Paul Vincent Craven
All Rights Reserved
The embedded computer for Sunrayce '95 was designed and built by Rick Steurer. It is based on the HC11, an 8-bit chip from Motorola well suited for embedded systems. While compilers exist for the HC11, we only had available an assembler for the processor. Assembler is a tedious language in which to program. Complicating the process, math calculations with 8-bit are difficult. The chip can only handle integer numbers raging from -128 to 127, or 0 to 255. Since the sensor readings require larger numbers, complicated work-arounds are required.
A second board was attached to the HC11 computer board via a digital I/O bus. On this board were two high-end analog to digital (A/D) converters. They take the voltages from the sensors and convert them into digital values that the computer can read. Each conversion is done to 12 bits of accuracy, which is an acceptable accuracy. This produces readings of roughly three significant digits. Anything over 16 bits would be too precise for our usage, as sixteen bits gives 4½ digits of accuracy, about the limit of most A/D converters.
During Sunrayce '95, we were unable to get sensor values from the solar car computer. Programs written on the HC11 seemed to execute correctly, but the values returned for the sensors were incorrect. Both Rob Ziegler and I looked long and hard at the assembler code, but we were unable to detect a software fault. At the time of this writing, no one is yet certain if the problem we had was hardware or software related. Similar errors had resulted earlier due to interference with the computer, so some Radio Frequency Interference (RFI) problems might be to blame.
Pre-built computer control systems are available costing a third of what we paid to develop our own (about $3,000). These systems have been tried in many applications, and take no student time to design. The downside is that they will not have the resolution of what we had hoped for in the system that we designed.
Those of us involved with the power management project decided to take a two-pronged approach. We would get a commercial computer system allowing immediate testing of power management theories. At the same time, a high-quality A/D system would be designed by students. Our risk would be the cost of the commercial computer system, versus not having a working student-designed system for an indefinite period of time. Once the financial support is available, the UMR team plans on getting a commercial system.