My career began when I founded Dual Systems Corp. with my PhD research advisor at U.C. Berkeley, Jay Singer. I was fortunate to lead a great team there that developed the first commercially successful MC68000/Unix "supermicro" and the first disk controller with fully hardware-implemented zero-latency transfers. Our disk controller transferred data faster than competitors at the time, acquiring an entire track of data in one rotation regardless of where on the track the disk head first settled.
My job designing VME bus disk controllers also entailed helping customers integrate these boards into large systems--a tough task requiring travel to customer sites on short notice. I really wanted a tester board we could ship to the site that would identify a specific error in the system, saving me the trip. This led me to develop the first bus anomaly trigger (automatic bus protocol/timing violation detection system), or "BAT." Its parallel triggering architecture could concurrently recognize 105 different VME bus timing or protocol violations in real time and trigger a logic analyzer within 70 ns of detecting an error. Dual wasn't interested in selling BAT, predicting a market of only 50 units per year. So in 1987 I bought the rights, founded Ultraview Corp., and sold over 2000 BATs.
Recently at Ultraview, I designed a line of very high-speed PCI data-acquisition boards, including a new A/D board that can acquire 8 Gbytes of uninterrupted data at an aggregate speed of up to 2 Gsamples/s, making it the industry's deepest A/D board. We achieved these memory depths using a novel SDRAM-based memory architecture by Tom Watkins. It is satisfying to see our A/D boards used in high-end scientific, defense, and intelligence programs, where their large record length and bandwidth enable capture of many seconds worth of wideband spectra.
Two years ago, Steve Chacko and I formed Furaxa Inc. to bring to market a novel method of producing ultra-short (10 ps) electrical pulses and sample apertures (snapshots of a voltage at narrow points in time). Contrary to earlier methods using step recovery diodes and nonlinear transmission lines to make fast pulses and apertures, our method uses a simple transistor architecture suitable for integration into matched arrays on monolithic chips. Our technology generates both the rising (start) and falling (end) edges of a pulse or sampling aperture in response to a single edge of a clock signal slewing between two dc voltages that can be adjusted to alter the width and position of each aperture or pulse with sub-picosecond precision. (See www.furaxa.com for more details.)
Luckily, Hal Levitt of the Naval Research Lab recognized the technology's potential, and we received an NRL contract to fabricate a run of gallium-arsenide ICs. We put eight of these pulsers on a pc board and built custom waveshapes out of 125-ps wide pulses on 125-ps delay taps, creating an 8-Gsample/s arbitrary waveform. This led to a second NRL contract, for which we are currently designing much faster indium-phosphide ICs, each with four pulsers or four samplers. These should achieve 10-ps pulse bursts, hopefully allowing signal acquisition or generation at rates of 100 Gsamples/s!
We have begun talking with partners who might incorporate our Pulser/Sampler IP core into their ICs. Applications could include ultra-wideband (UWB) transmitters, in which custom-shaped, changeable UWB pulses could be made of bursts of 16 of our "micro-pulses." This would yield tailored frequency spectra that compensate for antenna and transmission channel distortion, thereby increasing signal-to-noise ratio (SNR) and range and eliminating out-of-band filters.
I've been very fortunate to find the resources that allow me to be an inventor. While engineers don't often get the acknowledgement they deserve, the ultimate reward for our efforts comes through the process of bringing new technology into the world. On occasion, we bring a new capability to the industry, something never accomplished before, and it flourishes. For me, the ultimate thrill is bringing novel electronics to life.