1339 Open Alliance Build Thread

[Nick_Kieft]

Team 1339 is excited to be participating in the open alliance this year!

We’re a mid-level team hoping to share some useful information about our robots this season.

Almost all of our CAD, testing, and decisions will be available at one of the following links:
Offseason Onshape
2022 Onshape
Youtube
Open Alliance Thread

To kick off our open build, we’d like to share what we’ve been up to during the offseason. Since much of our offseason was spent planning and preparing for KCMT, we decided not to be too ambitious and instead focused on constructing a solid WCD drivetrain. The robot served as a training platform for newer students and will be used to quickly mount and test prototypes during the regular season.

Robot Images: (Apologies for vertical orientation)

We used Thrifty Bot COTS gussets to assemble the frame.

The robot features two custom gearboxes, with a pair of polycarbonate plates on each side. We’ve had great success using polycarb for similar gearboxes in the past, and so far these have held up great. Their gear ratios are 6.67:1

To transmit power to the outer wheels, we opted for 15mm wide HTD belts inside of our 2x2 drive rails. We 3D-printed some custom double pulleys for the center axle. The double pulleys sandwich around an 18t gear to avoid damaging their hex bores.

We’re strongly considering using GT3 belts and pulleys for our 2022 robot. Although they’re more difficult to source, their torque ratings are significantly better than those of HTD. If we do opt to use GT3, we’ll likely use a 5mm pitch so that we can quickly swap it with HTD if necessary

After uploading code and taking the robot out for a spin, it became apparent that it was drifting wayyy too much, so we swapped the rear omnis with another set of WCD traction wheels. After this change, it drove great.

To make this robot a more effective platform for testing code in the future, we threw a limelight camera onto it with some 3D-printed mounts. The mount was mostly inspired by this post, and is adjustable but can be firmly held in place by tightening the bolts at its pivot.

The last thing I’d like to share in this post is a design calculator that I’ve been working on. I mostly made it for fun but figured I’d post it here anyway in case any other teams find it useful.
1339 Design Calculator
These design tools greatly helped me make my own:
ReCalc
Julia’s Calculator
FRC Mechanical Design Calculator
AMB Design Spreadsheet
JVN Design Calculator

Once again, the CAD for this drivetrain can be found here.

Good luck in the 2022 season!

[josesantos]

@Nick_Kieft

We 3D-printed some custom double pulleys for the center axle. The double pulleys sandwich around an 18t gear to avoid damaging their hex bores.

What filament and print settings were used for the double pulley? How well have these pulleys held up and have you tested them at close to full robot weight?

All in all, looks like a solid offseason project!

[Nick_Kieft]

The pulleys in this robot were printed in PETG at a 0.15mm layer height with 3 perimeters and 50% infill. For our 2022 robot, we’re planning on printing them in onyx.

So far, they’ve held up great with this robot (~50 lbs). We haven’t yet tested them at full robot weight, but that’s definitely something we hope to do in the future. Even without testing, though, I feel pretty confident that they’ll be strong enough based on the experiences of other teams and this video. As seen in the video, gears of a size similar to that of a 24t HTD pulley can usually handle at least 5Nm, no matter the material. Considering the fact that the pulleys will have 12 teeth engaged rather than 1-2, I don’t think we’ll run into any issues.

[mrnoble]

Well, that was fun! Every single one of my game predictions was wrong, but I’m very glad about most of them. It looks like the GDC has created a very well-balanced game overall, with many game elements that harken back to old games, and some that are essentially new challenges.

1339 spent the day reviewing the rules (we read them out loud as a group, then split into smaller groups for analysis) and trying to determine how scoring will break down, and what our most effective robot will do. Here are some of our first takes.

GAME

• Neither the climb, nor autonomous, nor teleoperated ball scoring is overvalued. There doesn’t seem to be a “if you don’t do this then you will lose” item this year.
• The amount of resources that need to be dedicated to a mechanism increase exponentially for both scoring balls accurately, consistently, and quickly in the high goal, and for climbing past the mid-rung in the end game. Teams (ours included) need to carefully consider how much they can dedicate to these difficult goals, rather than to having a robust drive train and simpler mechanisms, and getting in lots of practice early on.
• Opportunities for a single robot to win on it’s own through additional RP’s have been purposely reduced this year. No buddy climbs, and probably no solo RP in autonomous. A great robot still needs minimally competent partners to get those ranking points.

OUR PRIORITIES

• Rule Zero: get lots of drive practice, including cycling
• 1: Drive base needs to be stable, capable of fast cycles, and NEVER BREAK DOWN. There will be fast, hard hits on this flat and open field.
• 2: Must control two balls, with an intake mechanism that is both “touch it, own it” and able to take very hard hits by natural retraction and flexibility.
• 3: Fast cycles in the low goal, with 100% accuracy.
• 4: Minimum 2-ball auto, prioritize this in programming to try to get 3 or 4 balls.
• 5: Mid-rung climb, which we think can emulate mechanisms from previous years (including 2020).
• 6: High goal shot, with high accuracy (not worth it at 66%).
• 7: Don’t know yet, this is where we get into diminishing returns, or rather increased investment. Is a level 3 or level 4 climb more worth it? Or would it be better to build a turret and hood? Yet to be determined.

We have moderate machining capabilities this year, at least for the beginning of the season, and will be relying a lot on COTS solutions for the climber and shooter. We also recognize that we have some advantages that many teams currently do not (meeting in person, for now), and even so we think we should aim more for simple, robust and done rather than complex and untested.

More to come! Have a great week.

Mr. N

[mrnoble]

Good morning everyone.

I have been thinking a lot about how 1339 can best meet the needs of the community through our affiliation with the Open Alliance. Part of that question is answered already: we will be open. Y’all can ask us questions, view all of our CAD drawings and code, see our documentation, and watch us prototype, build, and test. We will be participating with FIRST Updates Now several times through the build season, including this coming week.

Still, we want to make sure that this is useful for folks. We are a mid-level team, so we do not have much insight into the ideas or process of elite teams; you won’t find here the secret sauce that wins every event, or the design that breaks the game. We will have prototypes and documents, but honestly there are other teams that have much more extensive and detailed processes than us who have been publishing for many years. Prominent in my mind (though this isn’t an exhaustive list) are teams 95 (Grasshoppers), 2363 (Triple Helix), and 3847 (Spectrum). 1339 has relied on these three teams and many others to help us make decisions and improve our designs over the years and I highly recommend that you do the same.

Here’s what I think you’ll be likely to get from our team this year that isn’t just the daily grind.

GUIDEPOSTS. When we see something cool that influences our decisions and designs, we will post them here. We aren’t the heroes, we are just thirsty people telling you where we found water, so to speak. These will often include work that other teams are doing or have done in the past, but might include non-robot stuff, like industrial machines. If you don’t already know where to look for inspiration from the past, we can help point you there.
INSIGHT. There are some aspects of this game that we think we’ve figured out, or will figure out, and when that happens we want to share it with you. Especially in cases where we think teams might end up wasting time; we’d like to save you unnecessary effort if that’s helpful to you.
DESIGN IDEAS. Some of the things we do regularly work very well and don’t require fanciness or a lot of equipment. This makes our process a lot more streamlined, and maybe knowing these things will help your teams this year or in the future to build reliable robots more quickly. We also might (MIGHT) have some ideas that will help you create a robot that does a little more than the bare minimum, and makes you more competitive.

Thanks for your time. Please contact us here or in the FRC Discord or through our emails, angelbots1339@gmail.com or jnoble@dpsk12.net if you have questions or want additional access.

Mr. N

[mrnoble]

Here’s one of those insights for you to ponder.

We are coming off of essentially a two-year hiatus in FRC, and Covid still makes so much about this build and competition season unpredictable at best. Supply chain issues, lack of meeting time, loss of institutional knowledge, and loss of financial sponsors are all impacting what teams have to work with this year.

I think that FIRST took all these into account when finalizing the game design. The movement of balls is something that any team has access to. There’s a vast, and recent, pool of knowledge on how to pick up and place or shoot balls. In addition, after having tested the ball surfaces against a variety of materials (different wheels, for example), I don’t think that there are very many materials that would be bad at picking up or launching these balls. Performance-wise, the difference between a 50A AndyMark compliant wheel 3D printed vectored intake wheel seem almost negligible. These balls in particular seem very forgiving regardless of what you choose to move them with or shoot them with (assuming you launch them with a wheeled shooter; not the only choice but a common one that seems to work well enough). It also doesn’t seem that the balls will be jamming up inside your hopper or singulator this year, as they don’t squish together like the 2020 balls did.

In summary, I think the GDC gave us a mostly solved game piece on purpose, so that more teams could find more success in the midst of all the hardships we are facing. If you need to or want to take the intake wheels off of your 2020 robot and reuse them here, they will probably work fine. You shouldn’t fret too much about finding that little bit of performance difference that takes ages of time, and lots of money, to find. I think we will all be OKAY with this game piece.

Mr. N

[mrnoble]

End of Week 1 update for y’all.

Strategy and Meta- Discussion: Our biggest conclusion is that this game is remarkably forgiving. I’m of the opinion that this was on purpose, others aren’t so sure, but in any case we have a game piece that is familiar enough to any students or mentors who participated in the 2020 or 2021 game, but is more forgiving in many ways. It seems like the climb challenge is also accessible for any teams that climbed in the last game, and may be easier, in fact (not referring to the HIGH or TRAVERSE climb levels, that’s different). This is great news for FIRST teams around the world, given the realities of living with Covid. For us it means that our prototyping doesn’t need to focus on the basics (how do we pick this up? How do we move it? How do we climb?) but rather on optimization.
We expect fairly low scoring games, with lots of challenges for cycling effectively and quickly. This means that are putting more effort toward an effective level 4 climb. Compared to 2013, that high climb was worth 2.5 full cycles, and this is worth 3.75, a significant difference. After much discussion about ways to sneak past the HIGH level we have begun to converge on a more flexible climb style that would require being supported on each level in order, but would be more stable and more flexible in terms of the requirements of each match. We intend to have the capability to climb to level 4 and are designing the robot with this in mind.
Here are our priorities, as they now stand:
MUST DO:

• Drive without fail
• Go over floor obstacles without beaching
• Quick start and stop, fast cycles
• Fit under the Low bar
• Be able to climb the Mid bar 100%
• Be able climb the Low bar 100%, as necessary
• Over-the-bumper, full-width intake
• Control two balls
• Accurate low hub shooter from set positions
• 2 ball auto
  SHOULD DO:
• Accurate high hub shooter in auto and set positions, including the Fender
• Intake and output on opposite sides
• Intake retracts on contact with other robot or other obstacles
• Climb to High bar
• 3 ball auto
• Reliable vision-based control
  WOULD BE NICE:
• Accurate shots while moving
• Traverse climb
  NOPE:
• Turret
• Swerve
• Complex or risky climb
  Our current CAD model (here) reflects these priorities.

Plan for the Weekend:

• We have just received and constructed a new CNC router, which we will begin testing on Saturday. We are excited to once again (after two years off) have this capability in-house.
• We will be constructing an easy version of the high/low hub out of PVC pipe and Pex tubing.
• Create and test prototypes for indexing and vertical ball movement
• Prototype two-wheeled hood shooter
• Make changes to existing test drive base to reflect likely motor and controller choices on the final robot
• Begin cutting of drive base components

Hope y’all are doing great. Keep it up, this will be a fun and accessible game for everyone!

Mr. N

[Nick.kremer]

Holy guacamole y’all are fast with the CAD (also thanks for posting CAD)!

[ClayTownR]

I poked around with the CAD, and noticed that your shooter hood doesn’t have a lot of wrap, so there would be less time to get a ball up to speed and might also lead to increased variability in your shots.

Do you have any plans on increasing hood wrap, or otherwise pre-accelerating the balls? Or did your prototyping just show that you don’t actually need much distance to accelerate?

[mrnoble]

Thanks, good questions. The hood you see is a placeholder, but even so we want a very high launch angle. We are prototyping a dual-wheel hood (with one on top) tomorrow using similar geometry to what you see, so we shall see if we keep it or change it.

The balls will have some pre-acceleration as well; the holding wheel is located about a foot below the hood, and the HTD belts (or whatever works) will move the ball into the hood upon release.

[InfiniteHeight]

I might be mistaken, but I don’t see a gearbox driving the extension of your climber arms. How do y’all plan on doing that?

[mrnoble]

The location of the spools for the telescoping arms is TBD I think, but they will be motor driven on each side, using the new Max Planetary boxes from Rev.

[Smtopps]

I’m curious from how far away have you found you can make reliable shots. We are in a situation right now where we can’t do after school stuff for another 2 weeks. So the only things we can do is prepare lots of CAD and CAD prototypes based on the testing other teams share. We can do a bit with the robotics course that our lead mentor runs at the school but it’s challenging.

Also @ClayTownR we may consider pre-accelerating the balls to over build it as well as designing the shooter to be adjustable after construction. Are there any recommendations for pre-accelerating since we didn’t do that in 2020. Should it be driven off of the main flywheel or a separate motor? Is there an Ideal surface speed ratio for the feeder wheel in relation to the main flywheel.

[ClayTownR]

I can’t say I know much about using a pre-accelerator on shooters. Some teams call them “kicker wheels,” so you could try working with that term instead and seeing if any CD searches get you anywhere.

I’ll also add a different option - you can get extra hood wrap on the bottom as well as the top, such as in 971’s 2017 shooter:

[Laaba_80]

@mrnoble

Compared to 2013, that high climb was worth 2.5 full cycles, and this is worth 3.75, a significant difference.

Be careful here. Keep in mind, most L3 climbers in 2013 also scored (or tried to score) the discs at the top for a total of 50 points. This would be worth 4.17 cycles.

[mrnoble]

When Things Go Wrong

Part of being in the Open Alliance means sharing the defeats as well as the victories. Today had some victories but also some very frustrating defeats, some that maybe y’all can relate to.

The good stuff: Our Imagery sub-team got all their supplies in and were able to spend the day making the most gorgeous team shirts ever. We have long enjoyed choosing and changing themes each game, ever since our team was started with a large contingent of theater kids. This year the Imagery group has been incredibly ambitious, not only creating a theme and designing the shirt graphics, but also purchasing equipment to make their own shirts.

A Cricut vinyl cutter and a large vinyl heat press means that we will have amazing outfits.

Robot-wise, this wasn’t a great day. The CNC router we had used in the school woodshop for many years finally died in early 2020, and we just (finally) received shipment or our own CNC, a Shapeoko XXL Pro, several days ago. Three students and I spent some hours following the instructions to build the machine, and last night I stayed up late prepping the CAM files to go into the machine this morning for our high-fidelity prototypes. Well, that didn’t work out so well. The machine has some bugs, and until we can troubleshoot with the manufacturer next week it is useless. Our backup plans (a neighbor’s CNC, and the school’s laser cutter) both failed as well, and we had a full day with no prototypes to show for it. Our CAD broke, too; Onshape suddenly lost all the wheels on our robot and started sending a large error message any time the drawing was moved at all.
All told, we all got tired and felt like we made only a little progress. This is part of the process, though, and probably all of you can sympathize. We will just have to get back up and try again on Monday. Not tomorrow, though. We need to take a mental health break.

Hope y’all are doing well. Hang in there!

Mr. N

[MrBasse]

I experienced the same issue in my classes for almost half my students on Thursday and Friday in Onshape. It’s really hard to teach assemblies when the parts keep vanishing! Everything was back in order by Friday afternoon, so I hope yours does the same. A few kids replaced missing pieces, but the rest they just magically came back after a bit of waiting patiently.

[mrnoble]

Monday update:

Our new CNC still does not work, but it’s been a very productive day nevertheless.

Thanks to 4481
The 4481 build blog has been incredibly useful for us. There is nothing wrong with taking a great idea, folks, you are not required to literally invent wheels in FRC. We have been struggling with the exact shooting geometry of our robot design, as well as with packaging the climber and the transitional ball indexer (from horizontal to vertical). Yesterday a napkin sketch turned into change in the CAD, and then into an overnight revelation, facilitated by the kind folks at 4481 who pointed out that this was their approach to packaging their robot.

Today that design was prototyped and it worked a charm, putting two balls in quick succession into both the high and low goal from a position directly in front of the fender. Here are the specs for what worked:

• 3" Colson Performa wheels used as the feeder wheel, with 1/2" of compression on the balls (so 9" opening)
• Back wall for both sides of the “S” curve 1/16" clear polycarbonate, zip tied to supporting tubes
• Shooting wheel 2 black Colson Performa 4"x2" wheels with no flywheel, 1" of compression
• Feeder motor was a mini-CIM powered by a modified Makita 14v drill with the chuck removed
• Shooter motors were two NEOs on a 1:1 ratio
• High goal used 72% power, low goal 40% power

Here is our shooter test; this result was very consistent.

If you don’t know the “drill” trick, it’s a great way to quickly prototype using not programming but allowing for some level of speed control. Open up any old drill, remove the motor and chuck, and add Anderson connectors (or whatever you use on your motors) to the lead wires. Then leave the front of the drill housing open with the wires dangling out, and plug them into a brushed motor like a CIM. The photo above shows our setup pretty clearly.

Thanks, hope that this info is helpful to folks out there. Our biggest takeaway from this is that these balls are as resilient as we initially suspected. You should be able to adequately pick up, move, and shoot them with whatever you’ve got lying around from two years ago. Best of luck!

Mr. N

[RonnyV]

Awesome to read that it worked out for your team!

The shot is looking great and nice to get confirmation that with different parts but roughly the same variables and geometry you get nearly exact same results.

Hopefully you can all package it in nicely and get that thing to run. Good luck

[mrnoble]

It sure was fun to watch the kids do FIRST Updates Now for the first time.