HP Memories Forum
[ Return to the Index ]
[ Previous | Next ]
HP 9100 Environmental Controller and Telemetry Systems
Posted by Wirt Atmar on 10 Sept 2000, 1:03 p.m.
Recently, HP ran a trivia game where you were asked to guess the function of various products shown. One of the products was the HP 9100, which inspired Wirt Atmar of AICS Research to write this article which is reproduced here with his permission.
The world's first programmable scientific desktop calculator, the HP 9100A
could add, subtract, multiply, divide, take square roots with 10-digit accuracy,
compute logarithms and compute the full range of trigonometric functions
in all quadrants in either degrees or radians. In a unit the size of a
typewriter, with built-in, read-only memory that stored calculating and display
routines, and the ability to perform floating-point calculations, the HP
9100A could run programs recorded on wallet-size magnetic cards.
The HP9100A/B Scientific Calculator
The world's first programmable scientific desktop calculator, the HP 9100A could add, subtract, multiply, divide, take square roots with 10-digit accuracy, compute logarithms and compute the full range of trigonometric functions in all quadrants in either degrees or radians. In a unit the size of a typewriter, with built-in, read-only memory that stored calculating and display routines, and the ability to perform floating-point calculations, the HP 9100A could run programs recorded on wallet-size magnetic cards.
AICS Research's Text
In 1968, we purchased a HP9100 Calculator as soon as one became available. And then we performed a lobotomy on it. The case of the HP9100 was cast aluminum, which we very carefully sawed off.
The HP9100 was built prior to the use of integrated circuits, thus every transistor in the HP9100 was a discrete device and the "ROM" was composed of four stacked boards that completely consumed the base of the calculator. The logic of the ROM was created through the use of a resistor/diode matrix (not unlike the display/keyboard computer that was on board the Apollo spacecraft that was flying to the Moon at the same time, although that computer was completely encased in poured plexiglass for mechanical rigidity).
We opened the calculator up in order to use it for a purpose for which it was never designed: an environmental chamber controller. To do this, various signals had to be intercepted from the circuit boards and converted into then-new TTL logic levels. And an 8-bit, sharable bus structure was designed to allow a serially sharable talker/many listener protocol to transmit the calculator's programmed commands to up to eight environmental control chambers simultaneously and listen to their responses.
Our HP representatives at the time were Norm Matlock, Ralph Kotowski, Jim Kemp, and Bill Little. Norm was excited enough about the novel use of the HP9100 that he convinced a group of HP engineers from the Colorado Springs Division to fly down and see what had been done. Eventually, three groups of HP engineers visited the lab at separate times and asked if they could copy the design, although it was nothing that they wouldn't have devised on their own. With some modification, HP later released the design as an internal standard and called it HP-IB, which was later certified to become IEEE-488.
The level of environmental control that we could obtain within the chambers using the HP9100 calculator was impressive. The graph above is the output of a mechanical hydrothermograph. This specific data was an early test run to demonstrate the chambers' capabilities over a two-week test run (only three days are visible in the photo above).
Insects are extremely elaborate behavorial machines, more elaborate than the most complex mechanical machine ever built by humans. Much of their behavior is driven by abiotic environmental stimuli (temperature, humidity, daylength, light levels, etc.). These chambers were designed to simulate those stimuli in an effort to tease apart the factors that promote the onset of such key behaviors such as ovipositional (egg-laying) rates.
In the graph above, the temperature was commanded to cycle sinusoidally precisely plus and minus 15 degrees Farenheit from the room's ambient 75 degrees while maintaining a relative humidity (atmospheric water saturation percentage) fixed at 85%. As you can see, control was nearly absolute. This level of precision control was accomplished by running small multi-stage refrigeration units (one per chamber) in opposition to air and water heaters. Although the technique was energy-expensive, the level of control necessary to the task demanded this design.
The sensors in each of the chambers were idiosyncratic, thus the HP9100 maintained a different calibration table for each temperature and humidity sensor in the chamber field. But it was the ease of generating elaborate polynomial equations using the HP9100's intrinsic scientific calculation capabilities for complex curve-fittings that made the HP9100 such an ideal controller.
Every calculator manufactured from HP following the 9100 incorporated a controller interface and became more and more elaborate in its capacities to control instruments. The later calculators incorporated a very fine BASIC language as their operating system. Nonetheless, for years following the HP9100, when HP advertised their new controller calculators, the example they always used in their advertisements was that of controlling environmental control chambers.
In 1970, we took the same HP9100 calculator, repackaged it, and significantly upgraded its responsibilities. Not only did it continue to control the eight environmental control chambers, it also became the centerpiece of a remote environmental telemetry system that stretched 200 miles from the Lincoln National Forest in the north to the Plant Science Center south of Las Cruces, New Mexico.
Remote telemetry stations were designed and constructed so that up to 16 sensors per station could transmit abiotic (weather) information back to the HP9100A calculator. The two principal research efforts at the time concerned phytophagous cotton insects and bark beetle infestations in the Lincoln Forest.
In both instances, sensors were placed at various heights in the plant canopies to accurately record their various microclimates throughout the year, the only difference being that the top of the cotton canopy was approximately four feet tall at its highest and the pine forest canopy was 70 to 100 feet tall. The white container in the right photograph was a surplus missile warhead container that we obtained from White Sands Missile Range. These lockable containers proved necessary to protect the telemetry stations from hunters.
The data transmitted back to the lab was recorded and passed on to the HP2116C computer for plotting and numerical integration. Insects respond to a wide variety of abiotic signals, but generally in a summed fashion such that total heat input or duration of daylight trigger fundamental physiological changes in them. Handling the insects was generally performed during their "night" period, in red light, which is invisible to them, in order to minimize resetting their natural clocks.
The data returned to the lab was not only recorded by the HP9100 calculator, but could also be sent simultaneously to one or more of the environmental chambers so that the external environment could be recreated within the chambers themselves. Every few seconds, the HP9100 readjusted the settings in the eight environmental control chambers. Interleaved with this activity, once every few minutes, the HP9100 calculator would query each of the remote telemetry stations.
Each remote telemetry station generated eight pages of plotter output per week, if all 16 sensors were active. The qualities recorded were most generally soil, air, bark and leaf temperatures, humidities, soil water tension, rain, wind speed and wind direction, although any physical quality could have been measured. Indeed, we were just beginning to measure electrical field strength and electron density, both of which were suspected of being sensible by our insects, when the project came to an end in 1976.
Although it's difficult to remember nowadays, standard communication speeds for teletypes were 110 baud. And 1K of memory was actually considered a fair amount of writable store. "High-speed" 300 baud modems were just beginning to arrive in the late 1960s, but you couldn't connect a non-Bell System modem to your telephone lines at the time.
However, in our case, the Bell modems that were available weren't usable. They drew far too much power and wouldn't fit in our size constraints. So we built our own modems from scratch, designing the UARTs (using TTL flip-flops and NAND gates) and tone decoders, and having them integrated into the remote telemetry station circuit boards, all of which were hand wire-wrapped by the wildlife and biology students.
In order to connect these "modems" to the telephone network we had to undergo a reasonably rigorous certification process by Mountain Bell to guarantee that the modems met all Bell standards. In spite of the pro forma procedures associated with this certification, the local Bell people were always very helpful. But then again, we were paying them a great deal of money each month.
[ Return to the Message Index ]
Go back to the main exhibit hall