I've been having an email conversation with a knowledgeable friend of mine with significantly more mechanical repair and design experience than I have, and quite a bit of thought came up as a result of it, so this post will discuss the points made in it. Depending on how the balance of advantage/disadvantage works out after further thought and comment from you, the readers, I might do something along these lines instead.
I looked at some of your design images. Why not move the motor and second gear cluster to the rear wheel? have one set (the human driven) on the right, and the motor driven set on the left?I had not yet considered that option, but there's two big reasons not to:
First, derailers have handed-ness. Meaning, they're made to be mounted on only one side. All of mine are meant only for the right side. It's why my latest blog post "Oops" had to be made, because I *knew* that but had not thought about it when making up the 3D model and the temporary physical mockup, because originally I was going to do things a bit differently with the wheel mounting frame, and thus would have had everything still on the relative "left" side of that frame, since it would have been essentially completely backwards, wheel, frame, and all. But I can't do that because I found the sprockets would end up crunching on the main bike frame when I turn, and/or chew up cabling (electrical or mechanical) that jiggles too close to it by accident somehow (since currently there's no cover over the chain/sprockets). It's possible that some of the derailers out there are meant for left-side mounting, but I haven't seen any on bikes I've actually had or worked on, or in the few bike shops online I looked at via google. Either way, if they do exist, I'd have to buy them, and that's something to avoid where possible. There's no way to modify the existing ones because of the way they work inside, as far as I can tell. Certainly *I* could not do it, as there are parts inside (made of steel) that are one-way fits, and could not be assembled in the opposite direction, even if I could take the whole derailer apart and put it back together in mirrored fashion.
If I did not intend to be able to shift the motor gears to make it more efficient for speed vs torque, to get the most out of my battery life beyond what the PWM controller will do, I would not need to worry about the derailers, and would simply need to make the chain shorter so it would have the proper tension without the derailer/tensioner on the rear. But it would still leave the second problem.
The second is that I would have to disassemble the rear wheel, and manufacture a new hub out of the existing one that could hold the spokes from the left side more toward the center, as the right side does (they're not symmetrical currently), so that there would be room to add the sprockets necessary for the left side drivetrain. Otherwise there is no place to put the sprockets to drive the rear wheel from the motor, and that is a serious flaw. :-)
A few minutes more thinking after typing up the above leads to the possiblity of only a single rear sprocket "welded" (bolted, glued, etc) in some way to the spoke mounting plate just inside the bolt/fork assembly, which would fit and clear all the necessary parts around it so the chain would not catch on things and still drive the wheel. It eliminates the sprocket ratchet, because now the chain is directly connected to the wheel, and coasting will cause the wheel to drive the chain instead of the chain driving the wheel, which has several consequences.
- An advantage of not having a sprocket ratchet is that it would allow Regen braking to work, which it will not if I use the current system, that takes advantage of the ratchet system inside the wheel's sprocket hub that allows you to coast when not inputting any energy, without rotating the input device, because energy is only fed into the wheel, never back out of it to the input device, a necessary thing for Regen. Since I'm currently using a motor with a worm-gear reduction system for testing, it's irrelevant, because it is not possible to feed energy back into the system that way, either.
- A further advantage is that it eliminates the rear derailer/tensioner, so it breaks down one part of the first objection--no need to find a left-side derailer/tensioner.
- A disadvantage of not having a sprocket ratchet is that I could not use any gearbox that has a worm gear in it, because that would lock up the entire drive train if the motor is not being powered, and my max speed would be limited to whatever the motor was spinning at.
- A further disadvantage of not having a sprocket ratchet is that the motor would always be "in circuit" mechanically, and thus always placing a physical load on my pedalling. Even though I can (and would anyway) make a switch to manually or automatically electrically disconnect the motor from the rest of the system when it is not being powered, which would remove most of the resistance/braking that would occur, it would still cause some extra resistance for the bearings and such involved--more than the sprocket ratchet in the wheel does, which is currently all I'd have to fight against. Depending on how the gearbox is attached to the motor-end drive sprocket, it might be possible to lever it out of contact with the drive sprocket so it is thus out of circuit and eliminating this problem, but that makes the connection to the sprocket more complex, since it would have to detach. I already do have some ideas about that, though, because of how I am thinking of doing this connection already.
- Another disadvantage is that not having multiple rear sprockets and a rear derailer prevents any possible rear shifting, severely reducing the amount of help I can give the motor in matching it's output to the requirements of the speed I want at any particular time, at the most efficient ratio. I would only have two speeds, because that is all the sprockets I have available on the motor drive section. Better than only one speed, but not enough for what I want to do.
If I totally trash the idea of shifting at all, and only use motors with wheel gears and no worm gears, then it is relatively easy to put all the drivetrain on the rear left side. It might come down to doing just that, but it's a bit boring because it doesn't have any real change from how I've seen many project bikes done, in that I don't recall seeing any of them yet that have a way to "change gears" for the motor separately from the pedal system, without a custom-manufactured reduction gear set on the motor itself. I don't exactly have this as a specific goal, but it'd be interesting to make this project in such a way that anyone with the desire could build it themselves from the junk they have around, too, like I am trying to. I know there will be parts I must actually make or have made with the current preliminary design, such as the blocks to go between the fork and the attached frame section, that won't be makeable with just a hacksaw, :-) but there may even be alternate solutions to *that* I haven't thought of yet.
I bring this up because of the issues you might have with steering under power. That motor might cause some interesting precession if you load it during a turn.
Possibly true, which I've thought about, but can't test until I actually finish building some basic test version of the setup. Currently, because it's a power window crank motor with gearbox, the motor will be longitudinal to the body of the bike during normal forward motion, but that will change during a turn. I don't remember which direction the motor itself turns inside it's casing to drive the gearbox, so I dont' know which direction would have more trouble turning. It's currently a pretty small motor, and has low mass, so the centrifugal / gyroscopic force isn't as bad as it could be with a larger motor, especially a pancake style one, I suppose.
I suppose I *could* prevent the motor from ever actually running during a turn, but then I have to decide how far a turning of the fork is a turn, and how much just collision avoidance, where I might actually still *need* the power input. So I don't want to do that unless the problem is so bad I can't safely steer because of it.