Some people want to know — just what do you do, Dave?

The easy answer to that question was that back when I worked for Advanced Micro Devices (AMD) was that I worked in a cleanroom, and everybody at the time marvelled at Intel’s cheesy ads featuring the “dancing” fab guys.

Now imagine that — but sadly without the disco, loud color fab suits and the opaque facial screen. The recent Apple/Intel ads are pretty close to the real thing in terms of appearance… but anyone holding wafers the way they are in the ad would likely have been taken out to the woodchipper.

That was me. More specifically, I handled some of the photolithography manufacturing in their R&D line, which for the 90s was bleeding edge technology. It was a great job at a great time during the Internet days.

So the answer then was easy — I was a “dancing clean room guy, except the dancing part”.

Times changed, and I found myself transitioning over to Perkin Elmer Optoelectronics. Same thing, similar position — but different technology. I went from worrying about REALLY small stuff (i.e. nanometer sized objects, which are much much smaller than a hair) to well, just SMALL stuff (microns), but instead about worrying about a chip the size of a postage stamp (okay, a little smaller than a postage stamp, but close enough) to a half meter square piece of glass. Now I’m doing similar work on automotive sensors.

Processes were the same, but then the game changed. Rather than making thousands upon thousands of sellable product in a day, I would obsessing over a single sheet of glass which would later be put into an X-Ray machine. It was then that I learned the true magic of statistics.

Clean room manufacturing and the methods of controlling yield/minimizing scrap on existing, established processes are largely based on statistics, not necessarily direct process engineering.

Typically an engineer walking into a manufacturing environment with an existing process isn’t expected to “invent”, rather “improve”. This sort of engineering boils down to process knowledge + creativity + applied statistics.

In as small as a nutshell as I can tell it —

If you know how to make something and you have an understanding of what things can vary (hardware/software/people/environment/methodology — sometimes this is called the “knowing the M”s — material/method/man/mother nature/measurement), characterizing what and how things can vary will help to build a better product.

The best analogy I can strike for the non-math folks would be bowling or golf. But I like bowling better — so bowling the analogy shall be. 🙂

If your bowling game is off and you want to improve that first ball to a strike — you can brainstorm the variables: your stance, your approach, your release (method), your height, health and whether or not you’ve had a few beers (person), the ball weight, the ball material (hardware), the oil on the lane (environment), the condition of the boards on the way to the pins (environment), the conditions of the pins and how well the pins are placed (hardware, but not yours), etc.

Now determine what variables can be controlled and what’s noise. Yes, the quality of beer choice could very well figure into the whole mix.

Typically unless you have a LOT of money environmental variables are too hard to control. All of those controllable factors can be tested with enough patience and money, but also remember that some of them are not stand-alone. Combination of variable adjustments can often be overlooked factors.

Noise is just that. Some things are uncontrollable and unknowable. Usually in the corporate world, this translates into ‘beats us’ and ‘would be too expensive to chase’.

Sometimes chasing noise is what turns normally intelligent folks into geeks, heretics, and yes… also leads to great inventions and creative leaps.

But ultimately, bowling (okay, the first ball, the second one changes things up a bit) boils down to trying to do the same thing over and over.

And we’re not perfect beings, so sadly, a potential strike becomes a spare or a few pins left over. Those “few pins”, when applied to manufacturing, translate into the bottom line. The less leftover pins, the better the ballgame.

So… back to the original question: “So.. what is it you do, Dave?”

My first thought — “Do I give them the LONG answer, or the short one?”

At this point I usually get this distant look in my eye, which probably looks like Brent Spiner’s Data’s facial expression when he’s churning data in his skull. Fortunately, if Valerie’s around and sees that look, she knows I’m reaching for the long answer and gives me a semi-subtle tug or jab to remind me that there’s still an easy answer:

“Uh… I’m an engineer.” 🙂