For over 20 years, the VME specification has been remarkably resilient and amenable to enhancement. Over the years we have seen the inclusion of VME64, five-row DIN connectors, extension of its environmental capability including conduction cooling, the addition of the P0 connector, the introduction of PMC modules, and the addition of VME320 to the basic specification while we still retain a degree of backward compatibility at the connector level. But the last five years has seen momentous changes in the way systems are architected, particularly in the military environment.
Federated architectures with fixed-function subsystems are giving way to integrated and network-centric architectures that will make use of the new switched fabrics that are becoming economically available. These new fabrics, and the ever increasing signaling rates used in sensor, video, and I/O subsystems (such as Xilinx’s RocketIO) are beginning to show the limitations of the VME DIN connectors.
It is remarkable that the original DIN41612-style connector has survived so well all this time. What other piece of technology is still current and so eminently usable after 30 years or more?
The first moves towards addressing some of VME’s legacy issues was made by the VME Renaissance proposal of Motorola in the form of the VITA 41 effort which is now in the final approval stages. However, as the demands of VME grow to satisfy the new wave architectures, the streaming of raw sensor data into VME cards, and high-resolution digital video, the connector is running out of bandwidth. That, coupled with the increasing functional density of VME cards that require more I/O pins than are currently available, drove the defense and aerospace VME suppliers to perform a fundamental review of VME connectivity needs over the next 20 years.
Hence, the VITA 46 working group was convened by the leading defense and aerospace suppliers and consumers to address the needs of the next generations of products, and the systems into which they will be deployed. There was, and is, strong support for an evolutionary step retaining backward compatibility with VME64. But, as the requirements and available connector technologies were reviewed, it became increasingly clear that direct backward compatibility at the connector level had to be sacrificed in order to meet future requirements.
VITA 46 versus 41
VITA 46 is not to be confused with VITA 41, although they do have a number of major capabilities in common (Table 1).
From the table, it can be seen that while VITA 46 embraces the new features introduced with VITA 41, it necessarily has to go much further to properly become the new standard for rugged and harsh environment defense and aerospace applications. Even though VITA 41 maintains a level of backward compatibility at the connectors, VITA 46 has decided to take the revolutionary step of changing to a new type of connector to meet its extended set of requirements. The proposed connector is a seven-row MultigigRT2 from Tyco Electronics offering 192 differential pairs, and 48 single signal pins, for a total of 432 signal pins. This compares to 335 single signal pins on current five-row VME cards including P0. The general arrangement of a conduction cooled 6U VME card fitted with the proposed MultigigRT2 connectors is shown in Figure 1.
In order to preserve current VME chassis technology, including conduction-cooled air transport rack chassis, VITA 46 continues with the 6U and 3U Eurocard formats on 0.8-inch pitch, defined by the IEEE 1101.10 and IEEE 1101.2 standards. It is anticipated that existing chassis can be retrofitted from the VME64 standard to VITA 46 just by replacement of the backplane assembly, thereby allowing a quick and affordable upgrade to the new standard. It is also anticipated that during the introductory phase of VITA 46 products there will be hybrid systems that incorporate VME64 and VITA 46 on the same backplane. A hybrid backplane with six VME64 slots, and fourteen VITA 46 slots is shown in Figure 2.
VITA 46 is still in the developmental phase, as further connector choice evaluation is required. Characterization of the connector for VMEbus operation and interoperability with current VME64 implementations is yet to be completed, and similarly so for the backplane PCI bus. However, confidence is high, and typical scenarios have been developed for how the new connector may be assigned to various functions. One such scenario is described below:
- 32 pairs for a switched interconnect, supporting four ports
- 64 pairs for PMC I/O (two PMCs)
- 20 pairs for XMC I/O
- 116 pins for VME using both differential and single signal contacts
- 10 pins for utility signals, intelligent platform management interface, etc.
- 76 pins for general purpose I/O
This example uses all 432 pins of the new proposed connector, and it clearly illustrates the inadequacy of the current 335 pins of VME64 as this mapping does not even include a PCI backplane bus. With the increased functional density of currently developed devices (e.g., VME SBCs), it is easy to foresee the full use of all 76 general purpose I/O pins shown in the above example.
3U VME format
The 3U VME format is also addressed by VITA 46, giving it a new lease of life in potential conduction cooled and rugged applications where backplane I/O is a requirement. The MulitgigRT2 connector will support 80 differential pairs, in addition to 20 single signal pins. This will be enough for useful functionality through the backplane, supporting a single PMC/XMC site on a 3U SBC, and offering additional high- speed switched fabric connectivity as well. The concept of a PMC site on a 3U conduction cooled board is now well-established with a number of examples of CompactPCI products already available from such vendors as Dy 4, Radstone, and SBS.
VITA 46 product introduction
With the level of industrial backing that is driving VITA 46, and the efforts being made within the working group, it is only a matter of time before a spate of new product announcements are made. VITA 46 is not destined to be another abstract engineering exercise without the support of an aggressive, competitive industrial base. A reasonable first prediction is that the hybrid backplane approach will be key to economically sustaining legacy VME systems, while at the same time introducing the latest switched fabric technologies.
However, the hybrid approach will rapidly disappear from new system designs, as SBCs will continue to become ever more complex as they evolve into complete single board compute and I/O engines. Finally, more use will be made of backplane PCI and carrier cards with PMC modules leaving these new systems as pure VITA 46 implementations. The next logical evolution will be to maintain all the infrastructure and ecosystem of VITA 46, finally relinquishing the VMEbus interconnect in favor of more complex I/O and fabrics.
This is why VME must make the transition to VITA 46. It is not VMEbus technology that sustains today’s vibrant market. It is the breadth and depth of products and end use applications, plus the ultra-competitive supplier base that is continually innovating and evolving to provide what the customer in this marketplace really wants: stability of supplier base, ever increasing performance, less risk, and lower cost.
For more information about VITA 46: www.vita.com
Duncan Young was educated in the UK and has worked in the defense industry for almost 40 years. Duncan was part of the management buyout team that formed Radstone Technology, and he initiated product development of the world’s first conduction-cooled VMEbus modules. He has also served on a number of standardization committees. Duncan is now an independent consultant, but writes this column on behalf of Curtiss Wright Controls Embedded Computing (CWCEC).