by Robert Getsla
In 1990, I was hired by the Stanford Linear Accelerator Center (SLAC) in Menlo Park California, as a Science and Engineering Associate I. Shortly after I was hired, I given an old desktop computer and a shared office. But a few months later, I was "evicted" from my office and my computer was taken away, both for reasons I never understood.
Some time later, the engineer I worked with quit and the other S&E Associate I worked with retired, leaving only me to design and maintain all of the Broadband networks at SLAC. In all, there were five bi-directional CATV like networks, each carrying video and high speed data at SLAC. The largest of these Broadband networks was effectively the spinal cord of the lab, because it linked approximately 80 slave computers scattered along 6 miles of network cable, with the main control center VAX cluster, using high speed cable modems.
In 1993, SLAC won a competition to build a new experimental accelerator called the Asymmetric B-Meson Factory. Building the SLAC B-Factory involved removing and rebuilding two existing broadband networks in an abandoned ring tunnel some 1.6 miles in diameter and 50 feet underground. One of the rebuilt networks would connect the 15 or 16 slave computers necessary to run the B-Factory with the main control center VAX cluster. That network would also be carrying some 20 or 30 closed circuit television pictures to the control room for the accelerator operators to view. The other broadband network was to be built with an intentionally leaky coaxial trunk cable, to provide 2-way radio voice communications and support 1-way pagers throughout the B-Factory's tunnels.
The designs for the original versions of these networks had been done several years before by an engineer who no longer worked at SLAC. I learned that he used a program of his own design, written in BASIC, running on his desktop computer.
I was assigned to rebuild these broadband networks, but I did not have a PC. So I began by reading the Cable TV equipment catalogs and the Cable data books from several manufacturers. I found all of the formulas I needed in the backs of those catalogs and data books. I then wrote the programs for my TI-59 based on those formulas, so I could calculate the spacing between the required bi-directional amplifiers and calculate the other details necessary to building these networks.
I later refined my programs to do "running totals" of gains and losses at various frequencies throughout the proposed cable plant, so I could spot cumulative errors. I tested my programs by recalculating all the examples I could find in the catalogs and data books using my TI-59. My numbers matched those in the books, often within less than 1 dB.
I then used my TI-59 to design the two new networks, including several variations of each design, exploring several cable and amplifier options. In all, I did a total of 8 designs. When I was done, I then had to present each design to a committee for them to review. Meanwhile, I also had to price out the hardware and labor required to build each option. Of course I used my TI-59 for this, too. My TI-59 came through like a champ.
After two of my network designs were chosen, I then had to order the necessary parts and build the networks. Amazingly (I'll explain this later on in this note), every equalizer and attenuator setting came out within a dB or two of my calculated values, and both networks worked perfectly after one alignment pass in each direction with a sweep generator and spectrum analyzer.
But the history of Broadband networks at SLAC suggested a much different outcome, because the Controls Department CATV group was forced to continually "chase" the previous designs, "tweaking" them almost daily, especially in hot weather, just to keep the linear accelerator working.
I now believe that when I was given this assignment, someone was expecting me to fail. But I did the unexpected when I brought in my own TI-59 and when I wrote my own programs based on the system design formulas I found in those catalogs and data books.
About a year after I commissioned both of my Broadband networks, I found a diskette with what I believe was a copy of the BASIC program that had been used to design all of the Broadband networks at SLAC before I came to SLAC. I was curious about that program, so I took the diskette home and ran it on my own PC. I entered the same information into the program that I used in my own successful B-Factory network designs. The BASIC program produced amplifier spacings about 10% too far apart according to my TI-59 (in terms of cable length) but otherwise the results looked OK. I think THAT was precisely the problem -- the results from this BASIC program were believable, but wrong. When I listed the program, I found some of the cable loss formulas had suffered in translation from the printed page into BASIC code. Those translation errors apparently were the cause of the incorrect amplifier spacings and they led subsequently to all of our grief with those previous designs.
If only someone had run "known good numbers" through that BASIC program, they might have found those coding errors. I was able to easily proofread my TI-59 programs, so I managed to avoid making mistakes of this kind. I owe this to the algebraic formula entry system used in the TI-59.
Both of my Broadband networks are still running flawlessly several years later, and no further "tweaking" has been required. And to think, if I had been allowed to use a desktop computer, I might have used that same defective BASIC program, which would have resulted in two more unstable Broadband networks at SLAC.
I still have my TI-59, but the rubber roller inside that moves the magnetic cards across the magnetic head has become very sticky and gummy, so I can no longer use the card reader/writer. The roller in my PC-100A has also become very sticky and no longer usable, too. If you know where I can obtain replacements for these rollers, I would appreciate it. I have several dead battery packs for my TI-59, each with one or more shorted cells. With some luck, I might be able to repair one or two of my battery packs by replacing the original nicad cells with high capacity nickel-metal hydride AA cells.
Thank you for your web-site. I really appreciated seeing pictures all of those TI calculators (some of which I still own.)