Date: Thu, 15 Sep 2005 05:58:06 -0500 (CDT) Subject: computer setback X-UID: 88 Content-Type: IMAGE/JPEG; NAME="img1836.jpg" Content-Type: IMAGE/JPEG; NAME="img1837.jpg" One of the boards will not power up. The LEDs flash and then it's dead. If I wait for some capacitors to discharge, then I can repeat the phenomenon. The only difference is that it has a 5 port USB card plugged into it. This card is a universal PCI card with 3.3 and 5 volt signaling. I read however that some cards short the 3.3 and 5 volt lines together. This is my best guess at explaining the problem. You can see it on the DSO. In img1837, the blue trace rises to 5 volts. The yellow rises to 3.3 volts. This comes from the +5 and +3.3 GPIO pins on the mainboard. This is what is expected. In img1836, both traces overlay one another. The traces rise up to about 4 volts and then drop to nothing. Note the horizontal scale is different. Each division is 2.5 ms. So entire screen on the scope is about 30 ms. In contrast, the other trace is for 600 ms. My hope is that the voltage regulator (power supply) on the lower board isn't messed up. I might still be able to salvage the board even if it is by supplying my own power with an external circuit. Given that the voltages involved are not high and the duration of the transient on the order of a few milliseconds before something forces the board to power down, I'm hoping there's no damage. But electricity is very quick to fry electronics too. It just takes a moment of the wrong polarity, etc for something to fail. And it may not fail catastrophically but just sap the life out of parts which then slowly die. I'm not sure that I could have foreseen this without more experience. While the PC may have spawned a huge industry supplying largely compatible and interchangeable components, that really applies to PC's only. The mainboards I'm using are an embedded product. They're not PC's. So PC stuff works with other PC stuff. If you have something else, then compatibility is uncertain. The direct lesson is that universal PCI is a misnomer. I do have several fallback positions. I could use the two USB 1.1 ports on the mainboards. I never actually needed to have the extra 5 ports the PCI card offers. It was just a nice idea for expansion. It also allowed placing both cameras under direct control of one board. If I use the mainboard ports, then the cameras must be split between the boards. This is discouraging. And I'm really dead tired right now. I've made good progress on the software side. I can boot into the upper board and login. Instead of doing a very long sequence of difficult to remember commands, I'm starting to set up a build environment to do this for me. I can see how it would be possible to have a robot distribution of software with something like the Linux style menuconfig. This would be easy enough to use that it wouldn't sit isolated as a research project but be practical. I'm feeling better about the DSO - it is a useful tool. Its user interface design is idiosyncratic. Increasing a value means turning the know counterclockwise. Decreasing a value is done with a clockwise turn. This is the opposite of my intuition. However, I believe this may be Asian meets American culture. When I first started driving my Toyota, I found all of the switches were the exact opposite of the GM Saturn. This scope is a little like that too. Anyway, I think this is why 100 hour weeks are required. That admonition about working smarter, not harder, only applies when you are smart. Most people are incapable of working smarter. So they must work harder. I'll have to take the entire electronics box apart (that's a full day/night's work), remove the USB card, and try booting into the lower board. If this works, I'll leave it powered on for a few days to kind of validate that it's still reliable. Then I'll just have to put everything back together without the USB card, unless I can modify it somehow. I don't know how the electrical interfacing works on a PCI card so don't know if it is an easy fix or impossible. This is taking too long.