24-08-2012 15:52

steelerborn

Re: pic: 2012 FRC971 Transmission Actual

Really like this setup! You guys had a great bot this year!

Do you have an overall weight for the transmission? It looks very light weight, plus with AL gears.

24-08-2012 18:42

AllenGregoryIV

Re: pic: 2012 FRC971 Transmission Actual

This is a very nice looking gearbox.

How do you mount the two gears to the output of the CIM that allow you to do single reduction?

What is the part number/soruce for the pancake cylinder they seem really useful in this application.

24-08-2012 19:46

James Kuszmaul

Re: pic: 2012 FRC971 Transmission Actual

Not entirely sure what you are asking here, but the way the gears are mounted to the CIM shaft is by pressing a hex sleeve over the CIM shaft, then using the hex to mount the gears. If I remember correctly, then the smaller gear is 15 tooth, the larger is 30, and then the two gears in the second stage are 60 tooth and 45 tooth, respectively.
Bimba FO-020.3-3FMT

25-08-2012 01:05

AllenGregoryIV

Re: pic: 2012 FRC971 Transmission Actual

That's exactly what I was asking, thank you for the quick response.

25-08-2012 02:21

DampRobot

Re: pic: 2012 FRC971 Transmission Actual

If I were to design up a transmission like this, I would get a lot of flack from the programming leadership about not including space for an encoder. Did you end up including one that we can't see, or was there an argument for not including one?

25-08-2012 03:17

MichaelBick

Re: pic: 2012 FRC971 Transmission Actual

Most teams who do WCD end up just connecting an encoder to another wheel(any of the front or back wheels) and then zip tying the encoder wire down to the bellypan. I can speak for 971 but I assume they did this too.

25-08-2012 12:13

Chris is me

Re: pic: 2012 FRC971 Transmission Actual

971's drive wasn't a WCD, and since they used dead axles, I doubt this was the method they chose.

25-08-2012 15:13

Nick Lawrence

Re: pic: 2012 FRC971 Transmission Actual

This is actually beautiful.

Man, you guys make some of the nicest robots around.

-Nick

25-08-2012 15:23

kevincrispie

Re: pic: 2012 FRC971 Transmission Actual

The encoders aren't part of the transmission. They are off on the back wheels on a plastic gear I believe.

25-08-2012 15:50

DampRobot

Re: pic: 2012 FRC971 Transmission Actual

And just to be clear, did you ever end up using them? In my experience, they are often a pain to design in, and not that useful for the programming team.

25-08-2012 16:01

Cory

Re: pic: 2012 FRC971 Transmission Actual

Why do you find them to be not very useful?

25-08-2012 16:12

Garret

Re: pic: 2012 FRC971 Transmission Actual

If you just use the AM encoder bracket for the E4P ts really straight forward to design them in; from memory I think its something like .75" out to each side and .1875" down for the mounting holes with respect to the shaft.

25-08-2012 18:05

R.C.

Re: pic: 2012 FRC971 Transmission Actual

Pretty easy to design in. Just drill a 1/4 hole and shove a s4 in. Literally no work at all.

-RC

25-08-2012 19:51

DampRobot

Re: pic: 2012 FRC971 Transmission Actual

The theory always seems to be that they will get used for autonomous tracking or smoothing out driver control or something. Because the programming team seems to have so much trouble even getting basic functionality, they never end up getting used.

I don't mean that it's hard to design in two drilled holes. I mean that often both sides of the output shaft are used, so there is no easy place for an encoder.

Thanks for the tip about zip-tieing the encoder to the bellypan, and putting it on a wheel shaft. Wouldn't that cause them to rotate off axis?

25-08-2012 20:41

James Kuszmaul

Re: pic: 2012 FRC971 Transmission Actual

The encoders were/are used.

The encoders are mounted to a hole on the inside of the drive train, and are driven by a pair of plastic gears, one of which is mounted against the wheel, the other on the encoder. We have one encoder on each front wheel, although because of our ball intake and barrier-traversing skids, the space in the front of the robot is crowded, so it might've made sense to put the encoders on the back wheels.

25-08-2012 23:25

apalrd

Re: pic: 2012 FRC971 Transmission Actual

Encoders are very useful.


Common algorithms I've implemented with drivetrain encoders:

-Autonomous straight driving with speed control, distance thresholds (each side has a speed controller and stops when distance hits a target)

-Autonomous straight driving with a dual distance controller and distance error controllers - Each side has a distance controller. The steering input is calculated from the integrated distance error (sorta like using the distance delta between the sides as a heading input in an I-controller gyro steering loop).

-Autonomous straight driving with dual distance controller and gyro heading correction - This is my favorite. Each side has a distance controller, plus a single steering controller based on gyro feedback primarily using an I term. When properly implemented and tuned, this is my favorite autonomous driving algorithm.

-Autonomous stop decel controller - We drove fast enough to jump on decel, so we implemented a PI controller to do a controlled decel. We could play with the PI gains until the decel was repeatable.

-Teleoperated speed control - Each side has a controller for speed, allows you to "push through" in situations where you request partial power. You can also straight drive when one side is unable to meet the demand by lowering output power on the other side, which is sometimes also used by trans algorithms.

-Automated Trans control algorithms often rely on encoder speed feedback in addition to driver demands. Some downshifting cases are primarily based on vehicle speed and/or acceleration.

Some of these controllers are more useful than others. I highly recommend playing with some autonomous controls during the off season, since they are often a large factor in autonomous reliability. Simply measuring distance and stopping at the right time can make autonomous significantly more reliable in most games.

29-08-2012 14:42

GDG 2337

Re: pic: 2012 FRC971 Transmission Actual

To make the design even more compact rotate the CIM motors like Killer Bee’s (FRC 33) 2011 design and use the CIM motor bolts to hold the transmission halves together.

30-08-2012 10:54

Adrian Clark

Re: pic: 2012 FRC971 Transmission Actual

Assuming I understand you correctly and what you're saying is that the CIMs should be rotated so that one of the mounting bolts can be used to mount the standoff: if the CIM was rotated in such a way then one of the bolts would be behind that massive low-speed dog gear. Although mounting gearbox standoffs to the CIMs is a great way to simplify the design, in this gearbox because of the way they gear their CIMs, the CIMs are much closer to the dog gears making it so the CIM cannot be oriented in any way much different than its current orientation.

30-08-2012 12:11

GDG 2337

Re: pic: 2012 FRC971 Transmission Actual

In IanW's Robot Tutorials, see Making a 2-Speed Custom Gearbox for pictures of the concept. The long 10-32 screws can be found at specialty nut/bolt places. I bought ours at Mid-State Fasteners, they are pricey.

30-08-2012 13:25

M. Mellott

Re: pic: 2012 FRC971 Transmission Actual

Very nice design!

So it's my understanding that half of the final gear pair is attached to the wheel and the other attached to the transmission--please correct me if I misunderstood.

Was it easy to line up the gears and obtain/maintain the proper spacing when you first assembled the drivetrain and installed the transmission(just bolt it up and done), or did you have to make several tweeks to get the gears to line up just right?

30-08-2012 22:25

James Kuszmaul

Re: pic: 2012 FRC971 Transmission Actual

As I recall, assembling the transmissions went well, but when we mounted it, there were a couple things which caused small problems:

• For some reason, one of the holes for mounting the transmission was the wrong size (too small), so it had to be drilled out by hand. It works now, but there was a flaw in the design at some point.
• The mounting bolts are tricky to access, with the shooter conveyor and all the electronics. The ideal place for getting an allen wrench to unscrew the bolts happens to be by poking the allen through the belly pan, and unscrewing a couple of the bolts. This is something we could try and improve on.

Overall, it went reasonably well, and come competition, we were able to switch transmissions relatively easily.