h a l f b a k e r y
A hive of inactivity
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
The in-hub gears used in the bicycles of the discerning
an astonishingly rare example of an American invention
being perfected and made commercially successful by the
canny Brits <link>. They are clever devices which provide
geared output from an input with reference to a fixed
In a bicycle, the sprocket is the input, the frame
reference (usually via an antirotation axle) and the hub
shell the output. The input:output ratio may be anything
you like, several selectable ratios may also be crammed
into such a hub*.
I recently improved the design of the wheel <link>
by making it significantly more compact. I don't blame
George Cayley for his shoddy design work, it was simply
"needs must" solution to a specific problem, like
The idea was for a whole wheel to be replaced by two
quarter-wheels. A bicycle rolls forward from one
wheel to the other, as the weight is taken off the first
quarter-wheel it races 'round to be in front of the second
quarter wheel, then the whole cycle repeats. To my
eternal shame I proposed a solution involving sensors,
electric motors and all sorts of unnecessary gubbins.
sensibly, let's use hub gears.
The reference for one quarter-wheel is the left fork leg,
the reference for the other quarter-wheel is the right
leg. The input is the hub shell of the first quarter-wheel,
this drives the output at a 3:1 ratio. The output is the
second quarter-wheel. As you push the bicycle along,
torque applied to the first quarter-wheel drives the
all the way over the top, moving 270 degrees compared
the 90 degrees rotation of the first. Then the second
quarter-wheel drives the first over the top in the same
manner. Since the drive from one quarter-wheel to the
other is only ever unidirectional, the two can be
by simple freewheel arrangement. I suspect a bit of a
drive arrangement might be necessary to absorb the
opposing torque at the segment to segment transfer.
I haven't however, worked out how you apply drive to
wheel, so it might have to just be a scooter. Or possibly
*Although bicycles with more than 3 ratios have been
linked to indecision, unhappiness, hysteria and
ungentlemanly conduct. Such problems are at epidemic
proportions in continental Europe.
[bs0u0155, Sep 13 2016]
Needlessly complex predecessor.
[bs0u0155, Sep 13 2016]
||For a smoother ride, you could build a mechanically actuated pair of quarter wheels using elliptical or non-round gears. You might have to have two drive gears on the same shaft but out of phase, and the driven gears then connected to the wheel segments. It should be fairly simple to devise a gear profile with the desired variation in speed.
||I have been pondering this sort of thing, in light of
[bs]'s wise observations.
||The logical limit, of course, is not to have any
wheel-segments at all, but simply two pointy-
ended spokes which "walk" in tiny steps in a rapid
blur, creating the same effect as a very-many-
sided polygonal wheel (ie, a circle).
||But having the two spokes swing back and forth
(like legs) involves lots of reciprocating mass; and
having the rearmost leg swing back-and-up-and-
over each time involves lots of unnecessary
movement, as well as large ac- and de-
||Better, surely, to have the two spokes welded
together to form a very narrow inverted V (both
points on the ground). Then tilt the V very slightly
sideways (say, towards the midline of the vehicle).
Now, if you just spin the entire inverted V along its
long axis (that is, along the midline between the
two spokes), you will have achieved the required
result, with no reciprocating masses and no large
changes in velocity.
||This, of course, is exactly the way that one marks
out multiples of a given distance on a chart, using
a pair of compasses: the compasses remain at a
fixed spacing, and simply tilt-and-rotate to make
successive steps. A skilled user can "walk" thirty of
these steps faster than you can count.
||//Then tilt the V very slightly sideways (say, towards the
midline of the vehicle).//
||After brief experimentation with a pair of straight
forceps I happen to have lying around there's a problem.
My forceps are about 11 cm total length with about 1 cm
of separation at the tip. In order to gain 5 mm of ground
clearance for the non-active tip, I have to deflect the
top 6 cm. That's quite a lot. If you scale it up to vehicle
sizes you are faced with either a very wide V, or a very
very tall V just to clear a medium pebble.
||You lost me at // More sensibly//
||//In order to gain 5 mm of ground clearance for
the non-active tip, I have to deflect the top 6
||Ah, but why aim for 5mm of ground clearance?
Scale that up to a man-sized vehicle and it would
become several centimetres - far more than the
micro-gap which is fashionable for today's high-
||A greater problem, I suspect, is that the compasses
will have to be spun at an astonishing rate to
achieve significant forward speed.
Howevertheless, this simply means that we have
found the ideal automotive application for the jet
turbine, without the need for immense gearing