Journal00 mar
From Simreal
Contents |
March 2000
Fri, 17 March 2000
Well!
I spent a feverish week getting Boris fully functional for a Career Day presentation at a local grade school last week.
Man.
This was the first time out with the new Brain Boards. I've surpressed much of the memory of this setup; too painful.  I hooked up the brains to the valves and ran some initial tests. The communications system (UART over RS485 layer) ran beautifully. Perfect. The valves ran pretty good too. But there were eight valve circuits (out of 36) that either didn't work, or ran weakly.
After some poking around, I determined that my lack of clamping diodes on the MOSFETs was allowing the inductive valve load to blow the poor chips. After watching some active valves on the scope, I felt that must be so -- these guys were spiking one heck of an inductive charge.
So, the next day (after work, mind you, each of these days) I de-soldered the handful of MOSFETs and replaced them. I also added clamping diodes to the valves, and watched a beautiful power signal trace its way across my scope. That was plenty for one evening, so I went home full of optimism and anxious to test the system later.
Just a couple of days before the presentation. I go back and re-test each valve system and, to my dismay, each and every one that was bad, and replaced, was still bad!
After contemplating, and dismissing, thoughts of honorable suicide, I attacked the problem with a vengence. I chose one particular MOSFET system to pick on. After pulling the board, I tested continuity with the vohm... yup, there was a short through the MOSFET. So I desoldered the MOSFET. What? Still shorted!
A careful visual inspection with my 10x jeweler's loupe (a vital tool for this stuff) indicated no problems anywhere on the board, front or back, that could account for the short. No. It couldn't be. The header block?
I desoldered and pulled the header block. No short. Somehow, this particular style of header block was causing a short to the ground plane of the board! Death to these header blocks! You see, unlike other headers I have lying about, these blocks expose a bit too much metal on their delicate underbodies. ARGH!
Now, after the fun of resetting all offending blocks, I'm off to test again. And things work! After a few more iterations, a few more blocks vibrate into a shorting position, but I fix that by stuffing thin plastic bag material under them.
Whew!
Career Day itself was a blast -- four groups of munchkins gazing at Boris and another robot, Catherine the Creative (work of Erik Lundquist) who does a passable job of drawing pictures like a five-year-old. I guess they were talking about the "big spider robot" for a week after the event! We certainly created some new engineers that day.
The big show was South by Southwest, which is really three festivals in one. We (The Robot Group) were part of the Film / Interactive Trade Show in a donated booth. During this show, Boris ran for three days, four to six hours a day, and performed beautifully. There were some problems, of course, many of them I anticipated but some I had not.
Check out the SxSW photos and commentary on the Project: Boris Photographs page. There are even some MPG movie clips, so you can see Boris in action! We even got some coverage from CNN, though I missed that moment in the sun Derek Bridges did a great job of representing the group.
