Back in 1986, I was just starting my career in computer electronics. I took a job at E-Systems (which was bought by Raytheon some years later), and my first assignment was to design a computer card with three digital signal processors (DSP’s) on it. I was in heaven. The DSPs were ADSP2100’s, made by Analog Devices, Inc (ADI). I spent the next couple of years designing more DSP cards using those devices (and some of their successors). Once the cards were built (and debugged), I wrote software for them.

Around 1990, E-Systems decided to get out of the board-design business. Instead, they started buying commercial off-the-shelf (COTS) boards. I was disappointed, but not unemployed. Someone still needed to write code for those things. And writing code was fun too. We tried a couple of boards, but eventually got the chance to use one featuring ADI’s newest chip (at the time) – the ADSP21020. This was their first floating point chip, and it was a thing of beauty.

In the late 90’s the company decided to quit writing signal processing software, and integrate COTS software instead. Using Texas Instruments DSP’s instead of ADI’s. I thought that was a horrible decision, and it had been compounded by other, equally horrible decisions (not least among them was a switch from Unix to Windows).

Soon after, an opportunity came up for me to work where I am now. The guy who provided us with boards with ADI’s DSP’s was now the president of another company that used ADI chips. He offered me a job, and I jumped on it. That move eventually landed me here in New Hampshire.

Then a couple of years ago, ADI decided to quit making floating-point DSP’s. It was a scary time. The world was going with the less capable (and much cheaper) fixed-point DSPs. Heavy duty processing has been shifting to the PC architecture. The only ones still using floating point DSP’s are the military.

With no new floating-point processor, we had no roadmap to the future. Our president came up with a compelling new strategy based on combining FPGA’s with DSP’s, and we’ve stayed in business (and are doing well).

Today, the pres pulled me aside and said he wanted to talk to me for a few minutes. He introduced me to a guy who was one of the designers of ADI’s last floating-point processor, and he was now the principal of a start-up. His new company is designing a new floating-point DSP. It is essentially, a continuation of the roadmap.

The “few minutes” turned into three hours (I didn’t get home until after 7:00pm). We talked about his new chip’s architecture, and I told him some of the features I wanted to see in it. He seemed enthusiastic about most of my suggestions. I have my fingers crossed.

My dream is that I will get to design a new operating system for this chip. I have ideas.