Home made pedelec/pedal torque sensor

[Jeremy]

My experiments with the Crystalyte 405 motor on my recumbent have been interesting, but have convinced me that an "automatic" system would be a lot simpler to ride. I've been giving this some thought and have come up with the following simple scheme.

Right under the front edge of the seat there is an idler wheel that supports the top of the chain. If I were to add another, spring loaded, idler wheel above and in front of this, then increasing pedal effort would tend to move this idler upwards. It would be a simple matter to turn this small movement into a signal that could be processed to drive the controller throttle input.

The signal would need to be modified by a pedal speed signal, to prevent the controller giving power when the bike is stationary but with pressure on the pedals (which is common on a 'bent, due to the way you have to start off from a standstill, with the brakes on).

My thoughts are to measure chain speed at the idler, together with chain deflection (pedal torque) and combine these signals into a control voltage for the controller. This would be very easy to do using a small microcontroller, which would also allow the selection of varying levels of "gain" in the system, so effectively controlling the assistance level per unit pedal torque.

Overall, the system should be fairly straightforward to build and may well work on a conventional bike as well. The fact that there doesn't seem to be an add-on like this on the market makes me wonder if I'm on the right track with this. Any comments or observations would be welcomed.

Jeremy

 

[flecc]

That's certainly plausible Jeremy. There would be a small loss of efficiency due to the extra idler and the spring loading on it of course, but I doubt it would be serious.

The only question mark I have is with chain "thrashing". As you know, it can happen under certain circumstances and the incidence tends to rise on longer chains. If it occurred it could lead to rapid on/offs of the sensing signal. It's unlikely while pedal force is applied of course, but might happen in the "in between" states, giving a jerky on/off drive momentarily.
.

 

[Jeremy]

Thanks for that, Flecc. I think I should be able to compensate to some extent for the "bounce", provided that it doesn't get too severe, by filtering the response in the software. It would be nice to have a direct pedal force measure, but that would be difficult to engineer as an add-on.

I have thought about fitting strain gauges to the cranks, with wireless links back to the microcontroller, as this would then give cadence and pedal force data, but it looks like a big job to sort out the mechanical bits.

One way to do this might be to machine up a new pedal shaft (my guess is that measuring one pedal is good enough). This pedal shaft could be strain gauged to measure pedal force and it might be possible to build in a small PM generator to provide power for the electronics from pedal rotation. A simple wireless link could then pass the pedal strain data to the microcontroller.

For my recumbent it might even be possible to just crudely strain gauge the front tube that supports the "bottom bracket". The outward forces on this will be fairly representative of pedal force, but my guess is that the strain range will be very low.

Plenty of things to play with over the long Winter evenings!

Jeremy

 

[Miles]

Hi Jeremy,

Here are a few links to show the different approaches taken by the developers of bicycle power meters:

SRM
ERGOMO.DE: the brilliant bike computer
Saris Cycling Group
Bolton Engineering Project Notes
Technical Information
Power Meter Review

Miles

 

[Jeremy]

Many thanks for those links, Miles. I hadn't thought to look for info about power meter systems. The chain tension one, using chain natural frequency, looks a good start. For my needs I don't need a precisely calibrated output, just a signal that is broadly proportional to applied pedal force, so I think a few experiments are needed with a guitar pickup type sensor!

Jeremy

 

[mr-motorvator]

Pedal Torque Measurement

I'm normally a reader rather than participant, but this is a very interesting topic. I am designing a system to use as a retrofit with a hub motor: I have various samples from suppliers including the 'standard' chainwheel rotation/direction sensor (PAS), a chainwheel that incorporates a strain gauge and rotation sensor (PFS) and a (crude) chain deflection sensor.

Unfortunately in its current form the PFS is not really suitable as a retrofit because of its dimensions, weight and # teeth.

I am currently following the PAS + chain deflection route, and have a prototype design that measures both into a PIC micro to produce a power demand signal.

I would welcome more discussion, but I am not free openly to discuss algorithms because of their commercial sensitivity. If you would like to exchange emails and maybe meet up (I am in Marlborough and spent several months in Salisbury DoCSU) that would be great.

Tony Castles
tony@mr-motorvator.co.uk

 

[daniel.weck]

Bump !

I wonder where to source this kind of sensor from ? (it's not based on strain guages, instead they use Hall-effect sensors with dual magnet arrays to analyze differentials induced by slight torsional changes in the elastic material):

- EDIT - The name of the manufacturer/supplier in China is JC-EBike. Scroll-down or click here for more information: http://www.pedelecs.co.uk/forum/electric-bicycles/1172-home-made-pedelec-pedal-torque-sensor.html#post62520

Scroll to item "M25" at Island Earth - Save the planet hug an EV !

...or check this post on Endless-Sphere: Endless-sphere.com • View topic - Pedal Assist Sensors

 

[daniel.weck]

Some iZip e-bikes include a single chainring with built-in cadence and torque sensor (it looks the same as the above generic chinese chainring/crankset addon). Maybe the part could be reused with generic controllers, with minimal electronics fiddling:

- EDIT - image credits: What’s New at Currie Technologies - IZIP & EZIP » Blog Archive » PROJECT IZIP

 

[daniel.weck]

Ah, some e-bikes use the rear-dropout TMM technology instead of the crankset system (the new "iZip Enlightened", the "Giant Freedom" bike, the PoweredBicycles "Mantra" and "City Thunder" brands e-bikes, some "Gazelle" bikes, the "Sparta Ion"):

"Torque Measurement System

TMM4 Torque Measurement Technology reads subtle changes in chain line force and drives the motor for pedal assist power."

IDbike | TMM sensor unit

TranzX Power Support Technology

Izip Electric Bikes | Izip Electric Mountain Bikes | Izip Bikes Parts | izip hybrid electric bicycle

Electric Bikes By Poweredbicycles - Aluminium Alloy Frame - Lightweight - Mantra Ultra Light Torque Measurement Method ( TMM ) New 2010 Model

Powered bikes take the strain out of travel

UPDATE:

Can be purchased here:

Abus Granit, Fahrradzubehör, Fahrradschloss, Thule Ersatzteile, Klickfix, Fahrradteile, Fahrrad Zubehör

 

[daniel.weck]

Ah, I have found the manufacturer / main supplier in China. It's a firm called "Suzhou Victory Sincerity Technology Co., Ltd":

JC-EBike.com (Chino-English)

JC-EBike.com products overview (Chino-English)

...not that it really helps my much anyway

As fitted on the 720-Cycles Rush Trek:

 

[daniel.weck]

Urban Mover e-bikes also use the crankset / chainring torque sensor technology:

 

[daniel.weck]

I read comments (on here actually) that the UM torque sensing was kind of on/off instead of a smooth incremental drive. Getting the electronics right is probably no small task... Anyway, here's a video that demonstrates torque-sensing pretty well:

YouTube - PAS BICYCLE

 

[flecc]

The Giant Freedom e-bikes use the rear drop-out sensor system, exactly like the Izip one. A to B raised one
criticism of it, namely that it's very exposed to dirt and corrosion.
.

 

[daniel.weck]

An Ananda-based motor kit, with crankset/chainring torque sensor:

- EDIT - they have now stopped selling the torque-sensor component due to inconsistent quality and function. I wonder how iZip / Urban Mover manage to make something good out of the Chinese kits ?

Base information about electric bicycle(e-bike) conversion kit

 

[daniel.weck]

Bottom bracket integration

300+ GBP

ergomo sport - a new dimension in cycling - the sensor

 

[daniel.weck]

The Giant Freedom e-bikes use the rear drop-out sensor system, exactly like the Izip one. A to B raised one
criticism of it, namely that it's very exposed to dirt and corrosion.
.

Ok. Apparently several Gazelle models and the Sparta Ion are also equipped with the TMM4.

 

[daniel.weck]

Ah, I have found the manufacturer / main supplier in China. It's a firm called "Suzhou Victory Sincerity Technology Co., Ltd":

JC-EBike.com (Chino-English)

JC-EBike.com products overview (Chino-English)

Ananda also supply a similar device:

Intelligent PAS sensor for electric bike Manufacturer exporting direct from Shanghai, China

Intelligent torque sensor for electric bike Manufacturer exporting direct from Shanghai, China

PAS sensor for electric bike Manufacturer exporting direct from Shanghai, China

 

[emissions-free]

There's another company that goes by the name of Dapu, Japanese company but have manufacturing in China that uses a similar system to the above. I've tried it and to be honest I didn't notice any difference over the standard type crank speed sensor. I've also heard they can be unreliable. In my opinion the power needs to be fed in much more gently on pretty much every controller I've tried and that's more to do with the controller setup than the PAS sensor IMO.

I'm pretty sure that crank sensor is not manufactured by Ananda. It will just be bought in, as are their throttle, crank PAS sensor and e bike brakes. They manufacture controllers and motors. I've been to their offices and have a few of their motors that I'm testing.

 

[Tiberius]

Good work, Daniel, thanks.

Have you managed to find any datasheets or prices (apart from that silly one)?
I'd like to put together an automatic system along the lines that Jeremy was musing about.

Nick

BTW. Jeremy is no longer on the forum, but I am in touch with him occasionally.

 

[daniel.weck]

Thank you guys.

I have talked to Frank Scheftlein ( Das Elektrorad von Frank Scheftlein: Komplette Pedelec- / Elektrofahrrad Umbausätze und Ersatzteile von Bafang, Puma (eZee), Tongxin u.a. )( http://www.pedelecs.co.uk/forum/members/mechaniker.html ) ( Pedelec-Forum ) who wants to supply torque sensors in the future. He has no time now, but will be looking into it in January, once his baseline business is running smoothly from China (he is currently sorting out the payment and shipping infrastructure, I think).

Frank made a comment about the current quality of crankset / chainring torque sensors, and he is not happy with the quality, thus why he has not been supplying them. I have received disappointing remarks from another kit provider, they have now stopped selling the force sensor due to the poor function.

Based on the feedback so far, it looks like I should wait until a better kit comes out. There's no point wasting money or time on average hardware and inadequate software (the controller firmware algorithm must efficiently use cadence+torque information).

Any other information welcome.
Cheers, Daniel