'Interfacing' with a bicycle - sensing braking [solved]

Hi,

I need some inspiration.

I need a simple way to sense when I'm pulling the brake lever on a bike. Ideally with some sort of feedback on how large the tension in the wire actually is.

So far I have thought of this:

* Hall sensor at the brake lever

It's simple enough, doesn't need re-calibration as the brake pads wear down, but I would have to glue the sensors to the handles. Potentially looks messy. To get some sort of feedback on the wire tension, I'd have to use multiple sensors.

* Load cell inserted into the brake wire

That's a massive safety issue and I really don't like it. If I could deal with the signal (and fluctuations due to vibration), I'd have a real feedback of the tension in the wire.

* Load cell in parallel with the brake wire

I suspect the signal would be even smaller, much more SNR problems. Not ideal at all.

* Hall sensor underneath the brake wire + magnet attached to the wire

Would need some sort of guide to keep the wire from oscillating side to side. Would also need re-calibration as the brake pads wear down. Not good.

* Direct measurement of the wire tension

Similar to this geometry. To keep it small, I guess I could only use a tactile switch instead of a real force sensor. Would need proper initial calibration (moving parts), possibly issues with the switch after a while.

* Accelerometer

This would be the last way out, but I really don't like it. It only works if the velocity of the bike is > 0.

What else could be used?

It should be safe, simple, clean and reliable.

Assuming the brakes work, why not measure the rotation speed of the wheels with a rotary encoder?

That only works reliably when braking hard enough, and also not with v = 0.

The goal is to trigger an 'action' as soon as the brakes are used, not when the effect of using them becomes apparent (or not).

how about a pressure sensor that is located on the frame tube. so the sensor would be sandwiched between the frame and the cable housing, the harder you squeeze, the more the cable wants to straighten, reading a higher pressure on the sensor.

take a look at how a cable tension meter works. it figure the tension based on the resistance to be straight.

load cell might be the a good way to go, just look for one that is rated in the correct range. something else will break before the load cell if it sized correctly.

good luck

That only works reliably when braking hard enough, and also not with v = 0.

The goal is to trigger an 'action' as soon as the brakes are used, not when the effect of using them becomes apparent (or not).

The brake/handle has a rotation point => apply a potmeter there to see how hard the brake/handle is pinched
(does not say anything about the cable tension yet

you could add potmeters to the derailleur to "reads" its position.

Adding a potentiometer would be nice, but the axle is just a big bolt (riveted in place), which doesn't turn at all.

The load-cell/strain gage approach is too costly. Everything I can find is (of course) for industrial applications and costs a fortune or "won't be sold" to individuals in single quantity.

I think I will have to use the hall sensor at the brake handle approach, or shift the sensor evaluation back into my own brain and just push a button. The last option would give me first hand information about the tension as well (haha).

The brake pads are mounted on arms that make a scissor-like action when closed, yes?
Something like this:

So how about a reading of something that can slide back & forth as the lower arm gets pulled up to the cable holder on the frame,
or, add an arm up from each brake pad screw/bolt and measure the distance between the tops of the two arms, either optically, with a linear pot, etc.

I have V-brakes, so there's no axle there. Another war of the systems I guess.

But I think adding a linear potentiometer underneath the brake cable (where it runs in parallel to the top beam of the frame) might be a simple thing. I found some on mouser that are supposed to last 15k cycles. I usually don't use the brake much, so it might be ok :wink:

More like this?
Same idea would apply - you have 2 parts moving together & apart, mount one part firm and let the other move back & forth.
If easier to attach to a cable somewhere using the frame to support things, then go for that.

A magnet and a hall effect like an A1301 might be interesting for you - Arduino Playground - Hall Effect -
Probably needs some calibration, recall magnetic fields loose strength with 3rd power of distance so its range is not too far, and it might be affected by all the metal nearby.

Adhesive strain guage stuck onto brake lever or other part?

I've ordered a couple of linear pots for testing. If I can hook them up to the brake cable and they don't require too much force to operate everything is fine. They're quite affordable as well.

Strain gages would be nice as well, but I have my doubts regarding reliability. These things need to be properly glued to a surface and covered up well. Definitely on a bike that gets wet and will see quite a lot of dirt / salt in the winter. Otherwise Mr. Oxi Dation would take care of them quickly. And I don't have a cheap source for them as well, as I don't need 1000s.

Got the linear pots today. They're quite nice. The force required to operate them seems low enough to not compromise the proper working of the brake's 'return spring' - if that's a valid term. Total travel distance is 6cm, which is plenty enough. Now I need some sugru to glue it to the frame. Can't wait to get a local source for this stuff.

If you want to try yet another sensor, consider using a (hacked) optical mouse.
they are good at sensing movement, and need no connection to the moving part.
inspiration : http://www.bidouille.org/hack/mousecam

That would be great in a clean and enclosed environment, but optics is a tricky business when dirt and vibration come into play. Unfortunately I know this from excessive experience... Ideally I'd have to build a small enclosure (with special 'brushes' to keep the dust out) that securely holds the sensor and guides the cable. I have a tiny mill that could deal with aluminium, but it's not a CNC one. I might be able to make one such enclosure by hand with a lot of tranquillizer pills, but I'm pretty sure I'd mess it up somewhere in the process. I could do a conversion to CNC, but that would take more time... I'll have enough time for this when I'm dead I think. Then there's 3D printing (ponoko does it too I think), but waiting for at least 1-2 weeks per iteration... nah.

I think I'll go for the quick and dirty hack first.

I made a bike brake counter using a linear potentiometer and had fun building it. My uno checks the pot and when triggered by position change, writes date-time and value out to a SD file. The potentiometer is mounted along the bike top tube and clamps the brake cable with a brass wire stop nut. Easy to put on - take off the bike frame and no problems with brake interference...but my uno-SD hardware needs a better case and battery mount. I've not analyzed my data yet - other than to check that the numbers seem valid. My intent was to gather data to predict "brake pad wear" - not likely to happen quite yet, but good fun just getting the data so far.

sensor.jpg

sensorOnBike.jpg

That's NICE.

Was the case a custom milling job?

A switch for brake sensing comes with this cheap brake light kit:
http://www.dealextreme.com/p/3-in-1-electronic-bike-horns-with-brake-light-and-turning-signals-3631

I was tempted to buy it just for the switch! I look forward to hearing how you do with the linear pots.

custom milling? no, the case is just sheet styrene (.080 ") with gray primer. I glued together a basic box to hold the potentiometer plus longer tangs to attach the frame mounts. Used flexible bike light straps with extra long bolts to go around the larger frame tube. To mount and use, I have to undo each strap and thread it underneath the brake cable into position on the frame, then align the wire stop screw on the pot arm onto the brake cable and snug it down - easier than it sounds. But the whole rig is custom to my bike for sure.

sensor2.jpg

If you are using v-brakes, why not just put a flex sensor across the two arms (where the rubber boot is in the pic below):