Codex - modems, and the mathematics underneath them - A company whose product was signal processing sold as a box, in the window a regulator had just opened.

Codex built high-speed modems in the years after the interconnect decisions made it lawful for somebody other than the telephone company to attach equipment to a line. Its work sat on the mathematical side of data communications - getting more bits through a voice-grade circuit than the circuit looked capable of carrying - which is a different discipline from the switching and multiplexing most of this timeline is about.

Codex was incorporated in July 1962 by James Cryer and Arthur Kohlenberg, a director and a chief scientist at a Boston-area research division of Melpar, for an unremarkable reason: their employer was moving out of state and neither wanted to leave the area. They took their knowledge and contacts and lived on defence contracts, and it was the uncertainty of that - never knowing whether another contract would arrive - that pushed them toward building products instead.

The product came from a conversation. MIT's Robert Gallager told them a 9600 bit per second modem was possible, four times the fastest commercial modem AT&T then sold, and pointed them at Jerry Holsinger, a 1965 MIT doctorate whose thesis was high-speed transmission over telephone lines. Holsinger had a prototype developed with National Security Agency funding, a company called Teldata, and no success raising money. Codex bought it in May 1967.

The AE-96 shipped in May 1968 as the world's first 9600 bps modem, and its specification is the most useful thing in this entry. It weighed about 125 pounds, held sixty-six printed circuit boards of resistor-transistor logic, and every pair had to be manually tuned to the specific telephone circuit it ran over - then re-tuned whenever the line changed, which it always did. A first-of-its-kind product that is barely operable is still first, and the market leadership went to the second generation, the L Series developed under David Forney.

There is a second lesson underneath that one. Bell Labs had known since the mid-1950s that these speeds were achievable and had only managed 2400 bps, while the design that worked came from one person outside the largest communications research organisation in the world. That is not an argument that big laboratories cannot innovate; it is a reminder that a problem important to one determined individual can outrun a problem assigned to a department.

Motorola approached Codex in 1976 and acquired it, and it operated as a Motorola division for decades before becoming Vanguard Managed Solutions and then Vanguard Networks. Its last strategic decision as an independent line is worth recording too: in 1984 it ceased its own local networking efforts and resold Ungermann-Bass products instead. A company that had defined the high end of one market could not cross into the adjacent one, which is the outcome far more often than the crossing is.

Codex ↗

Sources