VME turns 30

Jack Kister and John Black of Motorola never perceived what would happen when they pinned out the 68000 microprocessor bus on an edge-card connector with a VERSAmodule card format back in 1979. In addition, Sven Rau and Max Loesel (both from Motorola/Europe) did not know what they had created when they took the VERSAbus pinouts and put them on a DIN pin-and-socket connector on a 6U card format in 1980. In 1981, John Black (Motorola), Craig MacKenna (Mostek), and Cecil Kaplinsky (Signetics) did not know that they would change the world of embedded computing when they created the first draft of the VMEbus Specification. In October 1981, at the System ’81 trade show in Munich, Germany, Motorola, Mostek, Signetics/Philips, and Thomson-CSF announced VME to the world and initiated the most successful open-computer architecture in history.

A little history

Along the way, many others were involved in the development of VME. Wayne Fischer (Motorola and then Force Computers) and Shlomo Pri-Tal (Motorola) ushered the VMEbus document through the Institute of Electrical and Electronic Engineers (IEEE) process and final approval in 1987. Eike Waltz (Schroff) configured the mechanical packaging for the air-cooled boards in Chapter 7 of the specification. Later on, Kim Clohessy (Dy 4 Systems Inc.) chaired the mechanicals for conduction-cooled boards (IEEE 1101.2). In 1989, John Peters (Performance Technologies) created the concept of VME64 (64-bit wide data and address buses added to VME). Kim Clohessy and I chaired the VME64 document and took it directly to the American National Standards Institute (ANSI) (ANSI/VITA 1.0) in 2002. Chau Pham (Motorola) created the Source Synchronous Transfer (SST) mechanism, which enhanced performance of VME data transfers, and took the document to ANSI for approval in 2003 (ANSI/VITA 1.5). Drew Berding (an independent consultant) then created VME320, a method of making the SST transfers run at even faster speeds. Hundreds of people, too many to name here, were involved in the creation and enhancements to VME over the years.

VME overwhelms DEC

The primary target market that VME exploited was the replacement of the expensive and cumbersome Digital Equipment Corporation (DEC) PDP series of minicomputers being used in many industrial controls applications. Also during the early 1990s, the military market adopted VME for their weapons, communications, radar, and sonar applications. Somewhere around 1992, the VMEbus merchant board market hit $1 billion in sales. In 1998, DEC went out of business and the remnants of the company were acquired by Compaq Computer. They could not compete with the VME ecosystem.

Today, VME board sales are still in the $600 million range, driven mostly by military system refreshes and upgrades. Industrial controls have moved on to inexpensive commodity PC boards. The remaining vendors in the market say they still get 70 percent to 80 percent of their business from VME. As we endure the defense budget changes this year and next, VME sales will probably grow to the $700 million range as existing VME card cages stay in place and newer, faster boards are inserted (refreshes and upgrades).

Why VME continues to grow

VME has been resilient and vibrant for the past 30 years because it has been an adaptive technology. It has integrated newer, faster processors (including DSPs), adapted to standardized PCI-based mezzanine boards, integrated newer and faster A/D and D/A chips, adopted FPGAs for I/O interconnects, and adapted serial I/O channels to the P0 connector, all while remaining backward-compatible with the previous VME boards used in the market. No other architecture in history has been able to protect the investments in older boards the users kept in place. But the primary reason that VME is still a viable technology after 30 years is the engineering leadership, creativity, and talent in our industry. The hundreds of people who have worked on VME over the past three decades are unmatched in their engineering skill and capability.

Into the future

VME will continue to be the architecture of choice for event-driven (interrupt-driven) applications, especially on military platforms, for many years to come. No other architecture can provide the real-time responses needed in critical embedded systems. Data-driven applications such as Signals Intelligence (SIGINT), Communications Intelligence (COMINT), radar, sonar, ISR, and so on) will begin their migration to high-speed serial links with the VPX specifications, the high-end replacement for VME now entering the marketplace.

Will VME boards still be available and heavily used in another 10 years? My bet is yes, it will still be a viable and vibrant architecture. As in the past, we will see a lot of cheap commodity junk electronics come and go within the next 10 years. But VME will still be alive and well on its 40th anniversary.

For more information, contact Ray at exec@vita.com.