Internally geared hubs as mid-drive transmissions?

I'm pretty sure we've had a number of discussions about using internal gear hubs on mid-drive setups, and the general consensus seemed to have been that using an IGH laced into the rear wheel on a vehicle running a large RC motor (something like the Astro Flight 3220 or the Turnigy C80100) would be a terrible idea because it would involve putting far more torque through the hub than it was designed to put up with. Running some numbers through Wolfram Alpha, a 130KV C80100 produces 11N*m of torque at the shaft if you feed it 150A. Assuming you run it at 72V, you're going to need to gear it down by something like 15:1 to get a usable top speed (in this case about 51.5mph), which produces 165 N*m of torque going into the hub. If we're taking the Shimano Nexus 8 as an example, then it's taking 165N*m in, and in its lowest gear (0.527) is putting 313N*m back out through the wheel; that's somewhere on the order of two to three times more than what you want to put through it and (given the significant expense associated with a nice IGH) you don't really want your $200+ hub to be a six-month wear item.

That said, (in theory at least) it should be safe to use a nice internal gear hub in a high-powered RC mid-drive vehicle as long as you use it before the main gear reduction, as long as the high RPMs aren't a huge concern. If we take the same system that I mentioned before, a 130KV melon at 72V and 150A will only put 11 N*m through the hub input, and in the lowest gear you'd only be putting out 20-21 N*m; you'd then be free to use much sturdier chain and sprockets on the post-reduction portion of the drivetrain, making it possible to get a wide range of speed/torque combinations in a >5lb package for under $250. The only possible catch I could see with this would be wear and tear on the bearings and gears at stupidly high RPMs; in a human-powered setup most IGHs wouldn't be seeing more than 1-200 RPM in and more than 4-500RPM out. A 130KV C80100 on a 72V nominal system will be spinning at anywhere between 8K and 12K mechanical RPM; increasing the number of machine cycles per mile of use by a factor of 10-20 might lead to significantly accelerated gear and bearing wear. I'm not sure whether or not this is significant enough to be concerned about, and I was wondering if people had any experience using IGHs in this manner.

I've run a nexus 3 speed as the final drive (in the wheel) of a prototype bike with mild power (1500 watts) with no problems, but we didn't out many miles on that bike.

We're now using the same nexus 3 as a Jackshaft middrive for an electric go kart, about 2000w peak input at 4500 rpm from the motor. Our run time won't be as much as a commuter might have, but our abuse level is very very high. I'll let you know how it goes.

One of the MIT RD at MIT (you can probably find his blog through etothepiplusone.net) used a Shimano alfine 8 as the Jackshaft for a reduction from a 4 or 5 kw peak Mellon motor. He has a one year report on it that reports no problems.

ttobiassen said:

I've run a nexus 3 speed as the final drive (in the wheel) of a prototype bike with mild power (1500 watts) with no problems, but we didn't out many miles on that bike.

We're now using the same nexus 3 as a Jackshaft middrive for an electric go kart, about 2000w peak input at 4500 rpm from the motor. Our run time won't be as much as a commuter might have, but our abuse level is very very high. I'll let you know how it goes.

One of the MIT RD at MIT (you can probably find his blog through etothepiplusone.net) used a Shimano alfine 8 as the Jackshaft for a reduction from a 4 or 5 kw peak Mellon motor. He has a one year report on it that reports no problems.

Click to expand...

Yep, that's Ben Katz, and that's the vehicle I was thinking of; a Radio Flyer-sized tricycle capable of 47mph on the flats. I was actually looking at setting up something like that with a Revolt RV-120 (basically an upsized melon from Israel with power ratings that are actually believable) at 15kW peak, 5-6kW continuous. Given that the RV120 is 45KV, a 200A peak current would put peak torque through the hub at 40-50 N*m, and in first gear the hub would put out 79-95 N*m of torque (which is still well below the sensible maximum torque limit for a quality IGH like the Alfine); similarly, an RV120 running on a 72V system should be pulling no more than 4000 RPM under load, which is maybe 80% or so of the maximum RPM of Ben Katz's rig. As far as I'm aware, at that point you can slam 15kW through a hub at that point without doing permanent damage.

Update: the drive train in the go kart has been switched to Transmag 532 brushless power, with a commensurate increase in power and torque. New momentary max is about 3.5kw. The New York Makerfaire is this weekend, where we're going to race the kart in the power racing series for a 75 minute endurance race, shifting irresponsibly the whole time

I'll check back in afterwards to report if we have any problems.

Please can we use torque, rather than power, when attempting to quantify the capacity of hub gears...... Preferably, torque and rpm.

toolman2 said:

...also i tried a red band 8speed nexus as the jackshaft with a few kw and it stripped gears (from rpm?) similar to others have found.

Click to expand...

http://endless-sphere.com/forums/viewtopic.php?f=28&t=52151&start=50#p830123

bobc said:

More good feedback re: the alfine IHG - we have used these for the last 3 years in the greenpower electric cars (winning 4 national championships in that time ;^) These cars put up to a kilowatt into the gearbox, average about half that. The alfines have never missed a beat & seem to be as good now as 3 years ago - better in fact - the efficiency has improved as they've run in. That's over 2000 miles of racing + lots of testing and training on each of the 2 cars.
Previously we ran cars with shimano nexus 7 speed IHGs - these DID break after a few thousand miles and gears 6 and 7 ceased to exist when you spun them over about 800rpm (centrifugal forces affecting internal ratchet pawls I think)

Click to expand...

http://endless-sphere.com/forums/viewtopic.php?f=28&t=26567&p=s#p384451

I was running 2x cyclone 1200w motors through a nuvinci hub 360n ,took about 3 months to start slipping ,well it was more of a judder like the chain was jumping ,but it was the internal bearings losing traction in the viscose oil,cycle analyst said 3200w
Gary

Quick question; were you running the NuVinci before or after the final gear reduction?

Update on Nexus 3spd in power racing series electric go kart:

Our max input speed to the Nexus with our previous motor (Ampflow A28-400, brushed, about 2kw peak) was about 1400 rpm, which was a 3.3:1 reduction from the motor output. The spoke flange was adapted to a sprocket driving the differential, at an additional reduction of 1.8:1, or a final drive ratio of perhaps 6:1. This was run for about 3 hours total runtime between testing and racing, primarily only using 1st and 2nd gears, shifting up and down every few seconds in between sharp curves on a very small track. Shifting was done under power from 1st to 2nd and power was removed when shifting from 2nd back to 1st. Total estimated shifting cycles during this period of racing/testing was approximately 1000 cycles.

Our max input speed to the Nexus with our current motor (Transmag 532, brushless, about 3.5kw peak) was about 1000 rpm, which was a 2.7:1 reduction from the motor output to the Nexus. Output from the spoke flange to the differential was again 1.8:1, for a final drive ratio of about 5:1. This has been run for about 3 hours of runtime as well, with shifting being performed similarly but this time utilizing all 3 of the gears. Cycles of shifting are also approximately 1000.

If you equate this to the amount of shifting you might do on, say, MY ebike commute every day: My commute is about 2.5 miles, with a stoplight, stopsign, or other FULL stop every .25 miles or so. There is also a hill on the way back which requires downshifting. This equates to perhaps 25-30 shifts all the way down and all the way up through the gears every day, and would probably be done more responsibly than in a 'race environment'. To get to 2000 cycles would therefore be about 3 months of regular weekday commuting, which is encouraging because we have NO signs of the transmission giving us any problems to date.

That said, our input rpm is relatively high (~1000rpm), our output wheel is small (9 inches or so), and the Transmag was current limited at startup to about 100amps, so torque was limited. Using a hub like this (or perhaps a Nexus 8?) seems to be ok as long as it is used as a jackshaft in a 2 stage reduction.