Ok, so my team’s tinkering around with the idea of making a continuous variable transmission after looking at this website. I checked out what CD authorities have to say on the subject and I’m fairly convinced that making a CVT for competition is beyond our current capabilities. It may make a fun ongoing project for our engineers and maybe some students (like myself), but, well, most veteran teams toss the idea aside for more conventional, practical, and cheaper designs. So before I go to my team and tell them to kill their motors on this one, I’d like to have a better design to suggest for us to start working on for next season. But as I browse through the forums and the white pages, I quickly make the following realization: I understand transmissions about as well as I do about how I wound up being in charge of our robot’s drive train in 2003: ‘Ooh! Shiny!’
Why’s this guy rambling on so much?
So I see there are lots of people on this board who know a thing or two about gear ratios and torque. My small favor to ask is can the many transmissions posted on these boards be sorted by some very intelligent and, dare-I-say, genius individuals who can make it clear for those teams with little (or no) experience with these technical perplexities? Which would be the best candidates to build, test, and try to improve upon? What’s the simplest? The cheapest? The best? Myself, my team (if they put the CVT on hold), and many other rookie and rookie-esque teams out there would be incredibly appreciative of this kind of help. If nothing else it’d be nice to see a debate of the latest transmissions all in one place for easy reference.
Oh, and I guess I should ask how does that design of a CVT rate among others?
Aw man, it would be soooo nice if we could find the sharpie.
That’s another thing that is a variable of a design which, rated, would help out newer teams who want to dab into trannies: which ones are easy to make with limited resources (in our case we have a little access to a machine shop (iffy to get work done for non-competition components (budget limitations)), our own drill press and saws out the wazoo (band, jig, hack, you name it). Here’s a picture of our robot to give you an idea of what we’re working with. Worth a 1000 words indeed. Most followed by ‘?!’.
I am not an expert in gear boxes, but I know enough to say that one of the easiest ways to make a gear box is to have an input shaft with two gears a given distance apart. One should be large and the other should be small. There should be an output shaft with a small and big gear such that they fit together perfectly in any situation, except the big and small gear are spaced so that they can be switched by a pneumatic. In one example, say (for this year’s game), you wanted arms so that they could go fast to manipulate the 2x ball and slowly to hang and attach the bar to a stable part of the robot (the arm just lifted it up). Then you would have the big to small (say 1:1.5) used for the small ball and the small to big (say 3 or 4 :1) used for hanging. when the pneumatic is in, it is in low gear (1:1.5), but when the pneumatic is out, it is in the high gear (4:1).
Warning: By SIMPLICITY, I meant simplicity to BUILD. That does NOT mean that it will have a low weight (especially if you have only a hand tool/sharpie system). To reduce weight, welding is an option, and so is using lighter materials, such as aluminum (use it only on the big sprockets, which have force per socket). This is not a design, you will have to play around with it, but it is a basic idea for a transmission. If you are a beginner, I suggest that you don’t try an expert transmission, such as the excellent one built by team 222 (The Tigertrons), without some engineering expertise.
I would suggest the 2003 Single Speed if you are looking to add dual motor capability. This gearbox was also run this year by Team 228 with just the CIM motors. The Bosch’s can be run by themselves and the TechnoKats have several designs that can be adapted to shift with a servo.
716’s 2004 Dual Speed can be built in a week with a manual lathe and miller. I did not realize how many pieces there are until I did the paper!
Dave Lavery’s shifting transmission can also be built with manual machines.
You might want to check on Joe Johnson’s hints about the DeWalt gearboxes.
Bottom line, you can build a transmission with a drill press but it would much easier if you had a lathe and mill.
We’re going ahead with the CVT, despite my loud protests. It was my idea in the first place, but I was only kidding, and now our uber-fabricators are getting carried-away and… ahem
The current plan is to build the KillerBee’s Transmission (FIRSTwiki:Four Speed Transmission), as a fail-safe in the (IMHO) likely event that the CVT doesn’t pan out. The killerbees tranny is probably your best bet, and they will be posting a white paper “any time now” (though a gentle reminder may be in order). The easiest way to get a sweet tranny, but forfeit your rights to any kudos because you are a talentless hack, is to buy one. You can get a DeWalt drill and remove its gearbox system.
If you want to plow through with a CVT, I’ve written a qausi-informative article on FIRSTwiki that includes a rendering of our design. A CVT will be heavy beyond reason and you’ll also need a differential unless you want TWO CVTs. You are really opening a can-o-worms if you take on this project…
A CVT is a good idea if you want a broad spectrum of speed vs. torque. And there isn’t just one way to make a CVT. There are a few on the FIRSTWiki, there’s also one at ThunderChickens.Org in our whitepapers. That was our design in 2002, which worked very well (once we got over the current issues).
most veteran teams toss the idea aside for more conventional, practical, and cheaper designs.
It all depends on what you’re trying to do. The ThunderChickens’ CVT is very good for pushing matches, but if you don’t need to push very much, there’s no point in having it. If you have a game for speed, you want a speedy robot, not a tough robot, and vice versa. This year our robot didn’t even shift gears, because a torque/speed tradeoff wasn’t necessary. Whatever you do though, you need to make sure that it’s **RELIABLE ** above all. A fancy transmission won’t get you very far if it doesn’t work when you need it.
So I see there are lots of people on this board who know a thing or two about gear ratios and torque.
There’s also a tutorial on drive trains in our whitepapers. It’s by Paul Copioli(also on these forums), and it’s very good. The tutorial taught me motors and gears better than my Physics teacher ever did. Also you can PM Paul for help with specific issues.
Which would be the best candidates to build, test, and try to improve upon? What’s the simplest? The cheapest? The best
Though not extremely cheap, I fell in love with Chief Delphi’s Dewalt transmissions this year. Very good design, durable, and pretty :).
Ok here is my thing about a cvt. most people who know me know that I have been sitting on an original design for a while now and I tried to create it at one point and failed. But this was not all a loss in the process of making it I found a bunch of problems with both the manufacturing and the concept.
Here are some of the problems with a cvt
Weight. Since they have to be solid metal they are going to be very heavy, and if you decide to use to that is twice the weight
cvt’s are not efficient at transferring torque. because they are not a firm connection between the there will be a lot of slipping, which is fine for their original application (snowmobiles, and other gas engines)
the gear ratio’s are infinite but only to a point. lets say you have 2 identical cones 3 inches on 1 side and 1 inch on the other the max ratio you can have is 3:1 and 1:3 and if you want this smaller then 6 inches it is a serious taper angle (hard/impossible to machine without cnc, most lathes have up to a 15 degree taper angle) so now that you have this short really severe taper with 1/4 inch chain and the movements would be small but the ratios are not worth it.
In my opinion you would be better off making a 2 speed shift on the fly tranny because they are simpler to build with less variation, they also have a better torque conversion and can give you any desired gear ratio (From design).
as for other gear boxes which is best this is a tricky subject with out knowledge of next years game. 2 speed tyrannies are good over all but have issues like weight and a lot of machining.
single speed tyrannies are good but you need to figure out for sure what you want by analyzing the game and figuring out what is more important speed or power.
then the last most common choice is chain gears, personally I like this method because of its simplicity. and you can change the gears easy if you want to change the ratio during the comp (trust me I have done this many times) it also allows you to focus on more important things on your robot like functionality.
I hope this helps and if you decide you would like more information on anything feel free to pm me
edited to say: there are many types of cvts and in this post i am refering to a cone cvt a mentioned in the thread starting post. the thunder chickens used a planetary gear cvt, wpi used a friction taper cone, and there are many others.
for a reasonably fast, powerful, and simple® drive train, i’d go with a 4 motor, single speed gearbox. coming from a team that has done many different transmission types (2 motor single and two speed, 4 motor single and two speed), i can say that gear shifting is a no longer a necessity once you put in four motors. it is handy and cool, but certainly not a must do. plus, the added complexity of gear shifting can be a pain if and when it breaks.
with four motors, you can make a single speed gearbox that(when properly synchronized) goes 8fps and can turn out just as much power as you would ever need in a FIRST robot. just remember that the standard 4 wheel skid steering is enormously inefficient at turning (a fact that leads some teams into their more complicated two speed transmissions to get more power). go with something more efficient (coughsix wheelcough) and stresses on your drive train will be greatly reduced.
I’d say the simplest very effective (fairly fast and fairly powerful) would be what 696 did in 2003 and 498 did in 2004. http://www.team696.org/forum/image.php?iid=164 You gear down the Chia 3.5:1 to match the drill in high gear. Then you gear down the drill whatever (ours was 4.8:1) to the wheels (we used 8 inch). It all uses chains and sprockets and kit pillow blocks. All you need to machine is shafts and spacers. (I suppose if you have keyway shaft you could cut and grind that to length to make it even less machining) Since it has multiple pillow blocks and mounting points and stuff that lie on the same shaft (and since we used compressable nylon spacers underneath), binding and misallignment became an unavoidable problem. Also, aligning and tensioning the chains was always a problem too, as was high current draw. But, this was a very easy to build very fast, powerful, and RELIABLE setup. There was a lot of maintainance due to sloppy contruction on our part, but the only time it actually failed is when we threw a chain (to the wheel) by hitting a wall at full speed. another drawback was the weight - with steel sprockets and chains and shafts and 5/8 bearings and thick pillow blocks, it was no lightweight. However, I think if it was mounted rigidly in good alighment with properly sized parts (like 498 did in 2004 improving on our 2003 design), it would work very well. It is very easy to make, and it will definitely work - and work fairly well too.
If you are looking for a great little lightweight gearbox with a nice low output speed and are up for a bit of machining, I too would recommend the 716 2003 Single Speed. Both 696 and 980 as well as others used it with great success in the 2004 season. It is a bit more involved to make than the above (it was the most machining we’d ever done for anything) but is still quite simple. The gearbox itself was (and still is) flawless all season long. We made a few changes to the original design but nothing major. One thing we did have problems with however was the reliablity of the Bosch drill transmissions in low gear. I think we had to perform a swap maybe four times over a scrimmage and two regionals.
If you have any questions on any of the above, let me know.
suneel112 is right about our transmission in that it couldn’t be built without a lathe and a mill. Our transmission is simple in the idea but takes time to fabricate…just to give an idea of how much we worked on building it (our first ever transmission), it took us 4 weeks outta the build season for just designing and building it. Will it ever take us that long to build again, no.
I guess what I’m trying to say is think of concepts, how everyday things work, there are some pretty simple ways to make a shifting transmission. I guarantee you that whatever you decide to build is gonna take you a while the first time you build it, but you’re going to learn lots, you’re going to find out what works and what doesn’t. The biggest thing is you don’t have to make your robot like another teams. Come up with something unique and go from there. RAISE THE BAR…start brainstorming/prototyping, you have plenty of time! Good Luck on your transmission!
All I can say, whatever you do, make sure you build a prototype. I saw a team cut apart half of their robot to pull their transmissions at a regional this year because it wouldn’t drive. I’d hate to see it happen again.
I don’t know if you saw it or not this year, but we built a very simple 4 speed, 4 motor transmission. All you really need is a lathe and a drill press with an x/y table (or just a mill). The 2 motor aspect is done by linking the drill in low and chip motors with chain and sprockets (3-1 reduction I think). It’s incredibly easy to build, we had a sophomore do most of it in two meetings. It’s cheap, too, and you can make a 2 speed by simply cutting it in half…or you could save yourself the trouble and only build half of it!
Out lead engineer, Jim Zondag just finished drawing it in inventor and “I’m still working on exploded assembly veiws, parts lists, detail drawings, etc.” It should be in the whitepaper section very soon.
He did, however, send me a picture of the drawing. It’s pretty
As my friend cliff has said, just look at every day things, watch how things work. You may see new and inovative to do things. As far as our transmission is concerned, we had a machinist who used ride motorcycles, and he was like hey guys check this out and we went from there, prototyping and building. I guess what im trying to say is, anything is possible, even if it doesnt work dont get discouraged, chalk it up to a learning experience and try it again. As long as the students learn something from the situation then it was wellworth the time.
Now understand that I mean no disrespect to the ThunderChickens at all, or to their design team, but the 2002 ThunderChickens’ CCT is not a CVT. IIRC, it is a clever way of combining two motors that run at different speeds that gives high torque when you need it and high speed when you need it. Toyota tried a similar trick on the original Prius, but was forced to change the name from “Planetary CVT” to “Hybrid Synergy Drive” when people discovered that it wasn’t a CVT (there is in fact no such thing as a “planetary CVT”).
The CCT, with all due respect to the ThunderChickens, was flawed in design because it required the use of a worm gear to prevent the high torque motor from backdriving the high speed motor. Since a worm gear only has about 70% efficiency (compared to 97% for a spur gear or sprocket), much of the extra torque gained was immediatly lost to heat and nosie.
Coming from a team that has build a CVT, I can tell you that it is very difficult. Although the 190 CVT worked perfectly mechanically (and never had any problems with slipping), programming it proved to be very difficult. Throughout the 2002 season, our CVT was run as a manual three speed transmission, and it wasn’t until well into the 2003 season that we got it working. The CVT itself is quite difficult to dynamically model, especially since the shifting behavior is non-linear, and figuring out how to drive the thing with the right amount of dampening to prevent it from oscillating was a huge task.
If you want a CVT, more power to you. It is great to see teams innovate and try to improve on the past. However, I would definately make sure that you have a backup transmission, because your CVT might not perform as you want it to until the 2006 season.
With all due respect to 190… :rolleyes:
You speak of efficiencies, and mention your CVT “worked perfectly mechanically”.
What efficiency did you achieve?
Have you ever dyno’d up that robot?
I was under the impression (correct me if I’m mistaken), that your CVT was of a toroidal style, which relied on friction (between 2 cast iron pieces) to transmit torque. How is this any more efficient?
You speak of energy loss to noise… well… I’ve heard the 190 CVT run.
To say it purrs like a kitten… would be a lie.
I fail to understand how the inefficiencies of 217’s worm gear cause it to be “flawed in design”.
As far as I know, by nationals in 2002, the CCT was a force to be reckoned with.
Though maybe Paul could more accurately answer your criticisms.
525 uses a very simple 4 motor single speed drive train. It uses the out of the box drill motor with the 2:1 right angle gearbox driving the 9" wheels at one end and a CIM with a double reduction spur gear set driving the 9" wheels at the other end. The number of teeth on the double reduction spur gear set are selected so the front & back wheels turn at the same no load speed. Then to make it easy to drive, there is a #35 chain connecting the front and back axles to force the front and back to turn the same speed.
The only precison machining needed is the gear centers for the CIM gear box.
We ran this drive train this year with zero problems.
Last year we used the same concept with the FP’s instead of the CIM’s. The only problem during 1 regional & nationals was one pair of burned out FP’s.
Picture attached - note the chain connecting front and back is not visible.
Of course I was not speaking of efficiencies, I was talking about function (and I know the CCT worked perfectly in the same sense once they got their current issues worked out). I don’t have the figures off hand for the CVT, but I know it was calculated by the group who designed it for their Major Qualifying Project.