The Institutional Weather Machine
The Rochester Institute of Technology Department of Computer Engineering's capstone undergraduate engineering course, Projects in Computer Engineering II (0306-657-01), gives students the chance to build a realtime microcontroller system. Affectionately referred to as Realtime II, the Projects I and II courses typically expand into a three-quarter engineering effort during which time many students live in the Computer Engineering building at RIT. Not so with my group! Back in Fall 2001, I worked with Bob Faber and Shawn Jackson to create The Institutional Weather Machine. We were done on time and under budget, and the project was a fantastic success! The Weather Machine was operational almost a month before the course's deadline, and we performed the official demonstration over a week early. Careful thought and planning, as well as a thoroughly debugged component-based design, made the development process almost painless.
The Institutional Weather Machine was located on the fourth floor of the James Gleason building on the RIT campus. That is, it was on the roof. The Weather Machine sampled a variety of meteorological instruments to measure wind speed, wind direction, temperature, humidity, and lightning every 20 seconds. Depending on the environmental conditions and operating mode, the surface observations were transmitted via a cellular telephony Short Message Service (SMS) message periodically. Rapidly changing conditions, such as a gust front or electrical storm, would generate more frequent reports. The SMS messages were received by another telephone attached to a base station computer running Linux, our custom daemon, and a web server to distribute the current weather conditions.
One of the most surprising aspects of this project was the discovery of engineering scale. The Institutional Weather Machine barely fit in the Realtime Systems Lab for assembly and final pre-deployment shakedown testing. Everyone who visited the lab commented on how big the project was, noticing that it was a mere three inches shorter than the ceiling. From the top of the Gleason building, however, the weather machine seemed very small. From the ground, the weather machine was absolutely tiny! Despite 20+ foot tall enclosures on top of the James Gleason building, the wind data from the weather machine was quite consistent with values from the Greater Rochester International Airport. The temperature and humidity, as well as the derived dew point, were practically perfect!
Institutional Weather Machine Microprocessor and Meteorology Array
The Institutional Weather Machine utilized an anemometer and wind vane manufactured by NovaLynx Corporation. The anemometer contained three reed switches activated by a magnet attached to a rotating shaft. Pulses from the anemometer were intercepted by the microprocessor's timer interrupt and debounced using software. The windvane, the simplest sensor on the weather machine, operated as a simple variable resistor. Temperature and humidity data were collected using a sensor from Honeywell Sensing and Control. The humidity measurement from the device was presented as a linear voltage proportional to the relative humidity. The same chip contained a Resistance Temperature Detector (RTD), interfaced using a differential amplifier requiring gain, inversion, and DC offset stages. While the analog RTD data was transmitted to the base station, the weather station also contained a digital thermometer chip used to retrieve the temperature for reporting purposes. The weather station also featured a Stormwise lightning detector, interfaced to the microprocessor using 10 feet of fiber optic cable for isolation.
All of the weather machine's electronic components, including the microprocessor, interface circuitry, cellular telephone, AC power transformers, and battery backup supply were contained in a 3M outdoor cable enclosure. This enclosure was designed to be used as a junction box for telephone wires hanging between poles. Mounted vertically, the enclosure was the perfect size for the project's hardware. The microprocessor board's physical user interface consisted of a series of LEDs describing the system status and a lightning strike test button. More detailed control was available with an interactive console mode available to administrators by dialing the cellular phone in the station using a terminal program and a modem. Developers with physical access to the machine could connect a laptop with a terminal emulator to a second serial port to view textual debugging and status information.
The weather machine transmitted surface data to the base station using the cellular Short Message Service via a Kyocera 2035a cellular telephone. The telephone was powered by an AC supply independent of the microprocessor and interface electronics. Perhaps the most important aspect of The Institutional Weather Machine was its extremely reliable power supply. In the unlikely event of a power failure, the telephone could operate using its internal battery, and a liberal supply of batteries provided backup power for the microprocessor and interface circuitry.
Of course, somebody must always, somehow, be thanked. (Umberto Eco said that!) The Instititional Weather Machine was made possible with corporate support, including cellular telephony equipment, documentation, and service from Verizon Wireless. Additional assistance was provided by Peter Faber from the landline division of Verizon Corporation. Mr. Brian Romanko created an excellent graphic design and web page layout for the data dissemination site. Finally, my father Ken Woitaszek helped with the equipment infrastructure that housed a microprocessor outside for almost a month without blowing over in the Rochester wind.