4096 Ctrl-Z "Seesaw" Drive 2017 and Drivetrain Evolution

Two years ago, our graduated drivetrain captain (Lyle) and I wanted to design and build something that would work better than the rather poor mecanum drivetrains we had built in 2014 and 2015, coasting off of our relative mecanum success in 2013. We designed and built the drivetrain outlined in the following post from May 2015: https://www.chiefdelphi.com/forums/showthread.php?t=137379&highlight=4096

A departure from our previously AndyMark drivetrains after a devastating experience with using the RAW Box for Recycle Rush, we made the switch to VexPro products (which we have been happy fans of ever since). We went with the first two-speed gearbox we ever used and bought two 3-CIM ball shifters as well as a Versaplanetary gearbox to power the strafe motor. Our core idea was to accomplish the same thing as 624’s grasshopper drive, a shifting drivetrain between a tank drive and an H drive with only two pistons, compared to the four pistons used by a butterfly drive. We also wanted reliability, ease of repair, and a low production cost, so those core concepts drove our design.

With our first foray into using belts, we definitely made more than one mistake, starting with the center distances. We ended up doing a +0.002 center add to the belt distance rather than the -0.002 we used this year, and it resulted in our belts being slightly overstretched and difficult to put on. In addition, our shifting mechanism would stretch the belt between the drive omni wheel axle and the dead axle tank wheel. On the other hand, as a prototype the drivetrain was a massive success. It took over two years of heavy use to notice any belt slippage, and it was the smoothest strafing drivetrain we had ever built. With the ability to shim down the hard mounted strafe wheel, we manually found the optimal height that let us have an amazing drive.

That Nonadrive drivetrain CAD was uploaded, with renderings, to Grabcad: https://grabcad.com/library/nonadrive-cad-with-electrical-1

After the success of that drive, we sketched up and CADed a new drivetrain, one using the sheetmetal frame we had started with in Recycle Rush instead of 8020 frames. The new drive also had a redesigned strafe module and moved the drive gearboxes to the back to leave more room for mechanisms. Sadly, (or in hindsight, rather luckily), we never got around to building it as the 2016 game rolled around and our small 4" wheel designs got thrown out the window.

Not to be deterred by some rough game pieces, I somehow came up with the idea to add another wheel and make our nonadrive a “decacanum”. Using the same core concept with our pneumatic shifting mechanism, we made a drivetrain that shifted between a 4 wheel mecanum and a 6 wheel tank, using all 8" wheels. To ensure that the mecanum worked, we had 4 single CIM single speed gearboxes from VexPro, each belted and then geared to power both the pneumatic wheels and the mecanums. The transmission also had the mecanums spin faster than the pneumatics to ensure a faster robot despite single speed gearboxes. It worked rather well, and we were able to strafe throughout competition, but with miscalculations on necessary piston force and improper usage of belts, the drivetrain had a lot of belt slippage and the shifting time was too long for comfort. It included a couple key improvements over the original nonadrive design, notably that all the shifting caused pivoting around the driven shaft, ensuring that the belt tension remained constant at all times. In addition, all the wheels on the robot were dead axle, which greatly reduced repair times from our 2015 season.

This year, Ctrl-Z continued the trend of our previous years in building a strafing drivetrain. With the large open field this year, we went with a greatly updated version of the Nonadrive we built after Recycle Rush. For lack of a better name, our electrical captain has dubbed it a “Seesaw” drive. My core design necessities were that the drivetrain was light, robust, and easy to maintain. As such, I kept the dead axle wheels of the previous year, but instead of belting the gears, had each main drive wheel be powered through gears, removing the necessity to slide belts on and off when replacing a wheel. To make repairs easy and the drivetrain robust, the 8 main wheels were split into 4 drive modules, each containing a traction wheel, an omni wheel, 3 gears, and a pivoting axle. This axle was powered by the gearbox through belts and powered the wheels through the gears, but also acted as the pivot point for the module. This made the drive easily serviceable, as any wheel could be replaced by removing a single bolt and dropping it out the bottom, and any module could be replaced by pulling out a single shaft and dropping the module out the bottom.

A full drivetrain rebuild, going from everything installed, to everything removed except the gearboxes and strafe module (taking out 8 wheels, two pistons, 8 pulleys, and 4 belts), and back to installed, took two students under an hour to do. The modules were designed in such a way that in tank mode, the omnis were barely off the ground, ensuring that the drivetrain couldn’t tip during a pushing match. The shifting piston was a small 1.0625" bore 0.5" stroke Flat-I piston, and with the 3 to 1 lever arm built into the drive modules, easily and instantaneously switched between the tank and H drive modes. Without pressure, the robot had all the wheels on the ground, which effectively became a tank drive with slightly less traction.

Since the new pivoting mechanism meant the the omnis and traction wheels were at almost the same height in both states, and subsequently the frame, I knew that we couldn’t keep the strafe module stationary as before. To make it easy to actuate and repair, I designed a self-enclosed module that had two 4" omni wheels on a hex shaft driven by two Mini-CIMs. 4 pistons situated around the corners of the module raised and lowered it within a set of linear sliders that kept it nice and even. While that required some fiddling to keep it from binding, we had no issues with it at all during the season. We also added originally unplanned springs to add more force on the strafe module, as at first the strafe wheels slipped while strafing across the floor. In total, we had 3 solenoids on the drivetrain controlling the 8 pistons inside the lower frame.

Looking back on my season, there are another few changes I would have made. For example, I would have put in another CIM into each of the drive assembly gearboxes, or even have designed a 4x775pro gearbox and used that instead. I would have also placed more strategic lightening holes for easy access with hex keys and also used 0.090" instead of 1/16" aluminum for the frame, as we encountered some issues with the frame bending at the Central Illinois Regional. We ended up adding a brace to the front end of the frame at Midwest after we bent it in a fall at the Central Illinois Regional and more structure to our ground feeder after a particularly painful collision with 2481. I also would have gone with some beefier pistons on the strafe module, and would have left more room in the frame for comfortable pneumatics tubing and wiring. We also ended up replacing all the pivot axles with tapped shafts instead of shaft collars as the collars kept slipping off the shafts. Lastly, I wish I had made the robot slightly faster than the 17ft/s high and 7.5 ft/s low free speed ended up being a lot slower than planned on in the real game.

In the end, I am extremely proud of the progress my team has made over the past couple seasons, and the success of our drivetrain during our two competitions was telling. In 14 matches at Midwest, our robot didn’t miss a single climb, the only robot at the competition to do so. Thanks to our maneuverable drivetrain and our strafing abilities, we were able to get to the rope every time and catch up and climb extremely quickly. We were also able to adjust our shooting on the fly to get the perfect arc. The durability of the drivetrain kept us the closest to unbroken that we have ever been throughout a competition, with our hardest fix requiring a couple taps with a mallet while in queue for our semifinal matches. We were able to easily align to score and feed gears and slip around defense, while keeping a relatively low complexity drivetrain that was lightweight and robust. The modular design turned heads in the pits and on the fields and our drivetrain setup left many pit scouters befuddled at what to write down. Most importantly our robot stayed reliable throughout the entire season and let us prove to ourselves that we can overcome challenges that we set ourselves.

This design challenge has left some of my team members wanting a swerve, and I feel that this drivetrain has helped our team bridge the gap between complexity and usability and make something more challenging like a swerve within reach. With our lead CADer (me) and our key coders leaving the team, I’m going to work through this offseason to ensure that next year they have at least a reliable, cheap, and easy to make versachassis WCD (preferably with some custom 775pro gearboxes). In that vein, I’m hoping to help work on designing a new robot from scratch for our offseason competitions utilizing a WCD, and prototyping and printing out some swerve modules as well. If our design evolution takes us to a 6 wheel tank next year, that will be the first time since Rebound Rumble our robots don’t move sideways, but I know that this team will have the technical knowledge and capabilities to build another drivetrain that supports us throughout the season.

Our evolution of drivetrains has been documented through a series of photos, videos, and renders in this album: https://goo.gl/photos/6DVzMF1MGjtu8Hyu8. Each picture or video has a description explaining what you are seeing, and I also included a small animation of our full robot. When I find some more free time, I’ll also post our full robot CAD, as well as put all the CAD versions onto Grabcad for anyone interested in taking a look. Until then, I’d love to hear your questions and comments, and will gratefully supply more renders and pictures and information at your request.

I don’t see any images. (Unless your robots all look like do not enter signs.)

Does the link at the bottom function?

I’ll also rehost the images linked in the text, thanks for pointing that out.

EDIT: thanks for pointing that out, everything works now.

These are some sexy renders. What program did you use, if I was too blind to notice what it was?

They look like Creo Parametric 3.0.

Thanks, that’s good to hear. It is in fact Creo Parametric 3.0 for all of the CAD. The little animation at the end of the photo album was made in Keyshot though.

$2150 for a year license :eek:

I’ll stick to SolidWorks 2015

Spot on, thanks for pointing that out.

Yay! I have images now. So…

My team does not have anywhere near the same manufacturing capacity as yours - a drill press is our precision tool. In any case, with belts, we have always done exact (as well as we can) C-C distance, with no cause for complaint. What influenced you to go + and then - on this?

It was basically something we heard. The other student I built the first drivetrain with heard somewhere that the C-C distance should have a center add of 2 thousandths, and that was our first belt drive. It worked well but it was pretty difficult to put the belts on. This year we had a new mentor who had more experience with belt drives, and he suggested we go with a center add of -0.002 instead.

Or, y’know, free for students and educators through their education program.

SW15 was available to me via my FRC team, so I understand the importance of education CAD discounts. Is PTC affiliated with FIRST the same way as Dassault, or is it just brandishing the education discount?

PTC is a strategic partner of FIRST.

TIL. Thanks for letting me know, I’ll look into it!

Maybe PTC Creo has improved, but my experience with Pro E was that it was something of a CAD version of the Turing tarpit - everything was possible, but nothing was easy.

Well said. I started using CAD with inventor, and looking back on it, everything is pretty easy but you can’t do much. Then my team got a sponsorship from PTC and have to use their software as a part of that. I’ve spent a couple thousand hours in Creo 2 and 3 so I’ve gotten pretty good and it’s extremely powerful software, but I’d stick to Solidworks if given a choice. Creo doesn’t make a lot of sense and has a steep learning curve whereas Solidworks makes sense from the start and after spending a couple hundred hours in it, I’m glad my college will be using Solidworks and not Creo.

There’s also more documentation for Solidworks as well as a larger market for someone skilled in that software.

Nothing has changed. Creo is still a really awful software to use for FRC. During the past three years I have worked with it I have maybe used 25% of it functions. I have absolutely no idea what 75% of the software does, or how to learn what it does. Since PTC pulled their funding from my team (of course they let us know a few weeks before build season so we didn’t have time to switch then) we are going to switch to Inventor or Solidworks for the offseason.


There are so many issues with Creo, from learning how to use it to file management. From using SolidWorks for four years to switching to Creo this past year, I now realize how privileged I was. The GrabCAD-SolidWorks collaboration tools make a very streamlined and powerful tool for most FRC teams, as well as professional applications/companies.

Gears should have a center add 0.002 - 0.003 (you want to have some slack so they don’t bind)

Belts need to have a center subtract -0.002 - 0.003. Those timing belts should not have banjo string tension.

On the strafe module: any issues with turning scrub? Linking the front and rear strafe wheels like that should cause turning scrub.

Bracket the side pistons attach to seems pretty flimsy.

Any issues binding the side pistons?