Extrasensory II
Extrasensory was
completely rebuilt with a chassis configuration that allowed the optical
mouse to be mounted closer and more parallel to the sumo ring surface.
The finished robot was under 2 inches tall and 20 grams under the 1 Kg
limit.
Specifics of the rebuild are presented in excruciating detail below.
Since there was no other entries in the Lightweight Autonomous Sumo class, Extrasensory competed in the Heavyweight Autonomous Sumo class 2005 NEIRG. Although it did not come close to winning, Extrasensory did well against some sumos that were three times it weight.
Videos of the 2005 NEIRG Competition:
4A Extrasensory vs Scrappy (0:07 - 0.2Mb) 4B Extrasensory vs Scrappy (0:20 - 0.4Mb)
4C Extrasensory vs Scrappy (0:23 - 0.5Mb) 4D Extrasensory vs Scrappy (0:04 - 0.1Mb)
6B Extrasensory vs Sticky (0:31 - 0.6Mb) 6C Extrasensory vs Sticky (0:04 - 0.1Mb)
6E Extrasensory vs Sticky (0:03 - 0.1Mb) 10A Extrasensory vs Scrappy (0:07 - 0.2Mb)
10B Extrasensory vs Scrappy (0:13 - .3Mb) 10C Extrasensory vs Scrappy (0:07 - .2Mb)
The event showed that Extrasensory needed much more work on the software, but might eventually be competitive in the 3Kg class.
The
rebuild started with new sensors in the front bumper. This is the
back side of the aluminum bumper with machined mountings for the
electronics. Having machined recesses for the IR LEDs and PNA4602
sensors assures proper alignment. Due to the low mounting, the
sensors are all angled up 20 degrees from horizontal and are angled out 15 degrees and
30 degrees. An experienced machinist might not be awed by quality
of the piece, but two grown males who think that they know everything,
using a milling machine for the first time, was really quite amusing.
Luckily, the first piece was usable.
The
five Panasonic PNA4602 sensors were glued into their recesses with black RTV.
The small diodes were already attached. I've since found out that
the diodes are not needed. The open collector of the PNA4602
allows all of the outputs to be tied together without extra components.
The
+5 Volt and ground leads were wired to the sensors. A small
capacitor was added for a little power filtering.
A
little heat shrink tubing was added to isolate the diodes and the sensor
leads were wired together. Refer to the page on
Excuse
II for a schematic of the circuit.
A
10.0 K ohm 0805 size resistor was soldered between the gate and source
leads of each IRFD024 FET. A wire was attached to the gate lead
and the FETs were glued onto the bumper. The QEC113 IR LEDs were
glued in and care was taken to properly align their leads.
The
IR LED wiring was added along with the power and ground wires for the
FETs. The blue 10 mf tantalum can be seen, but the 33 ohm resistor
is covered with heat shrink. Note that the power and ground for
the FETs is completely separate from the PANA4602 power and ground.
The wiring is now completed.
The
wires were neatly arranged and the sensor array was tested. It is
difficult to see in the photo, but small pieces of black plastic
electrical tape were added to the rear surface of the PNA4602 sensors.
IR radiation can enter the rear of the sensors and give false readings.
At this stage there is no optical crosstalk, but normally a little electrical crosstalk between
the test wires (about 20 inches long). Once the bumper is mounted in
the robot, the wires are under 6 inches long and crosstalk is not a
problem.
The
electrical components are all sealed in black RTV. The RTV
not only physically protects the wiring, it also provides electrical
insulation and eliminates IR crosstalk.
A
new circuit board was produced to fit the tight packaging of
Extrasensory. This photo shows the layout of the FETs, IC sockets,
voltage regulator, and push buttons. The holes in the perforated
board had to be enlarged to fit the FETs and the sockets. The
board was cut to final size and the edges were finished with a file.
The top socket is for the PAK IV, the middle socket is for the BS2SX,
and the bottom socket is for the two Pololu Dual Serial Motor Controller
ICs. The IC sockets are held in with super glue.
The
FETs used here are the IRF4905L and the IRF3205ZL. In this
application, the FETs can handle about 10 amps continuous and the motors
are rated at less than 5 amps stall current. The TO262 package
allowed a little lower overall height for the finished robot.
The
SR106 diodes are 1 amp rated Schottky type. Schottky diodes are
preferable in this application due to their low forward voltage drop and
fast turn response time.
The
diodes, zeners, and resistors are added that allow battery voltages greater than
20 volts, but keep the FET gate to source voltage less that 15 volts.
Any gate to source voltage close to 20 volts will destroy the FETs.
I learned that the hard way.
Here
are two side views showing the diode, zener, and resistor mounts.
It gets a bit tight, so additional photos are helpful. See the
page on Excuse II for the schematic of this circuit.
The
H bridges were tested with spare motors. The orange wires are
attached to the gate circuits. Short any orange wire to positive
battery voltage and that motor will run. I like to check each
subassembly of the circuit because wiring errors get harder to find and
correct later in the build process.
Next,
the Datel 7805SR voltage regulator was added along with the filter
capacitors. Again the circuit
was tested for proper operation.
The
Pololu Dual Serial Motor Controller is wired next. 10K pull down resistors
were added to the SN754410 inputs as well as to the PIC12C508JN reset
line. If the PIC is held in reset for any length of
time, the SN754410 input voltages tend to float up, turning on both
sides of the FET H bridges. This has lead to release of
the magic FET smoke in the past. Again, see
Excuse II for the
schematic.
The
BS2SX, PAK IV, and the adjustable voltage regulator (for the IR LED
drive current) were wired. No, no one took a bite out of the
board. When mounting of the remote kill switch
receiver was attempted, it was discovered that two components required
cut-outs in the perforated board. Again, the operation of all of the circuits was tested.
The
final wiring...the kill switch receiver. See the page on
Flaming P'nut for the specifics. Slowly, we start to
see the rat's nest emerge from the perforated board.
The
chassis was assembled and the mouse, bumper, and motor wires were
connected to the circuit board. In this photo, you can see the
front bumper and optical mouse attached to one piece of
carbon fiber graphite. The front motors are epoxied to small
pieces of carbon fiber graphite and pivot on two bolts. The rear motors are actually mounted on a
separate chassis. The motors are
B62 gear motors from
Robot Marketplace. The 1.75" OD
tires are molded urethane similar to those used on Sticky and Excuse. 
The
wiring is folded over to start the final assembly. In this photo,
the rear motors and batteries can be seen mounted to their own
carbon fiber graphite chassis piece. The original piece of graphite was a little
too flexible, so reinforcement pieces were epoxied on to add stiffness.
The batteries are two
720mAh 11.1V 3-Cell Lithium-Polymer Packs
from Great Planes.
The
wiring was arranged to allow access to the mounting bolts. The
front mounting bolt can be seen between the mouse and the front motors.
The yellow wires run on one side of the bolt and the blue and green
wires run on the other side. The rear bolt isn't yet installed.
The hole for the bolt can be seen just behind the mouse. Also note
the two mounting grommets on the rear carbon fiber graphite chassis.
The
rear chassis is now placed on the mounting bolts. The rear bolt is
still not installed. I can't remember why it wasn't. Since carbon
fiber graphite is conductive, clear packing tape was placed on the top
of the rear chassis (not yet installed in this photo). We don't
want any shorts under the circuit board. That would not be fun.
Finally,
the assembly is complete. A rubber band was added to hold the
circuit board in place. Extrasensory is now ready for the next round of
electronic and mechanical testing. For some reason, the photo was
taken with the robot pointed to the right instead of the left.
Could it be a subconscious attempt to confuse the reader?
A
carbon fiber graphite front plate was added to the front bumper to
absorb and reflect opponents sonar and IR. It is held onto the
machined aluminum front bumper with double stick carpet tape.
The plate extended down to almost cover the sensors. A top plate
was added to deflect an opponent's sonar and protect Extrasensory's
electronics. As can be seen from the videos, Extrasensory spends a
lot of time under other robots. The top plate is held on with a tab that
fits in a slot in the front bumper, and industrial strength "hook and
eye" material at the rear. The top plate must be tilted up at the rear to
reach the pushbuttons for the start, and then pushed into place for the
match. More clear packing tape was added to the top to
electrically insulate the circuit board from the conductive carbon
fiber. I also need a note to remind me what the start routines
are.The attempts at stealth were not very successful. Typical sonar picks up Extrasensory easily. The next sumo should have the front plate at less than a 45 degree angle. Perhaps a 30 degree front plate will work better.