The LM-2 was a re-packaged MIT-Cadr with some minor improvements. A temporary stop-gap go get some revenue, a surprising number were made, around 80 as I recall.
The L-Machines (3600 Series)
The 3600 series was the first new design. A much faster architecture, many instructions (such as “push”) could be done in a single cycle.
The first version of the 3600 series was called the 3600. It was often referred to as “the refrigerator”, though in truth it was larger and heavier than many refrigerators.
The 3600 was repackaged and “slightly” redesigned, and named the 3670. The package was a little smaller. The redesign ran into numereous problems, however, as it was rushed into production. There were no effective tests. Bob Kerns, who was in Chatsworth to install new software for manufacturing, hurriedly wrote test suites, which most systems failed on numerous counts. After hurried repairs and board swaps, and a FEP software fix (there were numerous FEP software uprades, on ever larger PROM’s for the first year or so), only the network problems remained. These were later fixed in the field.
Eventually the early problems were overcome, and the 3670 became a reliable workhorse.
The promised IFU, however, was long delayed and plagued with reliability problems.
G-Machines (Gate-array 3600 architecture)
Line drawings done by (and courtesy of) Kent Pitman.