About the company: When I joined Sperry in 1979, it was a major corporation, with something like $4 billion / year in revenue. The main focus at the Roseville plant was medium and large scale mainframe computers. An easy-to-use software product called MAPPER (kind of a combination word processor/spreadsheet/database with scriptlike capabilities) was a major selling point. Typical customers were airlines and government agencies. After being bought out by Burroughs, the new Unisys combined this customer base with Burroughs' large foothold in the banking industry.

General Comments: This was the first place I worked after I got out of school. In fact, they actually sent recruiters down to our school in Iowa to beg people to work there. A group of us went up to the plant for a tour/interview, and all but 1 of us received job offers. Looking back, it was a pretty good place to work: good money, good benefits, decent tools and equipment, plus the relative anonymity of working in a large corporation. In fact, I naively planned to work there until I retired. After the so-called merger however, downsizing reared its ugly head, big time. I survived the first few massive rounds of layoffs, but then in '91, they finally got me. Being laid off sucks, by the way. Here's a tip for all you CEOs out there: If you know you need to get rid of x employees, just shut up and do it all at once. When you drag it out for month after bloody month, it just sucks the morale right out of everyone. But hey, I'm not bitter.

Stuff I did when I was there:

• 1100/80 Mainframe Checkout -- The 1100/80 series was a fairly mature product when I hired on, so most of the checkout and diagnostic procedures were streamlined and automated. If the unit being checked out didn't have any problems, all you had to do was sit there and let it test itself. So what do you do when you're bored on 2nd shift, with a $2 million mainframe at your disposal? Write a Life program for it, of course! (I didn't know about Mandelbrots yet.) I had been working on a Life program at home for my 8080-based system, and was disappointed with the speed. So I sat down and pounded in machine code (no assembler!) directly into the main storage unit, and set it up to display the results on a 24 x 80 text terminal that was tied directly into one of the I/O channels. But I couldn't understand why my initial colony of cells disappeared from the screen the instant I hit the RUN command. After single stepping through the program to find the "bug", I soon found out why. There was no bug. My colony of cells had grown, flourished, and died out, all in the blink of an eye. After adding a loop to waste 99.99% of the CPU cycles, I could finally watch the colony's growth at a speed a human could observe. It gave me some perspective on the power of the "big iron".
  
  
• Support Controller Microcode -- All mainframe computers have a capability called Scan/Set; that is, the ability (using another computer) to read and write all the various bits, registers, and storages inside the machine. This capability is vital for debug and test, both at the factory, and in the field, to diagnose failures. In Sperry's 1100/90 series of machines, this Scan/Set feature was supported by a small box called a Unit Support Controller (USC) that was installed in every unit of the system. The USC sorted out all the details of talking to whatever unit it was attached to, while providing a common serial interface back to the diagnostic computer. The USC itself was moderately complex. It had an AMD 2901 / 2910 microprogrammable processor with a 32 bit control word. And it even had its own Scan (but not Set) capability. 128 bits worth of scanned data to help the programmer (me) debug the microcode. Unfortunately, the only way to view this data was with a little metal box with a lot of switches and LEDs on it. Eight LEDs, to be exact. So in order to see all the data, you had to select which byte you wanted to see with a thumbwheel switch, and then figure out which bit it was you were interested in. By the time you found what you wanted, you forgot why you wanted it. Something had to be done. So I commandeered an old TRS80 Model 4 that I found in a store room, and built an interface box that would plug into the USC just like the metal box did, applied some of my best CP/M assembly code, and soon I was scanning the USC just like the big boys. I could view all the bits and registers at once, on a nicely formatted display screen. I even added a trace function that stepped the sequencer through 10 instructions at a time, showing the full set of scanned data after each step. After that, debugging USC microcode was a breeze; my code was done before the unit that it hooked up to was even built.
  
  
• Other Stuff I Did:
  • Set up, installed, and maintained about 20 PCs for our engineering group to use as CAD workstations.
  • Generated and maintained large databases of automated test lists
  • Built three large test fixtures for use in training field service personnel

General Comments: Since the bulk of the employees were a 2-hour drive away, the home office had a real small company feel. For example, the engineering department consisted of only 6 to 8 people at any given time. This gives you a perspective on business that you just don't get at a big corporation. Although the pay wasn't in line with my previous salary at Unisys, I was ready to take any job, since I hadn't worked in a year. As it turned out, the lower pay was made-up for in new skills learned and the wide variety of interesting projects.

Stuff I did when I was there:

• A Little of Everything -- In a company this size, there is always plenty to do, and I eventually did most of it at one time or another. At first, it was mostly grunt work; soldering prototypes, building up test fixtures, machining various little plastic and metal boxes, stuff like that. Pretty soon, I was designing and programming the test fixtures, writing ECOs, hacking on the MRP system, and consulting on and debugging product designs. Debug here was much different than I was used to at my old job. At Sperry, I knew the design was proven, it was just a matter of finding the failed component. Here, (in a design environment) you didn't know if the thing even could work. Everything I had forgotten from school about circuit theory came back to me.
  
  
• PCB Layout -- When our CAD guy quit, I immediately volunteered to fill the void. I soon got the hang of the schematic capture and PCB layout package they used, and found that I really enjoyed it. There's nothing like the feeling, after spending hours on a PCB design, of seeing that first green and silver board come back from the board house, soldering in all the parts, and watching it run the first time. (Well, most of the time, anyway.) Some of the more interesting projects I worked on:
  • An infrared light curtain. You hook these up to big nasty machines (shears, presses, etc.) so they don't take your operator's [insert body part here] off.
  • A compressed air vending machine.
  • A controller for a system of filters that cleans the exhaust from diesel bus and truck engines.
  • A little meter that goes on the small oxygen tanks that you see people carrying around, so they know how much O2 they have left.
  • Some high-tech medical gadgets that monitor a persons heart on various ways. (Medical electronics people are serious about isolation!)
  • A cryogenic tissue sample storage unit (dubbed "the Sperm Freezer")
  • An electronic fishing lure.
  • A "smart" battery charger.
  • A small terminal device used by dealers in casinos to track things that casinos like to track.
  • A device that reads those tiny electronic ID tags that you can have implanted under the skin of your dog or cat.

About the company: -- FSI International makes equipment for the semiconductor industry. In other words, chip manufacturers like Intel and Motorola (and dozens more) buy machines from FSI to make ICs. (Sorry if I over-explained that, but when you say "equipment for the semiconductor industry" to most people, you just get this blank stare.) They have three divisions: Microlithography (processing the wafers), Surface Conditioning (cleaning the wafers), and Chemical Management (providing the chemicals into which to dip the wafers). FSI is what I would call a medium-to-large sized company. The home offices are in Chaska, MN., but they have offices all over the country and the world (usually near their customers). They do about $300 million a year in sales.

General Comments: -- I was a little worried when I came in the first day. I, and another new guy, had desks that were literally in a hallway (by the copier, no less!). But my boss assured me that our group (Chem. Management) would be moving to "the new building" (just down the street) soon. At least I had a phone and a PC with a network drop. The first couple of weeks on any job are usually spent reading, training, and exploring the network anyway. About a month later the new building was ready, and boy was it worth the wait! I feel compelled at this point to list some of the things that this company does right:

• Office Furniture - These are the most luxurious cubicles (not an oxymoron!) I have ever had the privilege to occupy. Spacious, one-person, high-walled cubes with plenty of storage and well-lighted desk space. Scott Adams may call me a nerd, but these are some damn nice cubes. And the chairs!
• PC Policy - At some companies, when a new employee comes in, the boss gets a new computer. Then the PC trickle down syndrome begins: Each level of management hands down its cast off computers to its underlings, disrupting productivity all the way, until the new employee gets some underpowered piece of crap with an EGA monitor that's dim in one corner and a tiny hard drive that's already full of stuff he's afraid to delete because he doesn't know what it is and whether he needs it or not. Well, not at FSI. New employee? New computer. When I walked into my brand new cube, there was my brand new PC with its big processor, big hard drive and big monitor, all hooked up to the network and ready to go.
• Paging Policy - I don't know exactly who made this decision, but I'd like to shake their hand. The decision was this: "We got E-mail, we got voice mail. We're not going to allow indiscriminate paging in the new building." Now this may not seem like a big deal. In fact, you don't really even notice the absence of paging. But the minute you go back to a page-happy environment, you notice it, and you notice it bad. It's amazing how distracting those things are, even after you train yourself to ignore them. Your subconscious still processes the noise: Was that for me? What was the number? Who was that, anyway? Why do they want to talk to that person? I wonder where that person is? It's just got to affect productivity. Someone should do a study.
• Training Policy - FSI International takes employee training seriously. In fact, the number of training hours you've had is a major factor in the employee review process. They have in-house training facilities with full-time instructors, and a lab with most of the equipment FSI makes. Need training that's not available in-house? No problem. Just tell 'em when, where, and much the class is, and it's 'See ya when you get back!' No whining about the cost, or about "lost time" being away from the office. (Or if there was any whining, it was not apparent to me.)

Stuff I did when I was there:

• Program PLCs, mostly -- The Chemical Management division has two main products, both of which are PLC based. The first is what's called a Chemical Dispensing Module, or CDM. Simply stated, the job of the CDM is to take a chemical from the drums that it is shipped in, and pump it up to the device that needs it (referred to as The Tool). Sound simple? Well keep these design criteria in mind:
  • It has to be ultra-clean. It absolutely cannot add any particles to the chemical, and ideally it should remove some. This is done by using special Teflon tubing, pumps, filters, and other components specially made for ultra-pure applications.
  • It has to be ultra-reliable. A fab (the place where ICs are made) can be losing zillions of dollars for every hour of production that is lost, so it damn well better not be because the tools couldn't get chemical.
  • It has to withstand some very nasty chemicals. Making chips requires a wide range of acids, bases, and solvents.
  • It has to be safe. Since the chemicals involved tend to be pretty dangerous, it's a good idea to make sure none of this stuff is leaking out anywhere.

  Chemical Management's other main product is Control Systems. This is a PLC with a lot of I/O that controls and monitors the entire chemical distribution system in the fab. It keeps an eye on the CDMs, monitors leak sensors and valve box covers, and controls valves throughout the system. All of this information can also be displayed on a PC that gives the operator graphical overviews of how things are going, where a leak has been detected, or which CDM needs a new drum of chemical.

About the company:

General Comments:

Stuff I did when I was there: