I’d call the huni rex an upside-down penguin giraffe with a duplex chain. Far as I’m concerned, it will never be fast enough to justify the drivetrain. Lowering the pedals limits handling too much.
I’m getting a jackshaft build together. Probably more news in a few months. Curious about the hubs in each, all custom work right?
Great presentation, I really enjoyed it! I particularly liked the comment about chasing the bicyclists up hills. Great stuff.
Those custom-built jackshaft unis are great!
(A guy posted a picture a while back in the “Club Photos” thread of a person who had built a 29" wheel into a Huni-rex. Seemed like a pretty nice setup.)
The custom hub can actually be one of the easiest parts to make.
A square-udc-wide-hub can be peeled down to fit nice inside a mtb-hollow-front hub (20mm axel hole). Actually, these mtb front hubs are 100mm wide, so it could hardly match better! Put the original bearings for the udc hub back on, and everything is hold in place. A cog is then mounted to the disk-holes.
Thanks for those pics, I didn’t really understand how it worked till I saw them.
Is there any sort of bearing between bike hub and axle?
I’mworking on a slightly different approach, using the drift trike hub with the freewheel disabled if that proves doable. Same deal with disc mount cog however
It seems like there would be, otherwise there’s no way it would work. Part of the hub rotates around another piece on a bike so I imagine what this is doing is replacing that piece with a uni hub.
Those “tube” hubs are all pretty much built like this. (The shimano hub do not have the apropriate dimension, but it illustrates the idea with the “tube”, and the bearings.)
I found my way into this thread by way of your blog post, that I found by way of your blog entry about floating point determinism. Which in turn was recommended to me on another forum, where I complained about up to two bits of difference in double output depending on whether I compile my simulation using clang or gcc.
Which annoys me, because it makes it more difficult to validate the correctness of said output, i.e. I have to parse it and use fuzzy comparison instead of just hashing the output.
Any way, finally dusting off this account feels really good.
[QUOTE=kamikaze;1634644]
I found my way into this thread by way of your blog post, that I found by way of your blog entry about floating point determinism. Which in turn was recommended to me on another forum, where I complained about up to two bits of difference in double output depending on whether I compile my simulation using clang or gcc.
…QUOTE]
This may not be the best forum to discuss floating-point issues, but I enjoyed the tangled tale of how you got here.