg
to the other crank speed goes down.

It's not fool proof of course but might work...