I am indebted to Daniel Weed, whom I made the acquaintance of at http://www.hpmuseum.org’s online forum, for the following testimonial. Daniel worked at NASA and here is his recollection of how the HP-42S was instrumental in his work:
I worked at the Johnson Space Center as an engineer from about 1986 to 1995. I originally worked on the Space Shuttle program and much of that work after the Challenger disaster was to develop “Contingency Abort” options. These were scenarios involving 2 or 3 Space Shuttle Main Engine (SSME) failures during launch.
I used the 42s while designing a new shuttle ascent abort guidance algorithm. In those days, getting time on the mainframe simulation was slow. We had to be very sure of the changes made to the simulation. Then we could run it in batch mode, then analyze the results. I could only get time every few days, so it was not a good tool for iterative design work. What I needed was a tool to validate the equations of motion before going to the mainframe. This is where the HP-42S comes in. I found it a lot easier to do all the prototype work on the 42S. Only after I verified the equations of motion, did I code it into the main simulation.
The abort mode I designed was called “TAL Droop”. TAL is the acronym for “Trans-Atlantic Landing”, an abort mode where the Shuttle flies across the Atlantic and lands in Europe or North Africa. TAL Droop was a scenario where 2 SSMEs fail sometime soon after SRB sep (Solid Rocket Booster separation). The T/W (Thrust to Weight ratio) was initially so low, below one, that the vehicle would begin to loose altitude. Literally falling back to earth. After a while it would burn enough fuel that the thrust to weight ratio became greater than one and the Shuttle began rising again.
The minimum altitude it drooped to was critical. If too low, thermal stress on the ET (External Tank) was expected to cause the tank to explode. Marshall told us the critical altitude, if I recall correctly, was 270,000 feet. I had to predict what this minimum droop altitude would be well ahead of time and display it in the cockpit so the crew would know to either ride it out hoping for a TAL landing, or get off the tank before it was too late. This meant a ditch in the Atlantic. The algorithm also had to balance the need to stay above this altitude, by pitching the nose up, and the need to gain as much forward velocity as possible by thrusting more to the horizontal.
I started with a set of differential equations of motion under a single engine scenario and solved these for position and velocity so that the vehicle state could be integrated to the minimum droop altitude. I didn’t have computer tools to help with the integration, so I borrowed a slab of 17 inch fan-fold printer paper and started the task by hand. This is where the HP-42S came in. After integrating, I developed a simulation on the HP-42S containing these equations. I programmed it to print out the altitude over time on the thermal printer. With the 42S I was able to rapidly validate the equations and prove the overall algorithm. I could also assess the effect of different burn attitudes to find the optimal thrust vector to maximize forward velocity while staying above the droop altitude limit.
I used the thermal printer to create graphs of the shuttle altitude profile under different initial conditions. I’d end up with graphs about 6 inches long showing the altitude going up, then drooping down, then up again. It was a beautiful thing to see the first time it worked and this perfect altitude profile printed out.
I kept the program on the 42s for over a year, through at least one breathless battery change. Then I printed the program, cut it to fit a sheet of paper, taped it to the paper and photocopied it. I did that because the thermal printing fades over time. I probably have it in my records somewhere.
The TAL Droop algorithm was the largest and most important program I had built to support my work at NASA. I continued to use it for years of course. Later when I worked on a robotic lunar lander I used the 42S similarly to prototype trans-lunar injection burns, lunar orbit insertion burns, Hohman transfers, and lunar landing guidance. By that time though we had good simulations and tools hosted on desktop Unix workstations. This obviated much of the need to use the calculator for developing algorithms.
The calculator I used was purchased from EduCALC in February 1989 for $89.95. The same month I also purchased the HP leather case ($17.95), and the thermal printer (105.95).
In November of 1989 I discovered that when I removed the batteries the calculator immediately lost all program and memory content. That was a bummer. HP replaced the unit. The serial number of the original was 2840A-xxxx, and of the replaced unit is 2919S-xxxxx.
Written by Daniel Weed.