Methods of shifting gears

[lukedude43]

Great thanks i'm like 90% sure I understand now

[Tristan Lall]

The one you see most often nowadays is dog shifting: the dog gear is (with the teeth on the side face) engages a recess on one of two output gears. I've built one of these with 188; it worked great. I've also bought these off the shelf.

Captive-ball shifting is once again popular, with the release of VexPro's kit. Previously, 222 used this design. Here, balls are displaced in and out of recesses in the central shaft, to engage gears with slots in their bores.

You used to see axial-engagement ("crash") shifting more often. There, the gear teeth themselves slide in and out of mesh. I'm not a huge fan of this, but implemented with fairly large gear teeth, it works fine.

Rarely, you see tangential-engagement shifting. I tried it once with 188 and it didn't work properly (the gears wouldn't stay in mesh). However, knowing what I know now, I think I could probably make it work. However, I don't see any particular advantage to doing so.

You also used to see planetary shifters more often, when the Nothing But DeWalts modification was popular. (A DeWalt XRP drill transmission, shifted with a servo by sliding the device that locks and unlocks its planetary gearsets.) I've implemented this, again with 188, with two CIMs feeding a single DeWalt. Everything worked reasonably well, except the custom output shaft. (It needed to be hardened alloy steel, not hot-rolled free-machining steel.)

[Michael Hill]

I'd be interested if it were possible to use a linear electric solenoid to shift instead of a servo or pneumatics. The bad part is that they have maximum "on" times (duration the plunger can be "out"), which can be between 1 and 2 minutes or shorter (which is a problem in our application) because they need to cool down. It would be useful for teams who don't want to deal with pneumatics, especially if shifting would be their only use, and want to have fast shifting, because servos are sloooooow (in comparison)

[Dad1279]

Originally Posted by Michael Hill

I'd be interested if it were possible to use a linear electric solenoid to shift instead of a servo or pneumatics. The bad part is that they have maximum "on" times (duration the plunger can be "out"), which can be between 1 and 2 minutes or shorter (which is a problem in our application) because they need to cool down. It would be useful for teams who don't want to deal with pneumatics, especially if shifting would be their only use, and want to have fast shifting, because servos are sloooooow (in comparison)

Could use a window motor/cam linkages to shift both, or a faster motor with a leadscrew.

Dealing with pneumatics isn't that difficult. Don't need an on-board compressor.

[fb39ca4]

Has anyone tried using a shifting bicycle hub? The kind that bikes without derailleurs use?

[Daniel_LaFleur]

Originally Posted by fb39ca4

Has anyone tried using a shifting bicycle hub? The kind that bikes without derailleurs use?

The biggest issue with derailleurs is that it is a single direction gearbox, and most FIRST applications are bidirectional.

[apalrd]

Originally Posted by fb39ca4

The kind that bikes without derailleurs use?

He means something like this


As to the OP's question, I can think of 3 successful team-built shifting gearbox types (this excludes the planetaries in the Dewalts and drill transmissions):

-Mesh shifting ("Crash" shifting) - Sliding gears to mesh. The Killer Bees 4-speed design linked earlier is a mesh shifter, as were many early FRC shifting transmissions. They are relatively to easy to build, and with a large pitch gear and slightly more backlash they can work fairly well. While not space efficient it is possible to have a 3 or more speed mesh shifter with a single mesh. The 4-speed is essentially 2 2-speeds in series as this design is more space efficient. As the paper describes, we really wanted a 3-speed but the 4-speed was easier to build. Note that 2 2-speeds in series requires a double swap shift for 2<->3 shifts and that is not usually a 'good' shift to perform (if it's not perfectly synchronous you'll end up in 1st or 4th for a short period of time).

-Dog shifting - Sliding a dog to mesh. In this case, the two gears in question ride on the common shaft via bearings, and the mechanical coupling to that shaft is provided via the dog. The dog meshes with one gear at a time to pick a gear ratio. The dog design is limited to 2 gear ratios per dog. You can get 4 speeds by designing 2 2-speeds in series or having multiple dogs on a shaft (this would usually, but not always, require the ability to shift a dog into neutral)

-Ball shifting - Balls sit in cups of the shaft and are pushed out to engage the gears riding on the shaft to engage a gear. This design is capable of any number of gears in sequence adding only another pair of gears and balls and slightly lengthening the shaft, and can shift sequentially very easily. It is impossible to shift non-sequentially (each gear you step through must engage to let the plunger pass by it). For a 2-speed design like the Vex Pro this is irrelevant.

There is also the theoretical option of the planetary transmission. An automotive planetary uses planetaries and lockup/drive clutches to select a power path through the planetaries (which elements are connected to the input and which are grounded). Good designs allow jump-shifting nearby gears with a single clutch-to-clutch shift (exchanging one clutch for another). A drive or lockup clutch in an FRC application could be a part of a dog, mesh, or ball shifter, or a lockup band.

[henryBsick]

I've designed and built a few ball shifting transmission in the past. I really dig the new Vex Pro models that are out. They've got many of the 'wanted' features I've recognized over the years, but never implemented. Kudos to them for a solid design.
Linked are two pictures.
1) Project cover page listing team members and showing completed drive base.
2) Shifting detail on the transmissions. The large exploded view is a bit crap (my apologies), but the other simple illustrations outline function showing the shifting shaft, plunger shaft, shift gears, and ball bearings.
I'm happy to share any specific details if anyone is interested in developing their own ball shifter.

[A_Reed]

CVT anyone?

http://en.wikipedia.org/wiki/Continu...e_transmission

Not entirely practical for FRC applications but it certainly would make an interesting design project to see if you can make it work.

[joelg236]

Originally Posted by A_Reed

CVT anyone?

http://en.wikipedia.org/wiki/Continu...e_transmission

Not entirely practical for FRC applications but it certainly would make an interesting design project to see if you can make it work.

Imagine having one of these on your robot. http://upload.wikimedia.org/wikipedi...xtroid_CVT.jpg

[Tristan Lall]

Originally Posted by joelg236

Imagine having one of these on your robot. http://upload.wikimedia.org/wikipedi...xtroid_CVT.jpg

190 did a toroidal CVT with power take-off in 2002. (Search the forum for more information, e.g. this thread.)

[Joe Ross]

The Thunderchickens did a planetary CVT in 2002 and published a whitepaper: http://www.chiefdelphi.com/media/papers/1361

[lukedude43]

thanks to everyone for all the information it's been incredibly helpful. Something that I didn't realize I left out is that this is a project to help me learn solidworks and because off that i'm gonna go with dog shifting just because of the easier design. Also if by some chance we actually made this instead of just buying some AM or VEX shifters, dog shifting would be easier for my team to do because they would be our first machined parts and it seems much easier to design than a ball shifter.

[Gregor]

Originally Posted by lukedude43

thanks to everyone for all the information it's been incredibly helpful. Something that I didn't realize I left out is that this is a project to help me learn solidworks and because off that i'm gonna go with dog shifting just because of the easier design. Also if by some chance we actually made this instead of just buying some AM or VEX shifters, dog shifting would be easier for my team to do because they would be our first machined parts and it seems much easier to design than a ball shifter.

Unless you have a really, really good reason, stick to COTS gearboxes.

[mman1506]

Originally Posted by lukedude43

thanks to everyone for all the information it's been incredibly helpful. Something that I didn't realize I left out is that this is a project to help me learn solidworks and because off that i'm gonna go with dog shifting just because of the easier design. Also if by some chance we actually made this instead of just buying some AM or VEX shifters, dog shifting would be easier for my team to do because they would be our first machined parts and it seems much easier to design than a ball shifter.

Make sure you do it in the off season. Before COTS 2 speed gearboxes were available are team wasted a lot of time making 2 speeds that were mediocre. I'm not sure how well the AM shifters shift but our dog shifters take quite a bit of tuning to shift in a useful fashion.