Drive System

    The motor drive system is pretty simple.  The left and right sides are identical (although a mirror image on the mounting bracket).  Each consists of a Hsaing Neng  Geared Motor  P/N HN GH12 16341 (available from www.jameco.com as their part number 161381.  It has a no-load speed of 200 rpm and a max torque of 3.6 Kg-cm.  Plenty of power for Trinity. 

    The wheel hub assembly was made from scratch from a piece of cylindrical plexiglass as a hub attached to a circular piece of Sintra for a wheel.  A groove was put in the outer edge of the sintra to hold an O-ring as a tire.  Holes were drilled through the hub to accept the motor shaft (6 mm) and the encoder shaft (0.25 inch).  Holes were drilled in the side of the hub and tapped to accept setscrews to hold the motor and encoder (P/N 600CS-ND in www.digikey.com catalog) in place.

    The motors were driven by LMD 18201 H bridges which I mounted to a small PC board.  (if anyone wants the  PCboard layout, I can send them the file for ExpressPCB).

    The motors are rated at 12 vdc, and when I "upgraded" to the two NiCad batteries (which supplied about 16 vdc when fully charged), I stripped out a plastic gear in each motor.  They had previously worked reliably for months at 12 vdc.  Subsequently (meaning 2 new motors later), I restricted the PWM to a maximum of 75% duty cycle and didn't have any more such problems.  However, as a warning, John Edwards used the same motors in his robot and stripped out the gears on several at 12 vdc. 

    I used these motors for 2001 and 2002.  By 2003, I had found a source of Pittman gear motors with encoders.   The particular motors I got were rated at 30 vdc (but had LOTS of torque for a firefighting robot) when powered by 14.4 volts (2 RC race car batteries).   They had a 5.9 gear ratio.  With 3 inch diameter wheel geared to the motor at a 1 to 1 ratio, the robot could run at up to 50 inches per second (ips).   I restricted the top speed to 40 ips to leave a little power for control to maintain direction more accurately.

(click on picture to enlarge)

I didn't want to build a robot any larger than 10 inches in diameter and these motors are pretty big.  Hence, I came up with the mounting above.  The two sides were identical.  The motors drive an RC race car gear to power the wheel.  Using two identical gears for a 1 to 1 ratio.  The wheels are from in-line skates with the shape sanded flat on the bottom to allow it to run over ribbed rugs a little better.  It came out kind of nice in that there was just room between the two motors to mount two 7.2 vdc batteries.

Note that I started out using an omni-wheel for the casters.  I found that they had quite a bit of drag when trying to make the robot pivot on rugs causing the robot to turn less than it was commanded to. Turning was based on differential encoder clicks between the two wheels.  I ended up using teflon skids for the front and rear casters which seemed to slide more easily.

These teflon casters were not very optimal for the contest.   The "floor objects" used were painted with a black paint loaded with a course sand for traction.  The casters did not slide easily over the objects (as it had for my smoothly painted floor object at home.).  In the end, I had to reduce the top speed of the robot from 40 ips down to 20 to be able to go over it in a relatively smooth motion.  (as it was the floor object had to be taped in place, or the robot would just push it along).   For next year, I want to have a rolling front wheel which will easily climb up on such floor objects.

One consideration is that the encoders produced 512 clicks per motor revolution.  After the gear train, this resulted in about 3000 pulses per wheel revolution.  At 40 inches per second, each encoder issued about 13000 pulses per second.  This was a substantial amount of interrupt processing and took up to 10% of the available throughput on my 24 Mhz processor.