Date post: | 07-Jul-2018 |
Category: |
Documents |
Upload: | jared-hasenklein |
View: | 216 times |
Download: | 0 times |
of 28
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
1/28
Team 1836:
Milken Knights
2016 Engineering Resource
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
2/28
This document was created with the intent of helping other teams
fully understand our team culture, design process, and robot. We have
gotten inspiration and guidance from dozens of other teams and we
wanted to create an open source platform to share what we have
learned. We publish several other resources for teams on our website
including a blog, game test, FLL resources and Chairman’s resources.
This year these documents were used by over 500 teams. That is 1 out
of every 6 teams in FRC. We track this with a pop up on our website
asking people what team they are on. Please ask us questions! We
love helping other teams. And, let us know if there is something we
can do to help your team!
2
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
3/28
Table of Contents
Systems 4
Drivetrain 4
Intake 7
Electronics 10
Catapult 11
Iteration Cycles 15
Improvements 16
Cost Accounting Worksheet 27
(CAW)
Contact 28
3
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
4/28
Systems
Drivetrain
This year’s drive train is an improved version of the West Coast-style drivetrain that we
have been improving on for the past five years. It can traverse all of the driving
obstacles, park on the Batter, and accurately turn. We prototyped three different
drivetrains; a standard West Coast Drive design, an adjustable West Coast Drive
design, and a four wheeled suspension.
Wheels
The drivetrain is an eight wheel West Coast drive with 200mm (7.67 inch) inch
pneumatic wheels. Different amounts of wheels and types were tested, including six
wheel drives, 6 inch pneumatic wheels, and plastic Skyway wheels. This combination of
wheel size, amount, and type was found to most easily traverse the defenses.
Center to center
While past drivetrains have used sliding bearing blocks to tension the drive chains, this
year’s robot has set wheel center-to-center distances of 8.055 inches. By having set
center-to-center distances, we drastically simplified the previously complex drive rails
and bearing blocks, eliminating many hours of manufacturing time.
4
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
5/28
Chain
Three of the wheels on each side of the robot are driven using chain by the main
gearboxes. While many teams use #35 chain, like previous years, this year’s drivetrain
uses #25 chain to maximize the weight available for other parts of the robot. Drive
chains are installed by stretching the chain over the drive sprockets to the point that
they appear to be too tight. The next step is to “wear in” the drivetrain for 20 minutes.
While the chains are initially over tensioned, by the end of the run-in period they are
perfectly tensioned.
Bearing Blocks
The bearing blocks are custom designed and machined. We used a technique
commonly referred to as “drop center” to mount our bearing blocks at differents heights.
The hole of the outer bearing blocks are offset upwards by ¼ inch, which allows only
four of the eight wheels to touch the ground at any time. This reduces the drivetrains
effective wheelbase, which allows it to turn smoothly.
Shifting and Gearing
Each side of the drivetrain is driven by a modified 2 CIM Ball Shifter gearbox, geared for
16 fps(feet per second) in high gear, and 7 fps in low gear. The high gear was
specifically chosen as a balance of speed and acceleration to traverse this year’s field
quickly. The low gear is used to push defensive robots and park on the Batter.
Sensors
The drivetrain uses US Digital’s S4T360 encoders in each gearbox for position and
velocity sensing and a NavX IMU for heading. Closed loop, velocity PID, is used on
every control loop as the base layer that maintains consistent autonomous and
teleoperated performance. By “closing the loop”, feedback from the encoders ensures
that wheel velocity is consistent no matter battery or field conditions. On top of that is
5
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
6/28
layered position PID and turn PID to drive straight and turn accurately during
autonomous.
6
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
7/28
Intake
The intake is a motorized arm that can be raised and lowered in under a second to acquire
balls. It can also be used to cross the Cheval de Frise. Because the intake serves so many
purposes, our team went through an extra iteration and design work to ensure that the intake is
extremely robust. In particular, iteration to a chained gearbox and specific software and
mechanical failure modes has prevented in match failures. However, in the event of a worse
case scenario, the intake arm hard stops against the bumpers, allowing the robot to continue to
pick-up balls. If our intake does not work we can no longer shoot or low goal so we took extra
precautions to ensure it would work.
Chain
Because the intake accelerates quickly, large amounts of force are placed on the arm
gearbox. An initial prototype intake used a gear final stage, however the final stage
pinion sheared. The next iteration used a new gearbox with the final reduction stage
changed to #25 chain, which is far more robust. One issue with chain is that it stretches.
We utilized a sliding tensioning sprocket to keep the chain tight, eliminating backlash.
This allows us to move our intake with significantly more accuracy and precision. The
chain is easily installed and tensioned through this method and can be replaced in a five
minute match Timeout.
Fastening
Brass pins are used to geometrically locate the sprocket to our intake arm frame, and
then screws hold the entire assembly together. As a result, the intake assembly is
extremely rigid and strong, ensuring that the intake performs consistently and robustly
throughout the season.
Gearing
7
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
8/28
The intake rollers are powered by a Bag Motor on a 7:1 Vex Versaplanetary gearbox.
This results in a 10 fps surface speed, quick enough for a ball to be acquired the
moment it touches the front intake roller.
Funnel
On the intake, there are two HDPE (High Density PolyEthylene) strips funneling balls
through the bumper. We tested other materials and found that this slippery surface was
optimal. The full width intake provides a large sweet spot for drivers, easily acquiring
balls regardless of ball position.
Los Angeles Regional Upgrade
We were behind where we had anticipated being on stop build day so we made the
decision to bag the drivetrain and catapult and bring everything else in our 30lb
allowance. Because of this decision we had to basically assemble our robot there. Our
30lb allowance included all of our electronics, wheels, and intake. We came to this
decision because we didn't want to bag anything that wouldn’t be ready or up to par with
the products that we produce. Thursday at the LA regional was a long first day in the
pits. Before competition we had made a detailed plan of what order we were going to
assemble our robot. We prioritized our electronics first before we started to do anything
else. Once we got our electronics on and all of that was adequately tested we got the
wheels on and made sure that all of our modifications worked. Once that was fully
finished, we decided to add the intake.
After the Los Angeles Regional
After the Los Angeles Regional, we made many changes to our practice robot which will
be implemented at the Orange County Regional. After the LA regional we realised that
to be a top tier team, we needed to add or upgrade a few things. We acknowledged that
we needed to use our catapult during matches. In shop, while doing testing, our catapult
broke in half, which meant that we needed to re-machine and design it to be lighter and
8
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
9/28
more precise. We made the catapult slightly longer and lighter, by adding 3 holes that
decrease in size while going down the arm. We also realized that we needed to
redesign a new ball clamp for our robot because the existing ball clamp prototype added
too much weight to our catapult and negatively affected our shot. We are redesigning
the clamp and have a few new ideas. One of the existing ideas for a new ball clamp is
to have two actuating cylinders, 1 on each side of our drive train, that will actuate out to
hold the ball in place, and actuate back in right before we take the shot. The team is
also considering a passive solution that will allow the ball to bounce around but not fall
out.
Sensors/Programing
The intake arm position is controlled via a PID loop using feedback from an S4D
encoder on the final gearbox
stage. The intake encoder starts
the match against a hard stop,
which is the stowed, “zero”
position. In case of encoder slip
or robot blackout during the
match, the intake arm can be
slowly driven up and back into
the hardstop, rezeroing the arm
position. In the down position,
the control loop is voltage
limited in order to prevent
premature breakage from impacts with defensive robots and field obstacles. Finally, the
intake and catapult are coordinated to prevent the drivers from accidentally colliding the
mechanisms.
9
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
10/28
Electronics
Speed Controllers
The robot uses Talon SRX motor controllers, located on one of the electronics panels of
the robot. Twelve talons are on the right electronics panel in total, four of which power
the drive gearboxes, one of which powers the intake gearbox, and one of which powers
the catapult. Though not all speed controllers are being used at the moment, the team is
prepared for the case of more mechanisms being added to the robot. Having extra
talons pre-wired also allows us to easily switch in case of motor failures.
Battery Mount
To mount our battery this year, we used a group of four 1x1 tubes and a hook and loop
fastener strap. The tubes, riveted to the bellypan, form a tight fit around the battery, and
the strap holds it down. Together, they form a secure battery box that prevents the
battery from becoming dislodged.
Connectors
Every electronic connection on the robot is crimped instead of soldered. Crimped
connections are more reliably and quickly made than soldered connections. High
voltage, power runs, are connected with Anderson Powerpoles. These connectors can
transmit up to 45 amps of current, are colored to show polarity, and can quickly be
disconnected. For signal wires, PWM connectors are used, which have a male and
female side. A female connector on each wire plugs into male side on the RoboRIO.
10
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
11/28
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
12/28
instead to let our prototypes make our choices for us. The catapult’s accuracy and
simplicity ultimately made the decision for us.
Cam
A cam is a rotating piece that is used in a mechanical linkage to transform rotational
energy into linear energy. The snail dropcam was originally used in timekeeping devices
to shift between days at exactly midnight. The void created by the snail shell or comma
like geometry creates the firing effect. We decided to use this technique to actuate our
catapult for several reasons. Consistency and repeatability has been a huge design
consideration and our current iteration is only capable of one shot in order to prioritize
these outcomes. Tuning of this mechanism happens outside of the actuation with the
adjustment of spring tensioning and launch angle. By keeping the firing mechanism
constant, it allows us to keep
consistency on our shot throughout
the tuning process, isolating the firing
from other variables. This mechanism
has proven to be tremendously
reliable and repeatable. Our
prototype was hastily made out of
wood and used tape as spacers and
yet was still capable of a 95% shot
accuracy from 17 feet.
Sensors/Programing
The catapult utilizes both an encoder for position and velocity sensing and a hall effect
for zeroing the encoder. The zeroing process starts with a velocity PID loop to spin the
cam at a constant rotational velocity, regardless of the current position of the catapult.
Once the hall effect triggers the robot to zero the cam, a position PID loop takes over. It
12
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
13/28
feeds into the velocity PID loop, moving the cam to the reloaded position. When the
driver triggers a shot, the position loop moves the cam 360 degrees, and the process
starts over. The layered PID loops and zeroing processes ensures that the catapult
performs consistently in all conditions.
Hardstop
The hardstop for the catapult was created using sailing line. We tested various lengths
to optimize the ball’s release point.
Gearing
The gear box on the catapult was custom designed. We used a gear ratio of 273.32:1due to its high torque and fast reload time. Our reload time is approximately half a
second which allows us to quickly retract into a position to fit under the low bar defense.
Forces/Springs
In order to launch the catapult, we decided to use springs. The process of choosing the
correct springs was lengthy.
We started with surgicaltubing on our prototype robot.
We calculated how much
force the surgical tubing
applied. From there, we tried
out various combinations of
springs with the surgical
tubing as a guide. Wecalculated the work done by the springs, and used this to pick which exact spring we
wanted. We ended up using two springs with 38 pounds of force each.
Trajectory
13
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
14/28
Adjusting the hard stop in conjunction with the spring force allows any trajectory to be
achieved. The catapult’s trajectory was tuned to create the biggest sweet spot possible
from the 14 ft range, which is the distance from the middle of the outer works. In fact,
the chosen trajectory will score anywhere from 7 to 17 feet away. Additionally, the high
release point makes it difficult for a full height defender to block the shot without getting
a penalty.
14
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
15/28
Iteration Cycles
This year we have implemented a high iteration, fast prototyping process. We have a
saying on our team “fail fast and fail often.” Learning what doesn’t work helps us getcloser to finding out what does. We have made over 30 iterations and prototypes
throughout this season. This allows us to improve on our designs through testing, not
theory. We had a really difficult time last year making design decisions as a team. Our
solution to that is prototyping everything and allowing the mechanisms to make the
decision for us. This is a huge step for our team because it allows us to focus less on
making the perfect decision, and more on improving based on actual data. We have a
term we have coined “MVP” (Minimum Viable Product). This prevents us from getting
carried away and helps us keep our designs simple and elegant.
15
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
16/28
Improvements
Scrum
As a part of improving our team, we have adopted the Scrum method. Scrum is a way
of creating a team approach to problem solving and innovation. Scrum keeps our
leadership on track to make sure that we are accomplishing all of the goals we set.
Using this method, our leadership created a spreadsheet where the team could keep
track of all ongoing progress. The spreadsheet includes a brief description of the
project, what has been completed, what needs to be completed, and the percent
completed.
Performance > Appearance
16
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
17/28
After much discussion the leadership team unanimously decided to prioritize
performance over appearance. We have powder coated our robots before competition
every year which took approximately 4 days. Our bare metal and unpainted wood may
be less aesthetically appealing but it allows us to put our resources on other areas. After
the competition season we will powdercoat our robot, leaving us with the same end
product and more time during build season.
Decreased Tapping
This off season a great deal of reflection took place as we thoroughly evaluated every
aspect of our manufacturing process. During last year’s build season we used a tap
wrench to thread over 100 holes. This process was extremely time consuming. The
alternative our team decided on for this year is using a bolt and lock nut but requires
extra space and access holes. Putting a little more time into our designs ultimately
saved us manufacturing time.
Battery Cart
In the offseason, the team decided to build a
better battery management solution. In order tohave a fresh battery for every match, with spares
for alliance partners, the team brings 12 batteries
to competition. Along with these batteries, the
team must also bring 4 chargers. As a result,
packing batteries and their chargers, alone
weighing over 200 lbs, was a time consuming
process. The team decided to combine all this equipment into a single battery cart toease the packing process. This was also a good pre season project to prepare students
for build season and gain experience working with baltic birch.
17
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
18/28
The resulting battery cart design had a couple of goals. Firstly, it needed to be easy to
manufacture with the team’s in-shop resources, yet still robust. This was done by using
a largely wood, puzzle-like structure, with a few metal parts used as frame members.
The battery cart also needed to effectively store all the equipment, while being easy to
roll around and fit in the pit. The design is short enough to fit under our pit tables, yet
extremely compact.
Battery Cart Reflection
It was a bit tight to assemble, in part due to some manufacturing issues. In the future
the dimensions of the pockets/holes should be increased to make assembly easier.
The nuts like to twist in their spots, so making the nut slots thinner and more rectangular
could prevent this.
Moving the nuts closer or using longer screws and extending the screw clearance would
allow for the screws' interface with the nuts to occur earlier
Right now, the SB50 battery connectors are locked in place, so it is impossible to add
charger banks and adjust the SB50 connectors without disassembling the cart. This
could be fixed by tapping or using rivet nuts in place of the nut and bolt that holds the
SB50 in place. Drilling a hole in the shelf would allow the SB50 bolt to be accessible.
The hole drilled for the power strip should be added into the CAD. Putting a plastic
piece in the hole would prevent the power strip cord from wearing.
Adding an inset handle into the design could be helpful for optimum transportation, just
in case it needs to be placed flush.
18
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
19/28
The material left in the pocket is extremely thin. In the future, making it .1" thick , if not
.15", would be better.
The design should be modified to avoid having to disconnect SB50 connectors from the
chargers (via the fuse holders). 3 fuses broke in this process.
There isn't a lot of space for wires, especially ones with large plugs, and so the cart
must be assembled in a very careful process. Leaving more space for wires without
significantly increasing the cart height could be improved.
Leaving a cutout in the top piece would make it is easier to see the charger lights.
The cart is extremely rigid... maybe too rigid even. The biggest time sink was the bottom
2x1 added for rigidity, so there could be alternative options here.
There may have been issues with radii being exactly .125". The next step is looking into
this problem to see if making the radii .13" would help.
Locking casters would probably be a better choice, as it takes a good bit of force to
insert the battery connector
We could probably use less fasteners on the L battery backstop
There should be a better plan for painting/staining going into the assembly process
All in all, the team is extremely happy with how this project turned out and we enjoyed
testing it full time this year.
Pit Checklist
19
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
20/28
To improve our consistency preparing for matches we created a checklist that pit crew
would implement before and after every match. We also use this list to select and train
pit crew and run a series of time trials. We have set up our pit in our shop and practice
to prepare for competition.
20
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
21/28
Wood
Wood has a negative stigma in FIRST and is improperly associated with poor design.
Before this season we had almost never used wood. Wood has the major benefit of
machining approximately four times faster aluminum and in some cases, even faster.
Last year our intricate lattice structure belly pan design took 8 hours to machine out of
⅛” aluminum. In comparison, this year’s wood belly pan took only 15 minutes.
One common complaint about wood is the reduced strength. We are using grade B/BB
Baltic birch, significantly stronger than average, Home Depot plywood. 6061 aluminum
has an ultimate tensile strength of 45 ksi ( kilopound per square inch) while baltic birch
has an ultimate tensile strength of 10 ksi. 6061 aluminum has a modulus of elasticity of
10,000 ksi while comparatively baltic birch has a modulus of elasticity of 2,000 ksi. This
data appears to show that Aluminum is much stronger than baltic birch but when you
adjust for density they have the same ultimate tensile strength of 16.7 ksi. They also
have a similar modulus of elasticity when adjusted for density with aluminum being 3703
ksi and baltic birch being 3333 ksi. Instead of making plates out of 1/8 in 6061 aluminum
we now use a 6 mm (0.236 in) sheet of baltic birch. Plates that would have been made
out of ¼ in aluminum are now made from sheets of 12 mm (.472 in) baltic birch. These
baltic birch plates are approximately twice as thick as the aluminum equivalent which
was not always feasible on the robot due to the tight packaging constraints. The void free
core is more uniform than other plywoods and allows for a stable crossband lamination.
Our prototypes were completely made of baltic birch plates. As the design evolved
through the iterative design process we have switched some plates to metal. The belly
pan, upper structure support plates, electronics panels, intake plate, and catapult bucket
arm are all remaining as baltic birch for the final robot.
21
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
22/28
Preseason
Training Document
A challenge in the past has been deciding how to effectively prepare new
members. This offseason, our team captains compiled a list of every skill
necessary to participate fully on our FRC team. We compiled these tasks into 3
different categories: green, yellow, and red. Green corresponds to a “beginner
task”, yellow for “intermediate”, and red for “expert”. This system allowed us to
track all team members’ progress, not only ensuring that each team member was
proficient in every possible skill, but also that each and every team member can
point at a part of the robot and confidently say: “I made that.” This training
method is being used by other FRC teams and is even being used as part of a
curriculum that is being developed for students in China.
The first and most important section of our Training Document is Safety and
Cleaning. Students must have this entire section complete before participating in
any activities in the room. This section includes general safety for using all of the
machines in shop (lathe, mill, and router), battery safety and spill clean up, and
other general shop cleanliness items such as sweeping and knowing the place
for each tool.
The next section of the Training Document is the Machining and Mechanical
section. This section lets our students be prepared to use any tool in our shop.
The training starts with simple tools like hand drills and deburring tools all the
way up to being able to operate our big machines.
Other sections in our Document include the Electrical, Pneumatics, and
Programming sections. In these parts of the document, students learn how
decode status lights, connect pneumatic fittings, and even how the FRC field
works. In this section a student could also learn the basics of programming
starting with understanding PID and getting all the way up to motion profiles.
22
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
23/28
The last section of our document is the CAD and Design section. In this part of
the document students start off by learning the basics of CAD. They learn how to
navigate Solidworks and use all of the functions like sketch and extrude. After
learning to CAD, students move on to design. This year we had the pleasure of
viewing a design process presentation from Adam Heard where he took us
through all the different steps to designing a robot. This combined with the
different concepts in our Training Document helped our students be ready to
design a robot.
Practice Build Days
This year we have decided to take a much more aggressive approach to build season
by prototyping all possible mechanisms for the 2016 game during the first week of build
season. After we have built prototypes, we can discuss game strategy and make more
informed decisions on what our robot should do.
To practice this prototype heavy approach the legendary Adam Heard of team 973
joined us for three Sundays this fall to help us develop our prototyping abilities. We
thought of these days a practice build season work days. Working with Adam on these
projects was an incredible opportunity for our students and really helped us get ready
for an action packed build season full of prototyping and iteration.
On the first of three build season practice prototyping workshops, we had over 25
students in attendance and the day was an incredible success. We finished an arm and
an elevator prototype and both had running motors. This was probably one of our most
productive build days ever and was a great sign of things to come.
On the second build day we recreating Team 254’s 2014 robot. We used our router to
quickly create all of the necessary parts out of wood and worked and improving our
assembly speed. This day also helped us improve our understanding of how fly wheel
shooters work and created a good base for us when we tested them during build
season.
23
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
24/28
On the final of our three practice build days we created our own game similar to the
2006 FRC game to give our team a challenge that had not been completed yet. This
day was the ultimate preparation for build season. We had to CAD, machine, assemble,
and test a robot in one day. This was a great learning experience for build season and
how to work efficiently under time pressure.
Scouting
Improvements
Match scouting was designed this year to be simple and easy for scouters to record
data. In doing this we decided to ignore boulder data as it often can be difficult to see
who scored and if the balls went in. We also wanted to combine quantitative and
qualitative data. Last year we created an app and used tablets but because of the
issues that came up due to the inherent complexity of this simple we decided to test a
paper scouting system. Every team has one sheet that is reused for all of their matches.
A runner brings these sheets to the coach before each match to allow the drive team to
prepare. The most important part of pit match scouting is taking a picture of other
teams. This helps us remember what they look like and look closer at them during our
24
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
25/28
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
26/28
down several times during build season this year partially because they were not
maintained and set up as well as they could have been.
26
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
27/28
Cost Accounting Worksheet
(CAW)
27
Team 1836: Milken Knights
8/18/2019 Team 1836: Milken Knights Engineering Recourse 2016
28/28
Contact
Web
MilkenKnights.com
“Team 1836: The MilkenKnights”
@milkenknights
Milken Community Schools
15800 Zeldins’ Way
Los Angeles, California
90049
(310) 440-3500 x3436
Mentors
Al Noel Sansolis (Robotics Manager) [email protected]
Mark Mascadri (Robotics Coordinator) [email protected]
Roger Kassebaum (Director, MAST) [email protected]
Tanner Ragland (Director of Robotics) [email protected]
Team CaptainsMichael Bick [email protected]
Miranda Milner [email protected]
Austin Shalit [email protected]
Daniel Spar [email protected]
Team Members
Please contact all other team members through our contact form at
milkenknights.com/contact.