Tuesday, November 6, 2012

Big Dumb Science - Or How I Spent The Summer of '91 (Part 1)

You can find the other parts of this here : Part 1 and Part 3.
After the 'real' job I was working at the beginning of 1991 went away when the company I was working for failed rather spectacularly, I was desperate and out of work for a whole two weeks. I will probably write about that company in a future post, but there are aspects to that particular job that I should probably give some thought to before putting it down in the blog. In any case, I called the people I had worked with before and set up interviews, as well as sending my resume to the company’s advertised in the Sunday want ad supplement.

Job hunting back then was a very different thing then it is now. When looking for work I would scan the weekly want ads. For tech jobs, I basically had to wait until Sunday when the Washington Post would do it's technical want ad supplement.  While there were probably as many 'head hunter' outfits back then as now, most companies preferred to list their jobs in the Sunday supplement. I would mail or fax my resume to the company and hope for the best. Since there really wasn't an internet or e-mail, that was the way things were done. Yes - it does seem awkward. Ironically many of the engineers hired in that era invented the infrastructure that created the internet.

Being especially anxious, I resorted to looking for engineering companies in the phone book, and cold calling them to check for openings. At that point in my career, I had committed to being an embedded systems programmer, so I had little or no interest in the plethora of COBOL and DBASE jobs that were the primary programming hires at the that time in the Washington area. There were a great many government 'toy shops' in the area that did specialized devices, many of which were developed in custom hardware and required someone familiar with embedded or device level programming. On one of these calls, the gentleman told me he did indeed need a programmer and would I stop by his business for an interview.

When I arrived at the address he had given me I was immediately confused. This was a residential address. I knocked at the door and was greeted by a Indian man in his early 40s and cloying waft of curry. He invited me in and we did the usual interview thing there on his living room couch. He explained that he was a PhD that had been awarded a SBIR contract to construct a lightning detection, quick disconnect, and deflection device. The work would be performed in his basement and I would be working with a tech and an electrical engineer. He offered me a lower initial salary than I wanted, with the understanding that he would automatically increase my pay on a weekly basis up to a higher than expected salary. Being in need of a paying job, I took his offer.

I was familiar with the SBIR contracting process as I had been exposed to it at my previous job where one of the engineers I worked with had secured for the company a lucrative SBIR contract for a innovative spacecraft power supply. This gentleman's idea was a unique solution to solve a problem for the US Army at Fort Huachuca, AZ. It seems that they had a wooden tower there that they used to suspend tanks and other military hardware so they could do radar studies. The tower itself could not contain any metal, so it was an expensive asset to construct and maintain. Ft. Huachuca was unfortunately subject to violent storms which produced some amazing lightning. A lightning strike on the tower or the adjacent radar equipment could cost significant money for repair. 

Edwards solution was threefold. First, the system would sample the environment via ground sensors and automated weather reporting to determine a threat level. The increasing threat level would activate the other parts of the system. One part of the system would arm a quick disconnect relay that would detect a beginning of a surge and isolate equipment before the surge could travel to key equipment. Yes, this is equivalent to surge suppressors found in every power strip today. The other element of the system was more interesting. If an immediate threat was detected, the system would activate a water cooled laser to fire, ionizing a column of air much taller than the tower, and leading the impending lightning strike safely away from the tower. There were two critical elements to the laser. First it had to be at the proper wave length to ionize air. Secondly, it had to be of sufficient strength. In this case, that called for a water cooled 10 watt monster with a collimated beam not visible to the human eye.

The processor chosen for the system was an odd TI part. Specifically it was a TMS370, normally used in automotive applications. However, it had all the elements that we needed. It had A/D pins which we could use for the ground sensors, and a serial port that could be used to dial up a weather reporting service. Again, there was no internet there, so and Ethernet connection was not a real high priority. There was no operating system as the microprocessor was a simple design and code space was limited. Being a TI part, there was sufficient (free – important as cost containment was too important to my boss, as I will illustrate later) tools and sample code for the serial and A/D functions. The ground sensor was basically a long rod pounded into the ground, with an exposed wire at the top and bottom of the rod, opto-isolated from the system (in case of the inevitable lightening strike). The concept was that before a lightning strike, there is a buildup of electrical charge in the ground. Once the lightning releases, the charge dissipates. This charge / discharge pattern can be measured from some distance away. Or at least was the theory that I was presented with.

What followed that summer was probably some of the most challenging technical work I had done up to that point in my career, and ending in perhaps the most disappointing way.

Next: The System

No comments:

Post a Comment