Return to Home Page



Efforts to replace the Nicolet 1280

It was pretty obvious in 1983 that advances in technology were obsoleting the Nicolet 1280. There was a search to find a replacement. The 1280 was used by the Nicolet FTNMR and Fourier mass spectroscopy divisions, however, the infrared spectroscopy division was by far the largest and most successful. So, the needs of FTIR had high priority.

The 1280 had a raster display, a fast 16 bit ADC and a the 24 bit fast Fourier array processor. Nicolet 1280 software included an operating system and a lot of application software. All would have to be replaced—preferably with something better and cheaper.

Clearly, Intel's 8086 wasn't going to do it with its archaic 16 bit architecture and its baroque segmented addressing scheme. (I never would have guessed that the Personal Computer revolution would ride on this inadequate steed.) However, Motorola's 68000 was almost a Nicolet 1280 replacement. It was only 16 bits on the outside but it was a true 32 bit computer on the inside.

The VMEbus

Motorola had a non-proprietary standardized bus and cage system called the VMEbus. It evolved and by 1983, it had true 32 bit data and address busses. The VMEbus also had a prioritized direct memory access (DMA) scheme similar to the 1280. That mean the processor, the ADC and the hard driver could access the same memory simultaneously. The Motorola VMEbus launched several small companies. They manufactured card cages and VMEbus boards.

Motorola introduced the MC68020 in 1984. It had a 256 byte data cache, a 32 by 32 bit multiply instruction and it was 32 bits on the outside and the inside. The MC68020 could replace the 1280.

Nicolet adopted the VMEbus, but added a third connector. The card cage was also greatly enlarged. The result was called the NicoBus. Industry standard boards and Nicolet proprietary boards could be intermixed. Nicolet could buy a computer on a VMEbus card and plug it into the NicoBus. Nicolet could be out of the computer business.

The same team that developed the array processor, developed a raster graphics processor for the NicoBus. The graphics processor included a BitBlt processor and a dedicated Motorola 68000. A BitBlt processor performs Boolean operations between pixel planes as unit operations. It took a large number of TTL parts as well as programmable logic to implement the BitBlt processor in 1983. Today, virtually all desktop computers use BitBlt processors. Windows style graphics depend on BitBlt processors.

About the time we finished the graphics processor, it became apparent that the NicoBus was a bad idea. The graphics processor board was big—as big as the boards in the 1280. The FTIR spectrometer had shrunk to something the size of a microwave oven while the data system hardly shrunk at all. A NicoBus data system would have been bigger and more expensive than the spectrometer it was supposed to be embedded in. Eventually, the data system would have to shrink to one or two little boards stuck in a corner of the spectrometer.

The NuBus

The NuBus was very attractive to Nicolet in 1984 because it had a 32 bit data path. It seemed like a safe bet because Apple Computers used the NuBus in the Macintosh II. And Steve Jobs, a cofounder of Apple Computers, used NuBus in his NeXt computer. Nicolet's plan was to design an application specific NuBus board to interface to an FTIR. The Macintosh would serve as a data system. That plan went awry when Apple dropped the NuBus in favor of the PCI bus, causing the NuBus to disappear.

The Nicolet 1280 replaces itself

The Nicolet 1280 was only half way through its life in 1983. It would live until 1996 and 3,000 more would be sold. There were two basic reasons for this. The existence of large amounts of high quality application software was one. The other was the fast Fourier array processor which gave the Nicolet 1280 an unmatched performance advantage. The raster display may have been important to.

top