ExSpurt
I have avoided
building a mini sumo for several years while I concentrated on the 1Kg
and 3Kg classes. Once
Executioner was finished, I needed a new challenge. About
that time,
CIRC was dropping the 1Kg class to make
more time to get into line following contests. I really didn’t want to
drive 200 miles for one sumo class and a very old line follower that
played the same old song, so I decided to build a mini sumo specifically
for the
CIRC event.
The
Maxon motors on the
BaneBots web site appeared to be a good place to start a mini
sumo. Because of the length of the Maxons, designing a legal mini sumo
was going to take a new wheel/tire/hub solution. I thought it was
achievable, so construction was started on a test chassis. In the past,
I have been unable to properly judge the performance of a specific
motor, so the first mini sumo was built to be a test of construction
techniques and performance. That test chassis is the ExSpurt "Mule". The
photos here show the construction of the real ExSpurt which has yet to be run in a contest
because ExSpurt Mule has been so successful.
The
past several robots that I have construction have used a simple machined
HDPE (High Density Polyethylene) jig to hold the motors/gearboxes in place while epoxying the
chassis components together. For the mini sumo the jig was 2.5” wide,
5” long, and 1” thick. Two slots were milled to hold the Maxon motors
tightly aligned. A wheel base of 2” would provide sufficient room for a
small Lipo battery and a simple bumper. Other slots provided clearance
for electrical tabs and motor wires.
Since
access was going to be limited, the motors were wired first. For good
adhesion, the paint was scrapped off of the Maxon motors. In the past
the chassis was epoxied to the motors, but not the gear boxes,
permitting some disassembly. The same technique was used here. Also,
the plastic rear covers over the Maxon brushes were removed and the
motors were pushed tight together to save on width.
A
simple chassis was milled out of .032” carbon fiber material. Yes, it
was carbon fiber machined on a mill. High speed small diameter end mills
work much better that the old method of a cut-off wheel in a Moto-Tool.
The overall dimensions of the chassis are 3.15” long and 1.5” wide. The
chassis was cut down to .865” wide in the middle to allow more chassis
flex. A rectangular hole was provided to allow sensor wires to get to the circuit
board while still providing a very solid bumper support.
After
sanding the mounting surfaces of the chassis and cleaning all surfaces
with alcohol, the chassis was epoxied to the motors. I use Devcon
Plastic Welder No. 14300. I have never had a joint break when
using the Plastic Welder. Other
structural epoxies that I have tried are very sensitive to mixing
ratios, but not Plastic Welder. It is fool proof.
After
24 hour cure, the chassis and motors were removed from the jig and .25”
x 1.7” .032” carbon fiber pieces were added to support the bottom of the
motors. Also small fillets were built up around the motors and all
voids in the epoxy were filled.
Here
is the completed chassis, ready for bumper, batteries, control board,
and wheels and tires. It was surprising how fast the chassis went
together. After the long process of constructing Executioner, ExSpurt
was a refreshing change. The major difference between ExSpurt Mule and
the real ExSpurt is the chassis location. ExSpurt Mule has the chassis
under the motors while the real ExSpurt has the chassis above the
motors. The change was made for two reasons: First the chassis on top
gives more room to mount lead weights (and, possibly, an optical
mouse). Second, the top chassis makes for a much stronger bumper mount.
There
has to be one photo of the chassis with wheels and tires. It sure does
look silly at this stage.
ExSpurt
Mule’s bumper was machined out of .125” extruded aluminum. The problem
with the thin bumper was that two of the IR LEDs stuck through the
bumper and were exposed to contact. So the real ExSpurt’s bumper
started out as 0.500” 2024 aluminum. Slots were milled and holes were
drilled to mount the sensors. See other BrooksBots for details. Five
IR LEDs were mounted in a 15° pattern, giving ExSpurt the typical 75°
view. Only four sensors were used to save space. The two outer sensors
were angled out at the normal 30°, while the two inner sensors were
angled at 7°.
A
horizontal slot was milled in the back of the bumper to allow room for the sensor wires and provide a good
place to attach the chassis. Finally, the thickness of the aluminum was
reduced to .350” and the bottom was angled 14° so that the sensors
pointed up slightly. The 14° angle prevents the sensors from
seeing the surface, yet will detect a pencil laying on the surface. The bumper was finished.
The
motor control is similar to earlier sumo BrooksBots. The only change is
that everything is packaged a little tighter. The
Basic Stamp 2SX (A) runs the sensors
and uses serial out to the
Pololu PIC (B). A TI
SN754410NE (C) is used
to drive the
IRF3205ZL and
IRF4905L MOSFET bridges (D). A
rolling code receiver (E) from Oatley
is used for the kill switch. A
Datel 78SR, OOPS! it is now
C&D Technologies, switching voltage regulator
(F) provides 5 Volts and a
LM317L (G) provides adjustable voltage to the
IR LED sensors. There is still a reset push button (H) and an input
push button (I) along with the four wire programming plug (J) as has
been standard in all of my robots for many years. A red/green LED (K)
provides output information, a small green LED (L) monitors the remote
and a 5K ohm pot (M) provides sensor adjustment. There is also an
additional four pin connector (N) for the LCD touch screen interface.
The
initial running tests on the mule were performed without bumper or
control board. The motors were wired directly a three cell battery
pack of
MaxAmps 430 ma-hr Lipos. When
place on the basement floor, the little robot spurted across the room,
ricocheted off of three walls and buried itself in a corner, wheels
spinning. After that experience, the name ExSpurt stuck. The next test
was with the control board (still no bumper) programmed for a one second
full speed run. ExSpurt didn’t make it far. It just flipped over on
its back and spun its wheels upside down. The weight of a bumper was required to prevent the
180 degree wheelie.Finally the bumper was attached. The total weight of ExSpurt Mule was 325 grams…way too light. The one second speed run resulted in much wheel spin on acceleration and wheel lock up and sliding while braking. The total distance traveled was close to 100 inches. It must be dirty tires and dirty floor causing the wheel spin. So, the floor was mopped and the tires were cleaned in alcohol but the wheel spinning and lockup continued. We are just going to have to live with it. Make a note to reduce motor speed with the controller and never, never run full speed. A run at half speed from the control produced 47 inches per second.
The
program for ExSpurt was just the 2007 sumo program from Executioner. In
the drive subroutines, the motor speed was divided by two to give half
speed, the sensors were reset to the optimum frequencies for ExSpurt's
sensors, and the mouse, LCD touch screen, and spinning options were
commented out. It was surprising how close the two robots performed on
exactly the same run settings. This state of development is the way
ExSpurt Mule ran at the 2007 NEIRG…325 grams total weight. One competitor looked
like it could be a problem, but ExSpurt won easily.Videos of the 2007 NEIRG competition:
1A ExSpurt vs Top Pusher (0:06 - 0.6Mb) 1B ExSpurt vs Top Pusher (0:06 - 0.7Mb)
3A ExSpurt vs Time Bomb (0:05 - 0.6Mb) 3B ExSpurt vs Time Bomb (0:05 - 0.6Mb)
7A ExSpurt vs Razor (0:03 - 0.4Mb) 7B ExSpurt vs Razor (0:05 - 0.5Mb)
10A ExSpurt vs Razor (0:04 - 0.5Mb) 10B ExSpurt vs Razor (0:03 - 0.4Mb)
After
the NEIRG, a 100 gram weight was formed and glued in ExSpurt Mule
bringing the total weight up to 430 grams. Push tests with various
composition tires resulted in a peak of 1080 grams of shove. ExSpurt’s
push and speed seem to be a good combination.
The
next event was the 2007 CIRC. ExSpurt lost in an early pushing match to Wild Tire (is that double stick masking tape on the tires?) and had to
work its way up the looser bracket. A second match against Wild Tire,
carefully avoiding a head on confrontation and four wins against Orthos
produced another contest victory.Videos of the 2007 Central Illinois Bot Brawl competition:
1A ExSpurt vs Land Mine (0:05 - 0.5Mb) 1B ExSpurt vs Land Mine (0:05 - 0.5Mb)
7A Wild Tire vs ExSpurt (0:48 - 4.5Mb) 7B Wild Tire vs ExSpurt (0:31 - 2.9Mb)
9A ExSpurt vs ToxicAvenger(0:05 - 0.5Mb) 9B ExSpurt vs ToxicAvenger(0:05 - 0.5Mb)
11A ExSpurt vs RedOctober (0:06 - 0.7Mb) 11B ExSpurt vs RedOctober (0:07 - 0.8Mb)
13A ExSpurt vs Wild Tire (0:07 - 0.7Mb) 13B ExSpurt vs Wild Tire (0:06 - 0.6Mb)
14A ExSpurt vs Orthos (0:06 - 0.7Mb) 14B ExSpurt vs Orthos (0:06 - 0.6Mb)
15A ExSpurt vs Orthos (0:06 - 0.6Mb) 15B ExSpurt vs Orthos (0:06 - 0.7Mb)