Adventures with a "Lunchbox" Computer

Marc de Piolenc , Philippines

I was new to microcomputers. My first, a Polymorphic 8813, came with little packaged software, occupied an instrument cabinet and held a paltry 90kB per diskette. I bought a Kaypro 4 more or less on impulse - the idea of actually being able to take a computer with me was appealing. And then, by sheer luck, it turned out that something called a "spreadsheet" could help me with a short-fuzed engineering problem.

I have a mix of interests, nearly all in exotic areas of engineering and technology. Lighter than air flight (which this story involves), magnetic amplifiers, free-piston engines, tailless airplanes and so forth. I'm a packrat when it comes to technical reports, and actually make part of my living selling copies from my 6,000-item-plus technical library. I take a special interest in "shelved" technology - most such work is ahead of its time and needs to be revived at some point.

This is "old hat" now, but in 1986 it was a big deal, and I was proud of my initiative and its results.

I had been hired to work on an airship project based in Eugene, Oregon. The airship itself, a nonrigid airship or "blimp" with an air volume of about 140,000 cubic feet, had been erected and inflated in an old Navy blimp hangar at Tillamook, Oregon and was awaiting Federal Aviation Administration (FAA) inspection prior to its first flight.

There were, however, a few problems. The flight controls were hydraulically boosted - an unusual feature in lighter than air - and the system was "chattering." A controls expert was on the job, and a simple gain adjustment would eventually eliminate this problem. More serious was the fact that the aircraft structural assembly drawings were missing, so the project leader, who had been hired after the design of the ship was complete, had to route the bridles from the internal suspension system of the ship to the four attachment points on the "car" or gondola according to his best judgment. Also missing was the structural analysis of the internal suspension system as built; the available calculations were for a ten-lobed catenary, while that actually built had nine lobes. The net effect was that the control car had a noticeable down-angle in front, and its front edge was not flush with the "envelope" or gas-bag, as intended. Besides looking odd, this changed the thrust-line of the propeller, which was mounted on the rear of the car. It also reduced the clearance between the propeller and the envelope and might cause control problems in flight.

I was one of the few engineers with LTA knowledge - "helium heads" in LTA parlance - within easy reach, and my assignment, which I pondered as I drove hurriedly north from my home in San Diego, was to analyze the internal suspension structure and its attachment to the car, as built, and either prove that it was sound or provide data for modifications that would make it so. I was also to figure out how to get the car flush with the envelope.

Neither of these tasks was particularly difficult or challenging. Structural analysis of a nonrigid airship envelope is theoretically complicated because the structure is highly redundant and not rigid, but there was enough experience, accumulated over more than a century of nonrigid airship construction, to provide the necessary empirical corrections and reduce the problem to a simple statics calculation. The problem was time - or rather, lack of it. An appointment had already been made for the final preflight inspection of the ship and its engineering documentation. We had thirty-six hours to make the calculations and any necessary modifications.

The second task was simpler in concept, but possibly much more difficult in execution. The envelope's internal catenary curtains were connected to the car by three pairs of bridles, but the car only had two pairs of shackles to receive them. What had to be decided, absent the assembly drawings, was whether the middle pair of bridles was intended to be connected fore or aft on the car. The project leader, reasoning that the biggest weight item in the car - the engine - was mounted aft, had attached the middle set of bridles to the aft shackles on the car. This guess was obviously wrong, because the car now had a definite nose-down pitch - too much lift aft, not enough forward. Unfortunately, re-routing the bridles would take much more than the time available, so if that turned out to be inescapable the company would have to postpone the FAA visit. This could result in weeks of delay, since the Feds had other commitments and would not necessarily be able to come back as soon as we were ready for them. Not to mention embarrassment and loss of credibility.

So, two tasks:

1. Prove that the arrangement as built was safe, if not necessarily optimal;

2. Give the riggers information that would allow them to improve the car's incidence with respect to the envelope;

- in 36 hours or less.

Task 1 could be done with the tools available - a Texas Instruments programmable calculator - in the time allotted. It was a "direct" problem, not requiring iteration. All that was needed was to prove that the peak force in the most highly loaded suspension cable was low enough to leave an acceptable margin of safety.

Task 2, however, required me to find a set of cable tensions that would allow the car to sit level. That task would require repetitive calculations to find a solution that both satisfied static equilibrium of the structure as a whole and allowed the car to sit level. There was simply no way to accomplish this with the TI before the FAA visit.

With me in the borrowed minivan was a new purchase - a Kaypro 4 computer running the CP/M operating system and equipped with two floppy disk drives, no hard drive and its standard monochrome monitor - green letters on a black background. With it were disks containing a suite of office software, which I hadn't had much time to look over since the computer had been delivered only the day before the 'phone call from Eugene.

The blue-grey sheetmetal box with its sturdy handle did prove to contain the solution to my problem, however. That evening in my motel room, I started looking through the software manuals. There was a BASIC interpreter which looked pretty interesting - the language was modified to make it structured, like FORTRAN or Pascal. There was a database manager, a word processor, and...what the hell is a "spreadsheet?" As I went through the Perfect Calc manual an idea started to form, and it solidified when I learned that Perfect Calc had trigonometric functions built-in. This was the interactive tool I needed to do the job in time.

I spent much of that night experimenting with the software and building small spreadsheets to prove to myself that it worked. The following morning at the office, we pulled the envelope drawings and called Tillamook to get current tension readings. I quickly built the spreadsheet based on the cable angles in the drawing, adding the cable tensions as they came in. By early afternoon I had a static calculation that balanced and agreed with the tabulated weight of the car. I saved a copy of the sheet and began modifying it to show the conditions that would have to be met for the car to be level and flush with the envelope, including the pressure force between the car "shoe" and the envelope. This involved a numerical calculation to get the approximate position of the car shoe's center of pressure. By this time, I was so confident of my skill with the spreadsheet that I did that in Perfect Calc as well, instead of firing up the old TI.

Now came the acid test. I had to manually adjust the tabulated cable tensions to get equilibrium in the new, car-level configuration. This was shockingly easy, and I had plausible results within less than twenty minutes. A quick check of the tensions showed that there was an adequate margin of safety in even the most highly loaded cables - and those cables were the ones whose loads were reduced during steep climbs and descents. With some trepidation, I telephoned the riggers at Tillamook and gave them the new tension settings. Then we waited by the 'phone.

Forty-five minutes later we got a call from the crewchief, who just said "the car looks a lot better now!"

The next day, when we flew up to Tillamook, the Kaypro was in the rented Cessna 172 with me. It was the hero of the hour, and everybody wanted to see it and to see a demonstration of the spreadsheet. It was quickly forgotten, however, when it was discovered that we did not yet have a ripline deployment system. Minutes later the Kaypro was shut down and closed up and the sewing machine was chattering away on a hastily sketched ripline design...

Next Story »