ANSI/VITA 74.0-2017 was referenced within the Sensor Open Systems Architecture (SOSA) Technical Standard Version 1.0 in the fall of 2021 and the reference is proposed to be updated to VITA 90 in the upcoming SOSA 2.0 revision due in late 2022. A typical VITA 74 and VITA 90 module in its simplest form is approximately the size of a deck of ordinary playing cards (Figure 1.). Both have an electronic architecture similar to VITA 65 OpenVPX. For the last several years, there has been a lot of work done behind the scenes to transform VITA 74 into VITA 90 to meet the technical attributes of SOSA.
|Figure 1 | Typical VITA 74.0/VITA 90.0 19-mm baseline configuration module. Photo: Trident Infosol.|
Design considerations for VITA 74.0, VNX
Going back in time, the technology behind this SFF standard was initially developed by engineers at Themis Computer (now Mercury Systems). Building on the successful performance of the ANSI/VITA 57.1 FMC standard, engineers at Themis chose to utilize the Samtec high-speed, high-density SEARAY connector as its primary module-to-backplane interface, as well as the interface between the backplane and the front panel I/O transition board (IOTB). The right-angle female SEARAY connector was selected to be used on the module, and the corresponding straight male SEARAY connector was selected to be used on the backplane. This original NanoATR design defined two module sizes; a 19-mm module utilizing the 400-pin SEARAY, and a 12.5 mm module utilizing the 200-pin SEARAY.
The NanoATR concept was brought to the VITA Standards Organization (VSO) to be considered for a standard describing SFF computer and payload modules to be used in embedded systems targeted towards medium-sized unmanned aerial vehicle (UAV) and other rugged aerospace applications. The proposal gained both traction and sponsors, becoming VITA 74, one of three potential SFF standards being considered at the time: VITA 73, VITA 74, and VITA 75. Of the three draft proposed standards, only the VITA 74 working group completed the work and received public consensus approval to be published as ANSI/VITA 74.0-2017. The standard was given the moniker VNX, a name befitting the unofficial “Nano” derivative of VPX.
From its inception, VNX was never intended to replace VPX, but was instead meant to bring the essential tenets of the VPX architecture into rugged airborne, space, and ground platforms physically too small to accommodate a VPX system. A typical 4-slot avionics display processor system based on the original ANSI/VITA 74.0-2017 standard is shown in Figure 2.
|Figure 2 | VITA 74 image processor with optical MT interconnects. Photo: Elbit America.|
The original VNX standard was designed and optimized for rugged military/aerospace compute-intensive applications, such as small mission computers, display processors, MEMS inertial measurement units (IMUs), and I/O data concentrator applications. Early systems contained one or more single-board computers (SBCs) which generally employed Intel Architecture (IA) processors such as Intel’s Atom and AMD’s G-Series system-on-chip (SoC), both with integrated graphics and relatively low-power processors. These first-generation VNX systems often utilized a commercial off-the-shelf (COTS) computer-on-module (COM) mezzanine computer, I/O modules, and Flash storage mounted on VNX carriers hosting COM Express Type 2 and Type 10 computer modules, Mini PCI Express (mPCIe) I/O modules, and mSATA memory modules. The VNX standard fits well within the U.S. Department of Defense (DoD) Modular Open Systems Approach (MOSA) strategy, which was established to achieve competitive and affordable acquisition and sustainment over the life cycle of a system. A standard in its own right, VNX achieves high levels of future-proofing precisely because it often employs other relevant standards in building its family of modules and assemblies.
The ANSI/VITA 74.0-2017 base standard was released with a list of future dot-standards that were to be designed and implemented to optimize complex VNX system I/O solutions for future applications. Two of these dot-standards would add the concept of backplane “connector modules” required to facilitate high-speed / high-bandwidth optical and coaxial RF/video data transmission for both inter-slot and intra-system digital signaling. Another dot-standard would define options to include wedgelocks, address electromechanical considerations for single-module deployments, and define techniques for high-efficiency/high-power VNX module and system cooling.
Similarly, another dot-standard-defined VNX-specific power supply modules, VITA 74.4 – also known as SpaceVNX – documents specific electronic and mechanical considerations required to implement rad-hard and rad-tolerant VNX solutions in small spacecraft applications. In the process of developing these dot-standards, the VITA 74 Technical Committee reviewed the government and industry’s evolving module performance requirements and use cases and decided to go all in and fully optimize the original VITA 74.0 VNX compute and I/O module’s pin assignments for maximum signal integrity; the committee also implemented the OpenVPX-style utility plane, control plane, data plane, expansion plane, and I/O overlays.
Expanding VNX for high-speed fabrics and coaxial/optical signaling
The evolution of VNX from a simple module standard to a family of standards that will describe the technology required to assemble complex SFF systems has been driven by military and aerospace adopters of the ubiquitous VITA 65 OpenVPX family of standards within the VITA, SOSA, and Hardware Open Systems Technologies (HOST) communities. As a result of evolutionary changes to the SEARAY connector pin assignments, the signal integrity of all high-speed channels has noticeably improved, and the revamped planes can now exceed the signaling requirements necessary to support PCI Express 4.0 and other modern fabrics. As companions to the SEARAY connector, new VNX connector modules have been specifically designed and manufactured to support various combinations of high-speed signaling: fiber-optic connectivity using MT ferrules, coaxial connectivity for RF and video signals, and isolated copper contacts for applications such as providing high voltage to an RF amplifier’s power supply rails. An example connector module that hosts two 24-fiber MT ferrules, 16 #20 silos for Samtec’s 50-ohm RF edge-mount (module) or in-line (backplane) coaxial contacts, and four #16 silos for either Samtec’s future in-line (backplane and module) 50-ohm RF, 75-ohm video, or 20-gauge copper contacts, is shown in Figure 3.
These evolutionary changes were deemed to be such a revolutionary improvement over the capabilities of the ANSI/VITA-74.0 VNX base-standard, that both VITA and SOSA management decided that the new capabilities should be codified in a new family of standards, VITA 90, which is called VNX+.
|Figure 3 | VITA 90 VNX+ module basecard with 240-pin connector and full connector module.|
VNX+ not only includes Samtec’s 400-pin SEARAY connector as defined in the VITA 90.0 base draft proposed standard, but also includes new additions necessary to support use cases such as signal processors, radio transceivers, graphics processors, network and fabric switches, and other I/O modules requiring coaxial or optical MT signaling. These applications will use the 240-pin SEARAY high-speed data connector with a full-connector module, or the 320-Pin SEARAY connector with a half-connector module as defined in the VITA 90.2 dot-standard. VITA 90.3 introduces the VNX+ energy conversion module (power supply) using a 320-pin SEARAY connector centered in a VNX+ slot, plus an energy storage module (hold-up capacitor and/or battery) using a 240-pin SEARAY connector, similarly centered in a VNX+ slot.
With careful engineering of the mechanical and thermal considerations, VNX+ modules can fit in many tight spaces using conventional backplanes or micro-backplanes for single module deployments. For specific applications, additional space may be recovered by employing an equivalent cabled backplane.
Pod-mounted sensor processor deployments
Considering the tight space inside a 5-inch tube (like an AIM-9 Sidewinder-sized pod), previous systems used expensive custom hardware that was hard to upgrade as threats and requirements changed. Current thinking in the user communities requires the ability to upgrade or modernize hardware to keep up with the ever-evolving threats; that requirement drives them to specify and procure standard modules with known hardware and software interfaces.
Military and aerospace integrators have discovered that the VNX+ SWaP-C attributes make it a natural fit for not only SFF traditionally packaged ATR-style avionics boxes, but also for pod-mounted sensor and weapons systems requiring high-performance sensor interfaces in close proximity to FPGA and MPSoC signal processors, computers, radios, and platform I/O available as COTS/modified COTS MOSA modules with standardized electromechanical backplane interfaces. VNX+ is the only backplane-centric COTS MOSA implementation that can be deployed as vertically oriented conduction-cooled modules on a traditional horizontal backplane, mounted longitudinally (i.e., along the long axis), within the usable area of a 5-inch-diameter tube as used for an AIM-9X Sidewinder sized sensor pod, or within a 6-inch-diameter tube as used for the Coyote, as well as other similarly small payload pods or UAV fuselages. (Figure 4.)
|Figure 4 | VNX+ example system in Coyote & AIM-9X pod.|
Space use case considerations
VNX is also being designed into space applications. SpaceVNX (VITA 74.4) and SpaceVNX+ (VITA 90.5) were designed from VNX’s inception to target smallsats, including 1U to 12U CubeSats, and space rovers for critical applications that require high levels of computing and data-transfer performance. SpaceVNX+ follows the high-reliability philosophy of SpaceVPX but considers SWaP-C constraints related to smallsats-based missions.
The SpaceVNX+ in-band protocols such as SpaceFibre, SpaceWire, and Serial RapidIO and in-development high-performance compute modules will enable future missions such as multiple spacecraft flying in formation to create unprecedented telescope and interferometers for imaging fainter, smaller, and more distant objects. Swarms of SpaceVNX+ assets will enable complex and time-varying networks of spacecraft and sensors that are capable of sharing rich, near-real-time streams of information to enhance Space Situational Awareness and Autonomous Operations.
Utilizing an available rad-hard electrical power supply (EPS), a rad-hard system controller/monitor module along with various COTS, modified COTS, or rad-tolerant sensor, processor or I/O modules, lower-cost rad-tolerant systems are being developed by several organizations. (Figure 5.)
|Figure 5 | 1U cubesat with SpaceVNX modules.|
VNX+ thermal performance targets
When VITA 74 was first released, the designers of the VNX standard assumed that the practical power limit for the 19-mm module should be arbitrarily pegged at a conservative 20 watts. As time went on and integrators started building actual systems in ever-smaller platforms, performance demands inevitably increased, and the use of one or more 20 to 25 watt processor modules became commonplace. These systems employed heat sinks consisting of nominally sized heat-radiating fins coupled with minimal airflow. Testing proved that well-designed VNX systems could be qualified to meet typical military and aerospace environmental requirements.
But how small is small enough? Demands to fit in ever-tighter spaces – coupled with faster and wider data paths between sensors, signal processors, and compute modules – has made it necessary for VNX+ to support higher-power components and even denser module packaging. The power rails and grounds are enhanced in the new draft proposed standard. New options include a legacy balanced 3.3 volt, 5 volt, and 12 volt power system or a new unbalanced “12 volt Heavy” power system. The VNX+ “Energy Conversion Module” (ECM) dot-standard is proposed to allow up to three load-sharing power modules to be used in a system, with a companion energy-storage module providing input power dropout and transient protection. Some individual ECMs are capable of delivering well over 100 watts to the payload modules, but the ECM’s output is purposely derated to ensure that the individual supply’s maximum ratings are never exceeded. The integrated system must use appropriate cooling technologies for its specified environment.
To understand the thermal limits of individual modules constructed using differing cooling strategies, studies are being conducted to define the upper power-dissipation boundaries of practical VNX+ signal-processing modules. One study modeled three power dissipation levels – 35 watt (low power), 55 watt (medium power), and 95 watt (high power) – with initial results showing that the “all-aluminum” module shells with a simple thermal interface will result in unmanageable temperature rises at all specified power limits. The moderately advanced thermal mitigation techniques, using copper as the thermal conductor, limit the temperature rise to manageable levels for modules dissipating up to 50 to 60 watts. The highly advanced thermal-mitigation techniques show significant improvements in thermal transport, showing a near-20 °C improvement in temperature rise for modules dissipating up to 95 watts.
The example cited uses two nearly perfect heat sinks assumed to be in very good contact with two of the three available module cooling faces; i.e., using the two module side faces, but not the handle face. In practice, optimal results would be achieved by using all three cooling faces and minimizing the thermal resistance between the VNX module’s cooling faces and the corresponding heat sinks. The type and quality of the selected heat sink will vary depending on the space available and the system’s thermal-performance requirements. A practical system could be cooled using a highly efficient heat exchanger or cold plate, in contact with an optimized thermal-transfer material in contact with all three VNX module cooling faces. The thermal-transfer material selected can range from a layer of compressible thermal interface material (TIM) to a thin layer of highly efficient thermal grease.
Considering the mechanical envelope studies discussed earlier, coupled with a little algebraic extrapolation, it appears that a system requiring a reasonable number of high-power SFF modules, each dissipating 60 to 80 watts or more, which must be constrained in a very small package (such as a 5-inch diameter sensor pod), may now be built on a standards-based architecture using practical advanced cooling technologies, such as oscillating heat pipes (OHP), to carry the heat away from the module to the dissipating thermal interface. This technology has been proven its ability to transport heat with a small temperature gradient, and if properly positioned, to essentially create an isothermal interface to designated heat-transfer locations.
Use of OHP technology should allow multiple modules to each spread their thermal load evenly across their primary thermal interfaces, with a large number of VNX+ modules oriented vertically along the long axis of a small-diameter sensor pod, with the internal chassis cooled by conduction, ram air, fan-forced air, or a liquid-cooled heat exchanger. As the VNX+ thermal envelope is expanded, the advanced cooling technologies will be thoroughly characterized in a real-world VNX+ modular application. These high-power use cases are being verified through continued modeling and laboratory testing. Test results and use case examples showing the expanded limits of the VNX+ power dissipation envelope will be published.
Modules available or coming soon in the VNX and VNX+ family of standards
Many modules with functionality required for compute, signal processing, communications, and I/O are available or in process. Examples of modules that exist today, are currently evolving from existing VNX designs, or are new designs in-work include:
- Power-conversion and energy-storage modules for low-, medium-, and high-power systems
- IA compute modules with up to 11th-gen dual-core Intel Core i7 processors (formerly Tiger Lake), and current generation quad-core Intel Atom processors
- NVIDIA GPGPU modules using the Jetson AGX Xavier GPU processor
- FPGA modules using various SOCs and MPSoCs with Arm cores, using high-speed copper and optical backplane interfaces
- RF transceiver modules using RFSoC, as well as MPSoC and companion transceiver
- Rad-hard controller modules for space applications
- I/O modules for MIL-STD-1553, ARINC 429, and MIL-1394B/AS5643 data buses
- I/O modules with RS-232/422/485, CAN, and Gigabit Ethernet interfaces
- Gigabit Ethernet switches with L2/L3 and 10 GbE uplink ports
- Storage modules using mSATA SSDs
- MEMS IMUs with GPS
As the military and aerospace communities move to smaller, more intelligent platforms, the requirement to build smaller systems becomes ever more important. To get more “bang for the buck,” there is an enlarging paradigm shift away from custom electronics towards COTS and modified COTS solutions. To minimize the effort required to upgrade systems through the use of common hardware, communications, and control interfaces, it is necessary to build hardware that can be conformant with MOSA standards from VITA, SOSA, HOST, and other standards-development organizations. VNX+ is being designed and implemented with all of these requirements in mind.