Tolerance

The dog days of summer have come and I missed a post. Being that I'm not making a living on this blog and that the only people who read it are ones I talk to often...that is probably ok. I haven’t given up on the blog because I hope that a year from now there may be an audience and I can save on travel and presentations. With our main aviation program, the Boeing 787, on permanent hold, our business has slowed dramatically. Add to that the recession and credit freeze and a dependence on industrial products and we have a smaller COTSworking. But I think everyone at the company still feels there is something unique here. In the last few weeks our customers have come through to see that also and business is picking up.

In the three and half years that we've been in business, we have created a tremendous amount of stuff and value. Making fiber optic links work over an extended period of time and temperature involves research, trial and error, effort, and innovation. Not the innovation of the Telecomm world where cash and paranoia drive miniaturization and speed increases. Rather, the new ways of thinking about things at all levels. I give Luxtera and Infinera and Glimmerglass a lot of credit. Not the same for anyone who makes CWDM or DWDM gear or even transceivers …the business is so much of getting in line, not drawing a line. I'm very happy being Harold with a Purple Crayon and if the only folks who want to be in this book are companies like Protokraft and D-lightsys, so be it.

All of us who make parts for military or aerospace care about drawing that line, the new line to cross. We aren't chasing a predefined concept like PON or 40G, we're chasing more nebulous tasks like making aircraft lighter or creating multimedia shims to 30 year old warplanes. It's a hobby as much as a business sometimes. But that doesn’t keep us from paying attention to details. Human labor is always fraught with the inaccuracies and errors that we make. Building a transceiver with automated assembly lines (once setup) will yield a better part but when the volumes needed each year are only a few dozen to a few hundred, there is no automation. I'd like to think we'll find our way into some larger programs but for now, we are artisan transceiver manufacturers.

In highschool, I did a lot of painting and drawing. This moved to silkscreening and along with some friends, made our own press setup. I quickly learned about offset printing and the need for tight tolerances. To create an image in color requires at least four plates and getting them aligned was difficult. The shirt stretching was one thing, but the machine and frames another. At COTSWORKS, this translates into the tolerances of our transceivers. since i pay the bills I'm keenly aware of those tolerances and the costs to achieve them. 0.002" is stated specification and we achieve it on production parts. One transceiver from us might cost $100 while an equivalent telecomm part we recently bought only cost $18. How is that possible? The parts are designed very differently: we have separate power planes for the Tx and Rx and plenty of copper wherever we can get it in the PCB. We only use high quality parts and make sure everything is sealed. And, we're a little tighter on tolerances. I was struggling to put the low cost parts into a datacomm switch and only got them to work after two other technicians had tried and failed and spent several minutes at it myself. The problem? Tolerances...the parts were only rated to 0.008" but seemed further out than that. They didn’t line up right in the low-cost switch (Dell is a low cost switch damnit). And there it was in front of me, why our parts cost so much more. 30-45 minutes in a lab is nothing compared to several seconds in on a battlefield. One of our customers has a fixed time that they have to be able to bring up a redundant set of switches. Could they save $80 a part? Sure. And they would fail in the field.

I would like to have more time to work on this blog. it's not that I don't think about it. Priscilla Diem and I have talked about those tolerances and the ones that she encountered workign at NASA on the shuttle. Not only did I have to make sure we delivered parts to those tolerances which has kept me very busy, but I had to find a way to shut off the side of my brain that is capable of dealing with this kinds of things and switch to the side that lets me sit and type our a post. I had no measure for success or accuracy with this writing, just to keep going until i felt I'd made a point.

0.008, 0.002, or 3714 letters. the latter works tonight.

Ken