Date: Sun, 22 Jan 2006 23:53:23 -0600 (CST)
Subject: more mechanical assembly
X-UID: 126
Content-Type: IMAGE/JPEG; name="img2563.jpg"
Content-Type: IMAGE/JPEG; name="img2567.jpg"

The robot seems complicated even though it is comprised of relatively few 
parts. Like the "Borg Appliance" (small form factor computer), the design 
is somewhat tight. Everything fits together without much extra clearance. 
Disassembly might require removal of other assemblies to gain access.

Weight is about what I originally imagined, somewhere around 30 to 40 
pounds. Without the lifting bar, it would be almost impossible to lift the 
robot by hand. Note to self - all future robots must have handles and 
hardpoints for lifting.

This is pretty close to what I wanted a year ago when starting to design 
the robot.

The two large battery packs at the rear are for the drill motors. The rest 
of the batteries go under the electronics box in front of the drill 
motors. There is just enough space for them there.

The electronics box, WiFi router, and 900 MHz radio in the Penguin 
Peppermints tin is mounted with "industrial strength" velcro. For extra 
security, additional velcro straps will be used.

One disturbing phenomenon I have not seen before is seizing in the 
steering motor. It is mounted next to the batteries under the electronics 
box. The motor would not turn or turn part of a rotation and then become 
stuck. I think that lateral pressure on the gear motor caused it seize up. 
Turning the rotor by hand and carefully mounting it seemed to avoid this 
problem. The earlier design of six months ago used a flexible shaft 
between the steering bolt at the front and the motor. This shaft isolated 
the planetary gearbox of the motor from lateral forces. Unfortunately, 
given how the design evolved, the shaft also made the steering motor 
assembly too long to fit. So the shaft was removed. A potential side 
effect is inferior reliability in the current design.

I read that MTBF (mean time between failure) estimates for US Space 
Shuttle indicate catastrophic mission failure 1% of the time. This was 
always true. Every mission had a 1% chance of losing the orbiter with all 
hands. There were so many systems that had to work. Any one of them 
failing would be catastrophic.

You might think, how could you design a system that would fail so often? 
Unfortunately, I think that human beings do this a lot. Either we don't 
care or we just don't know how to do better. Everything is designed to a 
level of quality and acceptable failure rate whether openly admitted or 
not.

I didn't really understand how complicated this robot is until going 
through the stages of final assembly. Individually, each subsystem is 
simple. But together in the robot they are heavily interdependent. There 
is so much to go wrong.