Beginner's Luck

I was born in 1947 and grew up in Israel, graduating as an electrical engineer without suspecting I would spend many years in optics. In 1973, after completing my mandatory army service (Even the Israeli Army makes mistakes.), I moved to Vancouver, Canada.

I was a bit surprised to find very little high tech here compared to what I was familiar with in Israel. All I knew about Vancouver was that it is beautiful and not cold (In the pre-Internet days, I picked up this information from an atlas, based on the isotherms.). The Canadian Embassy in Tel Aviv was no help. They just sent me the standard tourist brochure emphasizing the endless wilderness of British Columbia, not exactly what I wanted to hear when arriving as a penniless immigrant. On the other hand, I knew that if a million people could make a living in Vancouver, so could one more.

I spent the first few years as a hardware designer (DSP and Telecom) at MDA, one of the best companies in Vancouver. At the time, shortly after my first son was born, I was looking for a project I could do at home, so I could spend more time with my infant son. (Such ideas come only with your first born.)

MDA was buying high-speed film recorders to record satellite imagery for about $400,000 apiece, so I boldly told the management I could design a recorder that would do a better job for a tenth of the price. Not being biased by any knowledge of the subject, I proposed a fresh approach. 

The management, free of similar biases, agreed and gave me one year. After a crash course in optics, I changed the design, but surprisingly managed to deliver a shippable prototype in 12 months with only one person working with me. I had a small metalworking workshop at home, many of the machines home-built, and this allowed me to fabricate most of the parts for the prototype myself.

I now have a wonderful CNC machine shop at home, but I don't have the boundless enthusiasm of those days. However, I still build all my prototypes myself, finding it to be faster than sending out drawings and waiting for parts.

 

Gelbart has a prototyping facility at his home, with a lathe and other equipment. Photos courtesy of Dan Gelbart.

That film recorder, sold under the name FIRE 240, became a hit, and the operation was spun out of MDA as a separate company, Cymbolic Sciences, which is using the same basic architecture today. One nice thing in optics is the very long life of products. Many good products enjoy a life of 30 and even 50 years with little fundamental change.

This initial "beginner's luck" plus my unbounded optimism (There are no pessimistic entrepreneurs.) caused me to search for a difficult problem combining optics, electronics, and mechanics in order to start a company around it. There are only three large commercial markets using products combining the three: optical data storage, semiconductor manufacturing, and printing. Together with another MDA fellow, Ken Spencer, we started Creo.

The modest mission of Creo was to solve the world's data storage problem by combining the high-areal storage density of optical disks with the volumetric efficiency of tape. (This was 1984, before the great advances in magnetic storage areal density.) The fact that an optical tape recorder was considered a very difficult project, tried, and given up on by large players such as RCA and Philips, only made it more interesting.

This was the beginning of a very interesting 21-year period in optics. The person whom I give credit for most of the optics I learned is David Kessler, who was recommended to me as an optical designer. I negotiated with him to join Creo, but when he finally declined, I was left without an optical designer and a project ready to go, so I went back to the optics books.

Being young and having no exposure to business (My father was a writer.), I confused technical difficulty with financial rewards. At Creo, we had an idea not to use private investors. Instead, we would be self-funded by doing custom R&D for other companies in parallel to developing our own products (another bad idea). We were also helped by the Canadian government, which gives money to any startup that has a good track record of successes and is seriously committed to spending it.

After a few years of hard work, we were able to start manufacturing optical tape recorders. The financial rewards were small compared to the heroic efforts that went into the development.

The whole process reminds me of the fellow whose Social Security number matched the winning number in the lottery. Amazing, but he got nothing. If there is a lesson to be learned here it is to start with a simple product, to generate cash soon, and use this cash to develop a more ambitious product.

The optical tape recorder was an unusual product for a startup in more than one way. It required components with extreme mechanical tolerances and servo components with extreme electro-mechanical performance. Since they were not commercially available, we had to develop and manufacture them. The optical tape was made by ICI in the UK, who are wiser but poorer for it.

While the optical tape recorder was a bad product choice, it was an excellent opportunity for generating new technology for other projects. A bit like the Russian proverb: "There is no such thing as a useless person; you can always use them as a bad example."

After this project, everything looked easy. When we wanted to get some R&D and production work from other companies, we just gave their technical people a demo of the optical tape recorder. After that, they had no doubts we could meet their requirements. This is how we got the contract to develop and manufacture a PCB film plotter for Orbotech and several imagesetters and proofers for Dai Nippon Screen. These machines, of which hundreds were made, funded Creo and created good profits.

Working on optical data storage made me aware of the resolution advantages of using very nonlinear recording materials. The film recorders I had worked on before used integrating media, a silver halide film which was subject to reciprocity. So even small amounts of stray light and aberration added up quickly and degraded the film. When recording on a non-integrating material such as an optical disk surface, the data bits look perfect even if the optical beam is not.

This is due to a combination of the sharp threshold of thermal processes (such as melting) and lack of integration: The stray heat of the previous bit evaporated before the new bit was written, so the recording process can be designed (by using 2D interleaving) to have complete freedom from "proximity effects."

At about the same time Kodak developed a thermal printing plate, and this gave Creo the opportunity to come out with the first successful "Computer-to-Plate" system using thermal plates and invisible light at 830 nm to achieve higher resolution than traditional systems at half the wavelength.

The writing is done using a 240-channel light valve developed by Creo. The cross section of each pixel exposure is almost square instead of Gaussian, giving exceptionally good process control.

Creo was sold to Kodak in 2005 for about $1 billion, and I retired from optics a happy man.

 

Gelbart reconditioned the power supply for Ted Maiman's original ruby laser and fired it for a tribute symposium on 16 May 2010.

Amos Michelson met Dan Gelbart in the 1980s when he (Amos) was general manager of Optrotech, which was reselling imagesetters from Germany at about $100K a pop. Amos was looking for an imagesetter that was twice as fast, twice as large, and offered four times the resolution.

David Nims, a colleague, suggests a tiny company in Vancouver, Canada, whose chief engineer was Dan Gelbart.

David and Amos take a plane from Israel to Vancouver to meet with Dan. They ask Dan how much money he needs to develop the photoplotter engine.

Dan says $500,000. Amos keeps a straight face.

Then they ask Dan how much each engine would cost. Dan says $35K.

Amos makes a noise, now almost completely convinced the guy was nuts. Dan misunderstands the grunt and then quotes $28K per engine.

Amos calls a timeout so he can chew David out for flying halfway around the world to meet with a wacko engineer. Somehow, David convinces Amos to take a chance on the guy.

Dan delivers the first engine in 1989, about a year after the first meeting. It's got a motor inside from a household furnace. The optical resolution changer was based on a model airplane servo unit.

Does it meet the spec? Baby, it exceeds the spec. Creo ends up shipping more than 300 units and supplies spare parts for machines in the field well into the 2000s.

-From DJ Dunkerley at www.prepresspilgrim.com.

Note: Dunkerley is a former Creo employee. Michelson became the CEO of Creo and is the current chairman of Kardium, where Gelbart is technology adviser.

Have a question or comment about this article? Write to us at spieprofessional@spie.org.