Friday, June 19, 2009

A Week of Work != Week of Results

In an ideal world, the amount you work on a problem would be directly proportional to the results you get.


In the real world, it ain't gonna happen that way.

Sometimes all your hard work can be lost on a mistake.

Two students in my office are currently working in the fab (clean room where silicon wafers are processed, like for making computer chips) to create some very hot science. They need to make a silicon wafer with a particular pattern on it. Of course, they are making five wafers, because they expect some will be damaged or not come out properly. It is expensive to buy wafers, expensive to use the clean room, and expensive to go through the multitude of steps it takes to process a wafer.

They had been working about two weeks and were about halfway through processing their wafers, when one day they arrived for lunch with slumped shoulders and disbelieving looks on their faces. I asked them why they were so dejected, and it turns out that they had just destroyed their wafers. Literally turned them into dust. Poof!

They had put their wafers in the spin dryer, and one of the components in the spin dryer was upside down. Nobody knows how the component got in there upside down. Previous student messed it up, perhaps? It didn't matter, the result was that all five wafers shattered and were spun into little shards of silicon and dust.

Sometimes the problem you were working on turns out to be moot.

I worked for a week straight back in February to model a pneumatic system. I had a huge, complicated model that included compressible flow, springs, dampers, inertia, hardstops, and sonic flow calculations. Then I realized I could make some simplifying assumptions, and it turns out that I could eliminate all the fancy parts of the model. Then I did some more digging and found that Simulink has a toolbox ESPECIALLY made for hydraulic systems, and I could have just used three programming blocks and been done with it.

So when I went to talk to my advisor that week, and I had a model simple enough to be almost trivial. It looked like I had just thrown it together that morning, but I had actually spent the whole week on it.

And then sometimes, a simple five minute fix solves a world of headache.

Our lab has bought a new micromilling machine, but it's been buggy. Micromilling machines are hot new technology, but the emphasis is on the "new." The software and hardware still have some kinks to work out. In fact, in the user's manual, the start-up proceedure lists "Please push button A to phase motors. If motors do not phase, push button again. Repeat." Lovely.

Software bugs I can handle, but a major problem was that the spindle kept stopping randomly, which immediately breaks the tiny little tool when it plows into your workpiece while not spinning. We have tried all sorts of trouble-shooting and possible fixes for this problem.

Finally, a representative of the micromill company instructed us over the phone to take apart the electronics of the controller and inspect some circuit boards. Turns out there were a couple jumpers missing on the card, and that was causing all the problems. My labmate ran to Radio Shack and picked up a pack of jumpers for $.99, applied them in the appropriate spots, and we've had no spindle problems since then.

A week of work for a $.99 fix...

So really, why work at all?

It's enough to make me throw up my hands. Who knows if *this* work will get me results, or *this* work is not worth doing? Ack! I guess all you can do is work your best and hope and cross your fingers...

1 comment:

  1. At least you're not doing organic synthesis...I don't see how people go on in that field, what with their 20-step-sequences that end with a few milligrams of material that can never be converted to the target...ouch.
    Hope next week is better. Bad lab days suck!