OT: Continuously Variable Gear Transmission


Thought some of you here might enjoy this story.


I think it's genius.


Looks neat.




It sounds like it's not really a valid CVT if you have to have a motor of variable speed and torque on the second input shaft.

I read an article from decades ago (1950's, IIRC) on what people expected for the future of bicycles. One of the things was that they would have CVTs. What the writers did not anticipate was that we would get close to that with much lower weight, better efficiency, better durability, etc. by having so many gears so closely spaced. (They're out to 11 cogs in the back, and you can see how close the spacing is if the number of teeth goes 12-13-14-15-16-17-18-19-20-21-23. There are two to three chainrings in the front to select the range. Shifting is practically instantaneous too, unlike 30 years ago and back.

Edited: 14 May 2010, 3:16 p.m.


Actually we had closer ranges in the 60s than today. (Nobody shifts front and rear simultaneously for a specific fine tuning of ratio, except in very rare circumstances--too complicated and slow). This is because the old freewheels had a smallest cog of 14t and now we have 11 tooth cogs. The change in ratio from a 44-11 to a 44-12 is much bigger than the change from a 56-14 to a 56-15.

But the amazing thing is that some people can actually use a 53-11. Nobody road raced with a 67-14!

Edited: 14 May 2010, 3:37 p.m.


You're referring to half-stepping (optionally plus granny), which I did use for years. The most even spacing comes from a 42-47 front and 13-16-20-25-31 rear, with each step being almost 12% above the last. I ran calculator programs on gobs of combinations looking for the perfect distribution of gears in those years. Actually the calculator programs came after having done the same thing with a slide rule and plotting the results on logarithmic number lines (which are easily made by putting the slide rule's slide down on the paper if you didn't have log paper). All of the step sizes in the 12-23 11-speed cassette are smaller though, and without shifting chainrings. In the 70's, we could choose exact numbers of teeth. Today, shifting, especially of chainrings, has been made quicker and more foolproof by, among other things, the addition of ramps and pins on the larger ring(s) which not only catch the chain to help it up, but are also located such that when the chain comes up, the rollers are already positioned perfectly to go between the teeth instead of onto the points (or other in-between orientations). However this requires that the rings be paired for each other, something you can't do if someone can order any random combination of sizes. I partly lament the fact that the old way is mostly gone; but in most ways the new does prove to be better.

Edited: 14 May 2010, 4:23 p.m.


Wow, I haven't talked about this in years! Half-step versus "alpine" and some other gearing ideas. I also made extensive calculations as a kid--by hand (except for my trusty hp) with "sawtooth" diagrams to show the shift sequence.

But in racing, all those ideas went out the window by the time I was racing in the 80s--and it was simply "straight block" . There was one race, in Hagerstown, in the rain, as a junior, when I rand a 15-17-19-21-24 but every other race I have ever done has been in a straight block at least for the 4 smallest cogs.

It is interesting to see old racing bikes from the 60s with close spaced front chainrings--49/52 even! And in the back, a 14-16-18-21-24.

By the way, I have a Huret jubilee, two Suntour Superbe pro, a SunTour Cyclone, and both a Nuovo and a Super Record rear derailleurs. The newest campagnolo carbon units are still heavier than the old cyclone and jubilee--though they sure do shift better.

Edited: 14 May 2010, 7:32 p.m.


I still have a Regina 14-15-16-17-18 freewheel in a box. Your 15-17-19-21-24 is part of an almost modern 12-27: 12-13-14-15-17-19-21-24-27. The 10-speed version adds the 16.

I still have all my old papers from having calculated lots of setups by hand.


The newest campagnolo carbon units are still heavier than the old cyclone and jubilee--though they sure do shift better.

Speaking of better shifting, the latest SRAM 10 speed cassettes have both asymmetrical and missing teeth in specific locations to aid in moving the chain off the gear faster. As first glance, they look like they've seen 20K miles and should be in the scrap bin, not on the wheel.


I think the CVT story started in the Netherlands by DAF cars.

The variomatic transmission in these cars had a gear for every speed.

See also: Link to DAF Museum early CVT construction.

The mechanism was based on a 'belt-pull' torque powertrain.
Evolving development resulted in a 'steel-belt-push' torque powertrain called CVT.
DAF cars reduced its portfolio to trucks but the transmission technology survived in VDT, later integrated in Bosch.

See Bosch CVT for the details

Maybe we should support Steve Durnin with some perfect HPxx programs for simulation and calculation .....

Edited: 14 May 2010, 6:10 p.m.


I can speak from experience - having had a 1999 Honda Civic with their version of the CVT. It's a little bit of a strange experience since you don't get the sensation of the gear changes as you accelerate. I think Honda used a steel belt CVT. I put over 230,000 miles on it and only got rid of it when the air conditioning failed & I didn't feel like spending a lot of money to fix it.

A lot of people had problems with the early demise of the Honda CVT transmission. Mine went at 98,500 miles but I had an extended warranty that put a new one in for only $25 out of pocket.

Turns out that you needed to use only Honda CVT transmission fluid and - this is the key item - change the fluid often. After the first transmission, I started changing the transmission fuild with every other oil change - and no more problems.

It was a great car and got 35-60 MPG in combination city/open road.

Just to make this HP related, thank goodness I used the HP calculsator to calculate the extended warranty plan.



I think CVT predate this by quite a bit. The earliest version I know about was by George Constantinescu who was quite an amazing inventor. See here for a discussion and description of his torque converter.

Mechanical CVT are available commercially e.g. zero-max.

- Pauli


Actually, there do exist CVT-like things for bikes:
hettlage. I haven't seen a single one on the street, but you can buy them.


Interesting. I hadn't seen that one before. It looks like it's not continuously variable though. The positions are indexed, apparently so the teeth don't hit on the wrong parts of the belt. As for their comment on the belt lasting five times as long as a chain, I use an unconventional dry lube method with paraffin and graphite that makes the chain last at least 20,000 miles of hard use (a lot of climbing).

Possibly Related Threads…
Thread Author Replies Views Last Post
  hp-prime solver and variable name fabrice48 22 8,338 12-10-2013, 03:25 AM
Last Post: fabrice48
  Prime: Scope of Variable and functions within programs Thomas Chrapkiewicz 9 3,342 11-18-2013, 03:39 PM
Last Post: Thomas Chrapkiewicz
  HP Prime - Variable problem bluesun08 12 3,549 11-07-2013, 11:34 AM
Last Post: bluesun08
  50G variable issue Matt Agajanian 8 2,455 05-21-2012, 03:14 PM
Last Post: Matt Agajanian
  Reminders/notes to help recall variable meanings in equations Chris C 11 3,017 02-16-2012, 06:53 PM
Last Post: Bart (UK)
  Idler gear used in HP-97 printer Gary Robert Bosworth 4 1,651 11-19-2011, 03:39 PM
Last Post: Randy
  Card reader gear lubrication - do or don't ? John Robinson 2 1,076 05-31-2011, 09:43 AM
Last Post: Randy
  HP-97 Printer Gear Replacement XYZT 3 1,412 12-23-2010, 03:30 PM
Last Post: XYZT
  HP-97 Aluminum Printer Gear aj04062 4 1,526 09-16-2010, 07:31 PM
Last Post: aj04062
  Optimizing programs: Writing a number VS calling a variable Pablo P (Spain) 4 1,471 07-12-2010, 11:25 AM
Last Post: Katie Wasserman

Forum Jump: