FRC 3506 - YETI Robotics - 2022 Build Thread

Welcome to the first ever 3506 YETI Robotics build blog!

This year we joined Open Alliance to share and support a great effort. We really believe sharing our processes among teams raises the visibility of veteran robot building techniques and helps positively support the competitive nature of FRC to raise the competitive ceiling for everyone.

At our home in Charlotte, North Carolina we have a full size field and workshop and we have always had an open door policy that any team can come into our space and learn. There are no secrets and we share everything with everyone! YETI is an experienced team that has helped start and support teams every single year since inception. We think that we have a lot to share with other teams and we are also really looking forward to the positive community critique of our processes to be a better team!

About our team:

YETI is a student led community based team based out of Charlotte, NC. We exist as a part of the Queen City Robotics Alliance (QCRA) along with several other FRC, FTC, and FLL teams. Since we do share our space and field with teams in the area, you might see other robots in our practice area and shots of other robots during our blog!

Our team composition this year is interesting because we have grown by about 50% as a team year over year. Here are some metrics about our team:

Student count 62
Mentor count 24

Who are we?

  • 68% minority
  • 28% of our team is female
  • 26% of our mentors are female
  • 26% of our mentors are minorities
  • 36% of our team are not public school students
  • The majority of our mentors are FRC Alumni from several teams
  • A whopping 96% of our team is grade 9-11 so our team is mostly comprised of rookie team members!
  • Only 2-3 students have been through a full FRC season without an ongoing pandemic so we are poised for a challenging and fun season!

Schedule:

Our team meets from Wednesday through Friday 6pm to 9pm Eastern and on Saturdays we usually go from 9am to 9pm on average, taking the day in shifts.

Open Alliance:

We plan to update and share our progress 2-3 times a week via this thread (and discord) with major milestones discussed in our Saturday postings. Our team is divided into sub teams for Controls and Mechanical which include the Programming, Electrical, CAD, and Fabrication skillsets.

We use SOLIDWORKS as our CAD solution and we use GrabCAD as our document management solution. CAD will be made available upon request to make it easier to share meaningful data.

Resources:
https://www.yetirobotics.org/

We hope teams will use this resource in ANY way they see fit. From learning about a new building technique or replicating one of our designs to get them past a design hurdle let us know how or what we did to inspire! Imitation is the highest form of flattery for our team.

In closing, we are looking forward to a season on the Open Alliance with you all!

12 Likes

YETI had a great kickoff day. We primarily brainstormed climber and shooter concepts. We heard a lot of wonderful concepts yesterday (attached below are some concept sketches). We are exited to continue the journey through build season with all of you.

1 Like

Hello all!! I am exited to share our progress with everyone. Today we expanded on intake and hopper design concepts. As we started the day we only had rough sketches and thoughts but I am glad to share that we now have two prototypes that are sound designs. Our intake will be able to pick up cargo off the ground. Our hopper does a great job of getting the ball into the neck. As of now our cargo handling strategy is to funnel the cargo straight through our robot and directly into the shooter. Here are some images of our intake and hopper/neck prototypes. We can’t forget to give a huge shoutout to our wonderful mentors who are building the field. Today they where able to construct the hangar and are currently in the progress of building the rest of the field. The field is also open for use to any team in NC. All you have to do is contact one of our lead mentors (one of them started this thread)

Edit: Added images

4 Likes

As I read other teams’ build logs I am quickly realizing some things we have forgotten to mention. The strategies we have decided on is as follows.

Would be nice to have:

  1. Drive
  2. Climb mid
  3. Shoot high
  4. Ground intake
  5. Open canopy
  6. Shoot low
  7. Climb high
  8. Climb traverse

What we decided on:

1.) Drive
2.) Climb high and traverse
3.) Shoot high and low
4.) Ground intake
5.) 2 ball auto
6.) Have an open canopy
7.) Shoot low goal

Keep in mind this seems very ambitious but we will most likely make changes to our order of priorities (the “what we decided on” list) and possibly not include some of those elements on our robot.

WHAT A THIRD BUILD DAY. Today we expanded our prototypes and started to ut some prototypes in cad.

DRIVETRAIN UPDATES:
We have decided on a 6 wheel drive west coast drive train. With the wheel orientation of 2 traction wheels in the back with one Omni in the front with no drop center for maximum stability. At this point, the gearbox has been mounted on the middle wheel instead of the rear. We are very near to having a completed CAD model of the drivetrain. (I will upload a ss of it when it is finished)

INTAKE UPDATES:
Our intake takes heavy inspiration from teams 148, 118, and 254. We also based our 2021 intake off of their three-bar folding intake style. We really like this intake style for its versatility.

NECK UPDATES
The system that moves the ball from the ground intake to the shooter (the neck) made immense amounts of progress. We figured out how much compression is ideal for cargo. We also determined that a straight line neck would be the best choice in order to reduce points for the cargo to get stuck and to prevent losing game pieces mid-game (after intaking).

SHOOTER UPDATES:
We are essentially repurposing our 2021 shooter for this game (I will attach a photo). We have yet to decide whether or not a turret is worth the risk but it is definitely something we are heavily considering.

CLIMBER UPDATES:
We made sketches in Solidworks with proper dimensions to prove that our concept would work for our high/traverse climb. The basic idea is that we will have 2 climbers and one climber will initially attach to the mid bar and the second will extend to the high bar once we have fully climbed on the mid bar. After the second climber has a grip on the high bar the mid bar climber will unattach itself from the mid bar and attach itself to the high bar and the whole process will repeat itself to get us to the traverse bar. We are still brainstorming the how but we will update this thread with our results as to what we decide on doing.

Neck design vv
Early neck design
2021 shooter vv
2021 shooter
Climber proof of concept sctech vv
Climber geometry sketch
Intake inspiration vvv


148 Intake inspiration vvv
148
254 Intake inspiration vvv
254

5 Likes

Our latest update comes with a ton of information so here goes!

Yeti has been working hard to prototype several solutions for the first week and we broke into smaller teams to accomplish this goal. Our thoughts on our design direction - use whatever we have available with a mostly rookie team and try to Recycle Rush our way to completed designs.

How we prototype is kind of weird to anyone passing by, or maybe not? We have several wood drive bases that are cut to scale that represent our robots. Someone recently called it “Wood Coast Drive” we might keep that!

For the last to years, every robot has started out as a little frame, including SubZero here with the wood twin.

The first thing we noticed about this ball during prototyping is that light compression seems to work quite ok. Our roller prototype was designed to see how much compression it takes to sling the ball through our various subsystems to decrease cycle time. Initial measurements say that about half an inch or so works with our little flex wheels but more tuning is required. With that in mind, we started prototyping intakes using some of our old Whiteout (2020 robot) intake copies. Turns out the small flex wheels and nitrile rubber wrapped around a WCP roller work pretty well!

whiteout intake testing

One of our highest priorities is the mid climb. I don’t want to slap a done sticker on this quite yet, but if we wanted to be finished right away with that we might as well be. We created an offseason climber for Whiteout our 2020 robot. This design was actually inspired by Spectrum’s offseason climber but I don’t know if they ever got theirs to work at competition. With our spin on the design, we were extremely successful. The robot climbed every match at THOR without fail and rocketed us to number one with a 2 second lift. The climber is too tall to be legal for this game, but if we lopped off the hook rods slightly, it would be legal and work.

Some details about this device…
whiteout offseason climb

It is a chain driven two stage elevator driven by two Falcon 500s (this we have found is overkill). Maybe Lance can post the ratio, this custom gearbox was his baby. This box worked better than I thought it would. We had our rookies work on it and there were certainly some egg shaped holes.

whiteout climber gearbox

The small air piston and WCP ratchet brake did the trick here. We threaded a small rod, stuck it on a shifter piston, and pushed a screw through it into the pawl. For those designing a brake like this, make sure you put this on the first stage so you don’t need to fight additional torque!

The sliders are the magic bullet for us and I am sure a lot of teams are wondering how they will do some of these custom elevators/telescopes. We 3D printed our blocks. Yep. We are insane. So is the HP 3D printer we have access to through. We are using Nylon 14 and at first we had our suspicions about these but ultimately even if we replaced 5 of them a competition we would still be cheaper in plastic than machining these bad boys. We have yet to replace a single printed part on this device after two offseason competitions and an extremely successful run at both. (it is time to replace a few, but you can see them in the photos working quite well.

As for how we attached the chain? The chain is a continuous loop. We used a WCP chain tensioner with a 1/2" tube running through it, compressed by a 1/4" bolt through the bottom of the main stage (you can see the bolt in the photo below at half extension)

It isn’t the prettiest design out there, but it is simple, effective, and easy for us to use the limit distance function to create soft limits. It will most likely make a modified appearance with some upgrades on our 2022 robot.

Our shooter has also been somewhat copy pasted from what we know. Last season our 2021 robot Subzero was the highest performing robot in NC at the shooting challenges so we decided to take it and improve it. We began by widening it, and giving a couple of options for hole patterns based on the compression test we did earlier in the week. The result was something that was easy enough to slap together, and it taught many of our team members how to tap rods properly.

We used 1/4" lexan and cut these out on our little XCarve in the back.

The pulleys are 1.3:1 if I am not mistaken and we are running two falcons on each side.

shooter concept 1.3

The flywheels we picked are identical to Subzero, but we are using four instead of two to increase the mass of the flywheel. The 4" colsons have been great for us as flywheels!


The resemblance to our 2021 robot is obvious in our design and I saw the robot peeking over as though we were mocking it for being small. We will probably experiment with another “bearing stack turret” since it will only take us a few minutes to design one based off of Subzero. The programming team is working hard on using this robot to master pathfinding for auto runs since the robot is very similar in drive system!

shooter concept 1.4

Are the smiling? I think they are smiling. Our rookies are awesome! We had to put this one together a few times to get it right. A big win as the big shooter looks at it’s older brother undergoing some maintenance as we got ready to strap the finished product onto the earlier prototype.

shooter concept 1.5

We decided to try the middle compression, only about half an inch (maybe 3/4"). Most shooter prototypes yeet the ball so far there is no way those shots will go in on the real field. We dialed our falcons down to a modest 50% to try a bump fire into the high goal and met with success our first shot high at that setting. A 30% shot allowed us to scoot one into the low goal.

From the pad we ran into a very frustrating problem. Our ceiling has a huge HVAC duct right in the middle of our field and there are also roofing nails exposed that we believe WILL catch one of our game pieces one day. We believe a 65-70% shot will land in the high goal from the pad IF we add a mobile hood and scrape below our ceiling. We have a hood design for this on Subzero, but we really don’t want to add a hood if we can control the ball trajectory with just power/speed of the falcons. So far our release angle I believe settled out to be about 15 degrees for a nice gentle arc with mild backspin.

We will post some videos of our shooter and intake tomorrow and hopefully a ton of additional content as we enter Saturday!

7 Likes

Saturday was a busy day for the team!

As promised, the videos from our last test show a high goal, low goal, and pad shot as we tested them. (settings for the low and high goal shots are posted above for anyone curious) We are confident our shooter can hit from the pad without a hood…we are just not confident we can test it inside our current ceiling height, so we started playing with a hood with a sheet of lexan to emulate tipping the ball to get a sharper angle. The jury is still out on whether we need a hood, but we are leaning towards no.

We worked on vision a bit to see what angle we needed to mount the limelight on this years robot. When we finished installing the vision strips we drove Whiteout around the field to try and gauge what angle we could mount the camera at and see the ring on the high goal from anywhere on the field. We need to adjust the angle from last year with some additional testing, but even short robots should hopefully be able to see the ring from any position on the field.

Hopefully that bit of information helps teams without a field debate whether vision is helpful or not this season. From what we can tell so far, filtering that ring out from the ambient lights for short robots is going to be VERY important.

The team continued to refine our climber design, trying a few configurations in SOLIDWORKS to attempt to resolve our current fear, if we intake from one side and shoot from the other then the shooter must clear the climber and the limelight needs to see over whatever climber system we stick on the robot. We created a sled with the climber pivot using an older frame to help us figure out the CG of our robot. We know we will need this as we get closer to a finished prototype.

climber base pivot prototype

The team also kept optimizing the tower. At one point we had a lot of dead zones in the tower that feeds the ball from the intake to the shooter so we decided to pause and do a layout. Going from a prototype to something workable, we found that the most efficient way to rid ourselves of the dead zone was to cut a ramp shape that formed to the shape of the ball. Four jigsaw shapes later and we were much happier with the shape and eliminated the dead zones by sliding some rollers around for a smoother system. The question remains for us still if we need belts or not, and we will probably go that way on the real one although for us it is faster to slap scrap roller chain on these.

tower optimization cad

The next issue to solve will be the side to side alignment that we think we can fix by using different sized compliant wheels. As soon as Fed-Ex coughs up our VERY late package we can test this. (Fun fact our gears for our drive system and some new WCP flywheels were also in this box lost in the void…)

Hopefully I can get someone to post the videos to our channel and update teams on that progress. So far, all of our compression seems to work well at around 1/2".

As we work through the other subsystems, smaller groups have been working on intake devices. We mocked up a common design used by many teams including us last year - a four bar intake. The upgrade we are trying to make here is that we want to go full lexan if possible.

We can cut out many spares and copies of that intake if we go this route, but we need to develop and practice a method of keeping the bearings pressed in during hard impacts. I have seen some teams use the head of a small screw threaded in next to the bearing to keep it from popping out so we will probably test this method soon too.

One of our rookie students was working very hard to get our intake design to a point that we can test. After a long day with a mentor over her shoulder teaching, we had a prototype sans air cylinders ready to go on the robot for testing and optimization. I don’t think she would have imagined a year ago that she could make something like this!

By the end of the meeting, I think almost every team member knows how to install roller chain and I believe I got to blow a few minds up with the age old “sprocket spacer” trick on some loose chains. Yes it works, try it :slight_smile:

Below all that plywood we have a full testbed system ready to merge the wood coast robot into a driving platform. The controls team was working diligently on our practice chassis mule we lovingly call Stumpy.

Stumpy has been a workhorse for us since about 2018. Stumpy is actually what is left of our competition robot Avalanche, but we have a full Avalanche clone sitting in storage. Stumpy has seen many gearboxes, many bumper rail repairs, and many walls from inexperienced drivers. We rue the day we need to retire Stumpy, but until that day he will continue to have our hopes and dreams carried on his tired little frame. Every team should have a Stumpy for programming, wire training, and to act as a prototype platform!

This week we got some winter weather. Anyone else under some ice and snow? Wish you were a Yeti? Hopefully this week after the roads thaw enough we can merge and drive around with our intake, tower, and shooter on one chassis to estimate real cycle times using Stumpy. By the end of week 2, we want to have the climber built for testing and a relatively finished scoring system so we can iterate on it.

Our next meeting is on Wednesday in person, until then we will work on what we can digitally and walk in with some parts ready to go for our Week 2 push! Stay warm and safe out there!

1 Like

Looks awesome! Were you able to find a solid fixed output angle that was capable of both a bump shot and launchpad? Or are you planning to trade one of those off for simplicity of not having an adjustable hood

I just want to say I laughed hysterically not expecting to hear slow motion voices in my headphones lol! Looking great you guys!

1 Like

We did find the settled angle of about 15-20 degrees! We were adjusting the flywheel power from Stumpy’s Rio (this was a proper test using a robot on blocks) as we ran those tests and not adjusting angle at all. You can see the bot didn’t move positions between our low and high goal shot, although we did slide to the pad to try and emulate that. We think driving the speed of the flywheels using encoders would allow us to not use a hood at all. We are really pushing to lower the complexity here, so that angle you see in the image is our set angle. Currently it is built from just churro standoffs which worked well enough for consistency.

This is a late update, from before Saturday, just to fill in the controls side of the week. YETI’s Controls team has been working on our testing robot: Stumpy!

We got the wiring into safe working order after about a year of sitting still in the shop since last season. In order to test prototype subsystems that we create, we have 2 extra long wires that can power motors outside of Stumpy. These extensions were used to power two falcon 500s that we used for the flywheels on our shooter subsystem. Here’s some links to videos showing demonstrating the shooter.

Just last meeting we got Stumpy to drive, after running into a lot of problems. Originally, we were working with past season’s Stumpy code and trying to fix the drivetrain subsystem. To get past the problems, we ended up creating a new project for 2022’s Stumpy and coding the drivetrain from scratch.

While that was going on, the rest of controls was working on characterizing Subzero, our 2021 robot, in order to get Pathweaver working. We’ve had a lot of trouble with it, but hopefully we can find a solution and have better autonomous this season.

2 Likes

It has been a while since our last update, but we have been slowly making progress through the weather and issues for our robot concept this season. Some of our sponsors and mentors have of course been affected by the pandemic and this has slowed progress on a few fronts (including a funeral that delayed this update). Through our network many have stepped us the help and keep us on track.

So how are we doing at the end of week 3 going into week 4?

Overall design concept:

Yeti has been prototyping and updating our concept through testing using Stumpy. We have recorded a few videos of our progress on this prototype, and from this we believe we are on track to have a robot that can do a multi-ball auto and do at the bare minimum a high bar climb with a stretch goal of a traverse. We completed the gearbox and main climber mechanism with a custom gearbox. This is actually our first fully metal custom gearbox worth mention for the team and we learned a lot on this design!

We have driven our prototype robot Stumpy around with different roller materials and shapes to try to get a performance that we are happy with between the intake and tower that leads to our shooter.

We have also worked heavily on using this prototype to get a lot of our programming done prior to completion of the robot.

We feel we are falling behind on some goals but through strategic prioritization on base line competitiveness we feel we are in a good place to be successful this season.

Drive system:

As of 1/29 during our last in person meeting, we had a huge amount of progress. We have completed our drive system on the ground with temporary wheels while we wait for back ordered items and the weather to get our parts in.

We decided to go with a different method of construction this year because we lost our sheet metal sponsor through the pandemic but gained some other sponsors that can machine components for us. This means we decided to deviate from our 2018-2020 design for our bumper system.

We took heavy inspiration from the new tube plugs from WCP and Team 78 when they posted this idea on Chief: Team 78 Offseason WCD Mule Chassis CAD

Our take on the method is virtually the same because even our 2021 robot used the same snap latches to get the job done. We went ahead and ordered our reversible bumpers and are waiting for the bumper plates to get machined at a sponsor to have our drive base 100% done.

The ratio discussion on FIRST Updates Now got us rethinking our sprint distance and after we finally got additional game pieces we ran a quick mock match and saw that our sprint distance is going to be much much lower. We tweaked our ratios a bit but ordered two gear sets for us to play with. Rolling the dice on our best performance we have this ratio currently installed:

Overall, the frame and drive system was designed and cut out by rookies with the help of our Ultimate West Coast Guide and some mentors who could teach the students to use the Bridgeport Mill so we think this was a success!

Turret and Shooter Subsystem:

The shooter from our 2021 robot was modified in our last update. The only difference here is that we completed the design 1/28 out of metal and printed a solid hood that has an integrated limelight mount instead of the churros we had on the prototype.

We also decided to test out the WCP flywheels since they don’t expand too much at the ratio we are running, and this resulted in a similar performance to the Colsons with a quick way to also increase mass of the flywheel.

The turret gear was fully 3D printed like last year, but we made some modifications to it the raise the shooter up off the gear a bit more. This was to allow for more clearance between the turret gear mesh and the bottom swing of the shooter.


20220129_112514

We decided that we wanted to go with an absolute encoder so we are going to use the integrated encoder from the versa planetary gearbox and bypass all of the complicated work we did in 2021 to zero the turret each time we started the robot.

I am mildly concerned with backlash and drift for this one but if we need to swap it we are prepared to use a Rev Through bore encoder with a 3D printed holster between the gearbox and turret plate as a backup plan.

The turret is driven by a BAG with Talon SRX instead of a Neo550 from 2021 since we cant use the mag encoder with the SparkMax. The turret plate was cut from .250 aluminum and we were using the same bearings as last year for our bearing stack. ( part numbers WCP-0041 and 217-3489 for the curious)

Tower:

We have been playing with the compression and location of the rollers to center the ball and power the ball consistently through the turret ring above it to get repeatable shots. As a test, we wrote some code using the Rev beam break sensors that runs our tower until a ball is present.

We can still run the intake separately and keep a ball lined up inside our robot to shoot. This should prove invaluable to us on the field as we run the robot blindly on the far side of the field. We will of course plan to use a camera for the driver to see where the balls are, but once inside the tower things will be hard to see from afar.

Here is a shot of what we expect to be a quick cycle if we get this right.

We decided to try belts on this design for the majority of the drive shafts so that we can transmit power at all times on the game piece and have no dead zones whatsoever. Through layout and experimentation, we know that these two lexan plates will probably change a few times until we are happy with the performance of the final tower.

I think they were tired and not happy about this last tower.

Climber

This subsystem is the most complicated by far and we are still developing it. We have iterated on the hook shapes about 10 times by now to try and satisfy the geometry that will let us high climb and traverse too.

We felt confident in the structure and ease of construction for the moving part of the climber since we had experience with the 2020/2021 version, so we decided to go ahead and make the gearbox out of pocketed aluminum sheet and use the first version to power us on and get us off the ground for further testing.

There are still some parts to machine and our final hook shapes are still in flux, but we have made great progress getting the first stage complete with gearbox and most of the second stage installed for testing.

If we wanted to just go for the high climb and ignore the traverse we think we have a geometry that will let us simply do that now. The traverse is the prize that will ultimately be the thing we either tweak a lot of things until we get it right, or we simply cut bait and stay with our stable high climb with less risk. Trust me, if we get that traverse climb we will be happy to post that victory without a blurry robot!

Intake

Ah yes, this thing. I don’t know why, but our team is never satisfied with our intakes.

Probably because they get stuck into the design last and we always but less experienced members on them? Either way, we have iterated on this intake so many times I have lost count already. If I had to guess I think our team has made about 15-20 shapes of this thing between cad and prototyping.

We are floating between two types of design that we like and running things over with Stumpy to see how they go. Every team should focus more on their intake this year because it will make or break those cycle times.

Some laughable times on this last one, this would have won champs in 2010 for juggling a ball:

For us, we want to be able to run our intake into a wall, and possibly even trap a ball against the wall and still be able to suck the ball in while other robots are hitting us. Currently we have plenty of room for our intake and we have reduced from three rollers down to two with a single strut running across.

This year looks like the year of the 4 bar linkage as most teams are trying to compact their intakes into tiny spaces, us included since we left about 4” of space for our own intake. We have a shape that works and if we went to competition with it, it would probably do ok. But we want speed! We have invested in the roller wheels that have the 1.125 inch hole in them so that we can put our intake wheels into a versa roller tube. That should allow us to plow into a wall and not bend those thunder hex rods into bananas.

Hopefully our next update wont take so long! We want to have the mechanical robot done this week for first testing and hand it over to the programmers while we optimize our shapes. The rest happens when we power on the robot and go drive into the field.

As always, if you have questions on our ratios or cad we can provide more information in the thread or on the Open Alliance Discord!

7 Likes

What material is the turret gear printed out of? And on what kind of machine? Its very nice looking!

2 Likes

If I am understanding your turret correctly, the metal gear on the right side of the picture is attached to the output of the Versaplanetary, and it turns the larger 3D printed gear. If that’s the case, I don’t understand how you can achieve the absolute positioning using the encoder on the Versaplanetary, like you suggest. For example, let’s say the small gear has 30 teeth and the big gear has 300. The Versaplanetary will tell you the absolute position of the small gear. But that absolute position will correspond with 10 different absolute positions of the big gear. So I think with this setup, there are only 3 different choices for absolute positioning:

-Zero the big gear ahead of time, which you are trying to avoid
-Always start the robot with the big gear in a semi-known position, within a 36 degree range, so that you which ”section” it’s in, and can then figure out it’s absolute position. That’s a pain, because you need to remember to do that every time.
-Limit the motion to just 36 degrees, which sounds pretty suboptimal

Are you planning on doing one of these, or am I misunderstanding something about your setup?

Regarding your comments about backlash and draft: you won’t see any drift in this setup, unless you are somehow skipping teeth on your gears, which seems unlikely if everything is manufactured well. You’ll see a small amount of backlash, but that backlash will be measure against the number of teeth on the big gear, not the small gear. The most you backlash could have is 1 tooth of the big gear (and realistically much less). That corresponds to 1/300th of the big gear, so just a bit more than 1 degree. That’s almost certainly fine, given the width of the goal this year.

2 Likes

This gear is printed on the HP Multi Jet Fusion 5200 out of Pa 12 (or Pa 11?) The parts come out of the powder and need to be blasted and dyed with black shoe dye to leave that appearance. Overall, it allows for large and stiff parts that hold up really well to abuse. I think the material cost for this part was right at $50 with pocketing removing powder. Since the extra is recycled, it keeps material costs low when we throw the parts in with other jobs.

If the parts fits in a 15x11x15 we get the parts back quickly (prints take no longer than 18 hours for a maximum build volume). I will try to take some photos of our printed pulleys this week which are next in the build if anyone is curious.

5 Likes

Regarding the encoder, we’re having a little trouble understanding what the issue would be here.

You’re correct in that the smaller gear will be turning the bigger gear (the turret ring), and that we are using the encoder on the VersaPlanetary. Can you elaborate on why only knowing the absolute position of the small gear won’t allow us to also know that of the big gear?

Continuing the example with the 30:300 ratio; in this case, couldn’t we take the value given from the encoder and divide by 10 to yield the position of the larger gear?

But that absolute position will correspond with 10 different absolute positions of the big gear.

If I’m understanding correctly, this would be true if the CPR of the encoder matched the number of teeth on the gear (i.e. 30); in our case it’s actually 1024, thus producing a fairly precise measurement of the position of the big gear.

Again, please correct me if I’m wrong. Even if we made a mistake, we’d love to know now so we can correct it ASAP. Thanks!

An absolute encoder has a position readout for any angle of the encoder from 0-360 deg. Since you have another ratio after the encoder, your encoder will rotate more than 360 degrees for one single rotation of your turett.

Typically absolute encoders are geared 1:1 to the final output which will allow you to know exactly what angle the turett is at. Because there is a ratio there will be multiple positions of the turett that correlate to one single absolute encoder position.

As you’re describing it you are essentially using your absolute encoder as a standard quadrature encoder. Seems better to have a turett zeroing sequence and then just use a quadrature instead of an absolute not geared 1:1 to the turett.

3 Likes

Let me try to demonstrate with a picture:
image

-The red circle represents your small gear
-the black circle represents the big gear on your turret.
-The blue line represents the direction your small gear is facing when your encoder is reading 0 degrees.
-The green line represents the direction your big gear is facing when your encoder is reading 0 degrees.

I drew the green and blue arrows both facing up in this picture to simplicity, but that’s arbitrary here with no loss of generality.

Now, suppose your small gear makes one full rotation clockwise, so it is pointing back to the same spot. So the encoder reads 0 degrees again (since you are using it absolute encoder mode). In doing that, your big gear turns 1/10th of a rotation, so now it is pointed towards the orange line.

The point here is that your encoder can’t tell the difference if your big gear is pointed towards the green line or the orange line. There are 10 such places that the big gear could be pointed towards that you wouldn’t be able to distinguish from each other.

Finally, note that there’s nothing special about the fact that I started with 0 degrees here. The same thing happens with any starting point. For example, when your encoder reads 180 degrees, you don’t know if the big gear is actually pointing 18 degrees counterclockwise from the green line, 54 degrees, 90 degrees, etc.

3 Likes

We thought about using one mag switch to zero off of. Last year we used two physical switches and would zero by starting the turret compressing one of them. It was super annoying to have to remember to zero it every time. I wonder if we can swap components here still and avoid it. We have the option of swapping to a neo550 with the throughbore still.

You could use an external absolute encoder or potentiometer with a printed gearset that will allow the sensor to be geared 1:1 with the turret gear.

Swapping to a Neo 550, or to the Rev Through Bore encoder won’t help. As long as you are not directly measuring the turret, or something geared 1:1 with the turret (like @Peyton_Yeung suggested), you’re going to have the exact problem that I described.

As for remembering to zero it, you don’t have to do that manually. In particular, when your robot first starts up, have the very first thing that it does be to rotate the turret slowly until the limit switch (or magnetic encoder) is triggered. Now you know the precise position of the turret, and everything can be measured in relative space after that.