Rear Gear Generator

samatman

10 µW
Joined
Sep 3, 2022
Messages
5
Hi! Long-time lurker, first time poster.

A month ago I purchased a Class 3 cargo bike, rear hub with a cassette and
derailleur. Overall I'm very happy with this choice.

The bike, a Blix Packa Genie, has a cadence sensor and thumb throttle. I'll
end up replacing the throttle with a half-twist for comfort, but the real
problem is with pedaling.

The issue is that the bike is geared for torque, making it usable (ish...)
without power. The result is that at reasonable speeds, not even crusing
speed, there's no resistance on the chain at all. It's uncomfortable to spin
air for any length of time, and safe handling of a bike is much reduced when
the pedals have no bite, anyone who has done long downhill rides knows that
handling suffers when frame angle can't be adjusted with pedaling.

I could add a larger front cassette, maybe a front derailleur to go with it,
and I grant that this is the easier way to go, but long chain runs and two
derailleurs is a recipe for dropped chains in my previous Xtracycle
experience.

So I had an idea, and I'm looking for pointers to prior art, or to get talked
out of it. The idea is to have a single-speed freewheeling gear on the hub,
and one on the crank, with an electromagnet armature surrounding the rear
gear on both sides. There would be a full chain case to keep grime out, so
tolerances could be appropriately tight.

The gear ratio would be that at which I can actually move the bike forward
while pedaling briskly, the lowest practical in other words. When the bike
is moving faster than that, which is almost always, magnetic resistance is
applied to the rear cassette and the resulting current is sent to the
engine.

Instead of a gear shift, I would put in a selector which sets the target
resistance for the generator. It seems plausible to get 80% recovery of
effort as torque, which is terrible in comparison to 98.5% for a good chain,
but I'm thinking about it like this: if I'm putting 150W into a chain and
getting 120W out, that means I need 30W from a battery for it to feel like a
natural bike. If I'm having a full electrical failure then I'm getting picked
up, but rate-limiting the last 100Wh in the batteries will get me as far as I
need to go. The packs are rated for 1200kWh, so I figure that gets me 1kWh of
full range and a long tail of getting places at pushbike speed.

Measuring the transient power output, and putting a photodiode sensor between
the gear teeth, should give me a high-quality measurement of both torque
and cadence, at any speed high enough that the gear isn't turning the wheel
mechanically.

I have access to a makerspace and community, I can wind armatures and write
firmware, but I can't help asking myself why I've never seen this before.

There's no show-stoppers in theory, the Schaefller Free Drive claims 85%
efficiency, although I know that hasn't been tested in the wild.

For my purposes, I'd get most of what I wanted just dumping the power as heat,
I want the bike to handle like a bike, and I like pedaling fast and working up
a sweat sometimes. Maximum efficiency doesn't matter here, because my stoker
puts out 750W continuous, and I don't.

It seems like a promising design, because it just drops into the space a full
cassette would take. None of the mechanical strain is on the custom parts,
which can probably be 3-d printed, making this a practical DIY project for a
lot of people.

Am I missing something obvious? I looked for threads on the forum but any
plausible search terms overlap with things which one can buy, so if the needle
is in the haystack, I didn't find it.
 
blix-packa-genie-electric-bike-review-1200x600-c-default.jpg

https://www.youtube.com/watch?v=SPnM08ZTCyM&ab_channel=ElectricBikeReview.com

Might want to try a larger front sprocket before you get into a exotic solution.
 
The losses from the generator system, the conversion from the output of that to the input of the battery (or motor controller), etc., are going to be pretty significant. If the chain was 98% efficient, the generator was 80%, and the generator-to-system-voltage conversion was 80%, and the motor controller was 80%, and the motor itself was 80% (almost certainly all of these but the chain will be less than that), then since they are all series conversions the final efficiency would be .98 x .8 x .8 x .8 x .8, which is about 40% efficient. If you don't need a generator-to-system-voltae conversion, then it's a bit better at about 50%.

But either way you're losing something like half of your pedal power (probably more) as waste heat.

So I would recommend just driving the wheel directly from your pedal chain via whatever gearing you need, to use the pedal power as efficiently as you can, because you'll get (significantly) more range that way than by trying to convert pedal power to electrical power and then back to wheel power.

Another issue is that a generator is going to have a speed vs load point at which it is most efficient, so having something that will have to work at widely varying speeds is likely to be a lot less efficient than if you were to completely disconnect the pedal chain from the wheel, and just directly drive a generator at a constant cadence, via gearing chosen to keep the generator at it's optimal RPM.

(the http://ebikes.ca/tools/simulator.html has a "mode" for using a motor as a generator, if you want to play with that to see how it works; the instructions for that are down in the Simulator FAQ).


If the design of the bike and/or assist system chosen prevents you using multiple shiftable gears from pedals to wheel to efficiently use your pedal power directly, you might consider instead using an assist system that lets you redesign the chainline around letting the pedals do their work as needed.

One option is that if you have a long chainline, split it somewhere in the middle with an IGH with a suitable gear range for pedalling in your usage scenario. The pedals feed the input sprocket, and an output sprocket gets mounted to the spoke flange on the right side, just inboard of the input sprocket. If alignment from that flange to the rear wheel input is an issue, you can offset the whole IGH halfway between the point of perfect alignment both front and rear, and probably still have good enough alignment to not have chain derailment problems, especially if you are using chain covers that also guide the chain and/or keep it from coming off the sprockets.






samatman said:
So I had an idea, and I'm looking for pointers to prior art, or to get talkeg
out of it. The idea is to have a single-speed freewheeling gear on the hub,
and one on the crank, with an electromagnet armature surrounding the rear
gear on both sides. There would be a full chain case to keep grime out, so
tolerances could be appropriately tight.

The gear ratio would be that at which I can actually move the bike forward
while pedaling briskly, the lowest practical in other words. When the bike
is moving faster than that, which is almost always, magnetic resistance is
applied to the rear cassette and the resulting current is sent to the
engine.

Instead of a gear shift, I would put in a selector which sets the target
resistance for the generator. It seems plausible to get 80% recovery of
effort as torque, which is terrible in comparison to 98.5% for a good chain,
but I'm thinking about it like this: if I'm putting 150W into a chain and
getting 120W out, that means I need 30W from a battery for it to feel like a
natural bike. If I'm having a full electrical failure then I'm getting picked
up, but rate-limiting the last 100Wh in the batteries will get me as far as I
need to go. The packs are rated for 1200kWh, so I figure that gets me 1kWh of
full range and a long tail of getting places at pushbike speed.
 
amberwolf said:
One option is that if you have a long chainline, split it somewhere in the middle with an IGH with a suitable gear range for pedalling in your usage scenario. The pedals feed the input sprocket, and an output sprocket gets mounted to the spoke flange on the right side, just inboard of the input sprocket.

I have seen several posts about that concept but has anyone ever done it (excluding trikes)?
 
Yes; it's been done in at least a few LWB and MWB (and maybe some SWB) recumbents I've run across pictures and pages of (not here on ES, I think).

I dont' have links to any I specifically saw back when I was researching it (2009? 10? 11? can't remember), but here's a few I found just now:
This one seems to mostly be documenting using jackshafts somewhere in the chainline, rather than specifically an IGH, but at least some could've done it with the IGH there almost as easily:
http://www.bentrideronline.com/messageboard/archive/index.php?t-102340.html

Motoredbikes forum seems to have a lot of threads about doing it, often for their gasoline engine but sometimes it doesn't say if it's for that or their pedals, but most of the images are missing and the text doesn't usually have enough info to tell if they actually did it or not.

bentrider seems to have references to it but I couldn't find one that actually said it did it that way in the couple of minutes I spent.



I started to work out how to do it on CrazyBike2 with an old SA 3speed, but got distracted by stuff, then the housefire happened. Somewhere between them I was building a whole new frame for CB2 that would've had a NuVinci IGH in there. IIRC in both of them I was going to be building the IGH at the center of a rear swingarm pivot, and the "chaingrowth problem" was the reason for doing it that way vs a derailer at the rear wheel. (but its' been over a decade and I don't recall the details all that well).

The only one I've *actually* implemented it on so far is the SB Cruiser trike, though if I end up ever building the Cloudwalker bike like I want to, it will have to have one for pedal gearing as it doesn't have space for anything on/in the wheel(s).
 
amberwolf said:
The losses from the generator system, the conversion from the output of that to the input of the battery (or motor controller), etc., are going to be pretty significant [...] you're losing something like half of your pedal power (probably more) as waste heat.

Another issue is that a generator is going to have a speed vs load point at which it is most efficient, so having something that will have to work at widely varying speeds is likely to be a lot less efficient than if you were to completely disconnect the pedal chain from the wheel, and just directly drive a generator at a constant cadence, via gearing chosen to keep the generator at it's optimal RPM.

(the http://ebikes.ca/tools/simulator.html has a "mode" for using a motor as a generator, if you want to play with that to see how it works; the instructions for that are down in the Simulator FAQ).

Thanks for the reply!

I'm ok with losing most of the efficiency, because my motive here isn't to propel the bike with leg power. I went into a bit of detail about that in the first post, briefly: I want an electric vehicle which behaves more like a pushbike, as a matter of user-interface, rather than caring much about capturing leg power at high wheel RPM. I could literally dump all the resistance as heat and get 90% of what I want out of this conversion.

So I'm mainly looking for prior art for a specific design, whether it's a great idea or not. There are some very quirky designs out there.

I maybe didn't get the idea across, because the whole idea is to wind it for efficiency at a steady but not punishing cadence. The rotor is the rear sprocket itself, with stators on both sides, flux introduced with electromagnets, possibly mount some permanents on the sprocket if that's necessary. I just checked, and a normal sprocket takes a magnet with considerable grip, so whatever alloy is involved is very much ferromagnetic, it should serve as at least an adequate load in such a configuration.

So such a generator would operate in pedal cadence ranges, with a complete cage around the drive chain the stators could be very close to the sprocket. That might buy me some efficiency, but I look at it this way: losing 50% of 150W of human energy means I need to draw 75W from the battery, the motor is a 750W so that's an 800W draw. This is not a problem. With the right firmware, a low-power 100W draw would give me a little extra kick and an hour's extended range for 100Wh of the 1200Wh nominal in the packs.
 
LewTwo said:
blix-packa-genie-electric-bike-review-1200x600-c-default.jpg

https://www.youtube.com/watch?v=SPnM08ZTCyM&ab_channel=ElectricBikeReview.com

Might want to try a larger front sprocket before you get into a exotic solution.

The bike loses traction on the pedals around 16mph and hits a cruising speed of 28mph unloaded/level.

That implies 1.75 teeth per one tooth on the current front sprocket, so closer to twice the diameter than not. I'm fairly sure that would bite the ground when taking a sharp right.

I might well add a reasonably-larger front sprocket, which would help, but it wouldn't solve the problem, not really.
 
samatman said:
Hi! Long-time lurker, first time poster.
The issue is that the bike is geared for torque, making it usable (ish...) without power. The result is that at reasonable speeds, not even crusing speed, there's no resistance on the chain at all.
I could add a larger front cassette, maybe a front derailleur to go with it, and I grant that this is the easier way to go, but long chain runs and two derailleurs is a recipe for dropped chains in my previous Xtracycle experience.
samatman,

A second chainring and front derailleur would be my choice. It will be cheaper, lighter, more efficient, and easily "tunable" if you need changes later. As for dropped chains, you can get "chainguards" which are just a flat round dish that mounts outside of the chainring, and a second "chainguard" that mounts on the inside chainring. That should make the the chain stay engaged, because there is literally no place for the chain to go except onto a chainring. Make sure the two chainrings are close together, enough to prevent the chain from getting in between the two chainrings. The bike frame and the rear wheel should do the same job for the rear cassette, but you can put a "chainguard" between the rear cassette and the rear wheel if needed, also.

The complexity and mechanical drag of your suggestion will be costly, both in money and in range available. I regard the "waste heat" option as serious battery-pack abuse.
I don't know if your bike has a front freewheel at the pedal-crank, but if so, then to me the simplest possible answer is to let the motor do the work whenever the motor outruns you; don't even try to pedal unless the bike speed is low enough to do that comfortably. I have seen YouTube videos about adding a front freewheel to the pedal-crank, if you need that. The ebike-builder stores can help with a pedal-crank freewheel, or they may have bolt-in solution for you there.
 
red said:
The complexity and mechanical drag of your suggestion will be costly, both in money and in range available. I regard the "waste heat" option as serious battery-pack abuse.

There would be no mechanical drag. Perhaps a bit of magnetic reluctance when the armature is completely unpowered, due to magnetizing the sprocket, but that's okay. Stator: around the sprocket. Rotor: freewheeling rear sprocket. Mechanical drag: none.

I have no idea why one would use the battery pack as a heat sink, sounds risky and expensive.

If I were inclined (I'm not) to dispose of the current generated by the proposed motor as heat, I'd wire it to,
I dunno, a big ol' bolt. A many-many Ohm resistor.

What would that look like? Well, it would be an electric motorcycle basically, with controllable resistance on the pedals. I could measure this resistance, the generated current, and use it as a strain sensor for controlling the motor. So as the rider, that would feel like a bike that goes fast, and keeps going fast when I set the throttle and stop pedaling. What we'd have there is a hybrid motorcycle and stationary exercise bike, but not stationary.

I would be fine with this, in fact. There's just no point in wasting the current, which can be applied to the motor, that circuitry is not the difficult part of wiring this up, that's winding the armatures to 1-2 mm of tolerance and designing the controller. I consider using that current a nice bonus. I'm already mostly pedaling to drive a motor, or just cruising on throttle; that's not what this mod would change.

I have real trouble believing that this has never been built, I know too many mechanical engineers. If no one around here has ever seen it, well, don't know where else to ask.
 
The only similar thing I've ever seen built is simply connecting the pedal chain to a motor being used as a generator (or a generator designed for the purpose), instead of connecting it to the wheel at all.

Usually the output of that is fed via some conversion electronics to the battery, and is then typically called a series hybrid (vs parallel hybrid where the pedals have a separate power path to the wheel(s) via the mechanical drivetrain).


Sometimes there is adjustability to the load via the conversion electronics.

Those electronics may vary widely from system to system depending on what it is intended to do.

Just some ideas / thoughts:

A simple brushed DC motor/generator with brushed controller used as a PWM control (to select and/or vary the current allowed from the generator to the battery or load) is one of the simplest.

A more complex version could use a cadence sensor on the pedals to vary the PWM control to adjust the loading of the motor based on pedal RPM, or a torque sensor on the pedals to vary the PWM to adjust and maintain a continuous regular torque on them within the limits of the ability of the conversion electronics to vary the load. The limits are wider if using a resistive load to just waste the power as heat, but if the conversion electronics have a wide range voltage input and output, it can probably do it even by feeding the battery charge port.
 
That I've seen as well. There's room in the Packa Genie's frame for a mid-drive, and I've given some thought to making that mod instead, haven't ruled it out.

If I were going for pure resistance, I might want to make or modify a rigid chain tensioning wheel and surround it with electromagnets. Put a photodiode between the teeth for cadence, let the heat dump into the sprocket and magnets, measure the resistance to get a strain reading, adjusting the current into the electromagnets to keep a steady resistance which I can dial up and down.

Which, swap the separate chain tensioner for the rear sprocket, add a second set of windings to drive the current off somewhere useful, and that's what I've got in mind.

There are a couple months in my climate where riding is something I don't plan to do, I've got the interest and the necessary tools, so who knows. Might be a short thread, might be a long one.
 
The gear ratio would be that at which I can actually move the bike forward
while pedaling briskly, the lowest practical in other words.
You did not mention how many 'speeds' you have on the rear cassette.

Put a 10 or 11 speed very wide range mountain bike cassette and appropriate derailleur on. Choose a front sprocket that gives you the feel you want in high gears, and only use the lowest 2 or 3 rear sprockets when your motor fails and you need to get home.

You have a 7 or 8 speed with a motor to smooth out the gaps, and some unused very low ratios for emergencies. Efficiency remains in the 'chain drive' range, which is generally better than anything else.
 
Back
Top