Small Form Factor (SFF) embedded systems have been gaining popularity due to their small size, weight, and power (SWaP) advantages. The SFF architectures are typically purpose-built, providing just a few specific functions. But, the performance level is often limited and the versatility and re-usability across multiple applications is typically very low. Is a balance available where there is high performance and versatility in an open standard architecture, but a compact size that is optimized for SWaP-C (C stands for cost)?
There are many SFF solutions in the market, but many are not based on an open standard. Why is this significant?
The critical importance of open standards is often overlooked, but there are many benefits to a Modular Open Standard Architecture (MOSA) design. One of the key reasons is to reduce the risks posed by a single source technology and obsolescence. With several vendors in the industry supporting an open standard, the design engineer is not putting all of their eggs in one basket. The Mil/Aero community has learned the lessons of choosing a server blade approach from a Fortune 100 corporation and getting locked into a single vendor solution. It backfired when that corporation sold their business unit to the Chinese. Had the program gone to a MOSA solution, even if their prime supplier sold the business, they would still have had many other vendors to choose from and an open specification available. Additionally, with competition between the various MOSA hardware vendors, the constant drive to innovate, upgrade/improve, and reduce costs is prevalent.
Finally, the scalability of open standards is a key element. There is tremendous cost and time savings in upgrading on a scalable architecture versus a stand-alone, purpose built approach, which may not leverage much in forward compatibility.
What is SFF?
What size do we consider SFF? Comparing it to the traditional VME and CompactPCI form factors of Eurocard (3U and 6U) may be the best start. Those are the most common sizes for legacy open-standard architecture equipment in the embedded market. A 3U Eurocard board is 100 mm high (133.5 mm including subrack) by 160 mm deep, and a 6U is 233.4 mm (266.70 mm with subrack) × 160 mm. Nearly half that size is MicroTCA, a high-performance embedded architecture that offers significantly more performance than most SFF systems. MicroTCA is 75 mm by 180 mm in the single module size and 150 mm × 180 mm in the double module size. See Figure 1 for size comparison of MicroTCA to Eurocard.
Comparing Open Standard Architectures
AdvancedTCA is often used for Telco applications where massive throughput is required and size, power, and cooling limits are much more flexible. The modules are a large 8U (355.6 mm x 280 mm) size with a 30.48 mm backplane pitch. They typically have dual socket high-end processors that require more board space and cooling. The other area for AdvancedTCA is Mil/Aero where the processing power outweighs the architecture’s SWaP limitations.
MicroTCA is an excellent fit for a wide range of applications because of its high performance-to-size ratio and versatility. The COTS (Commercial Off The Shelf) architecture has benefits over other form factors due to its robust system management, e-keying, and highreliability features. In addition, the open-standard COTS architecture is typically more interoperable and lower cost than competing standards.
VPX is mostly a MOTS (Military Off The Shelf) architecture, serving one market. The architecture was billed as the “next VME”, although ironically it is not standardly backwards-compatible. It does, however, share the Eurocard form factor. Like MicroTCA, it can easily be ruggedized for hardened applications. Figure 2 takes a closer look at MicroTCA, VPX, and one of the open standard SFF form factors.
As shown in Figure 2, MicroTCA modules are close to half the size, weight, and cost of VPX. For the power, it obviously depends on the application. But, MicroTCA systems often have lower power consumption as well.
The SFF specification shown in the chart is roughly the same board height as MicroTCA, but about half the depth and a thinner pitch. But, there is a big tradeoff for the small size and architecture – lower computing performance. With a 20W limit for even the larger SFF conduction-cooled modules, that limits the processor selection to mostly single core, lower-performing Atom™ and G-Series ™ types of chipsets. For graphics modules, many of the upper-end chipsets will exceed 30W. Most higher-performance FPGAs such as Stratix-IV, Stratix-V, Virtex-6, Virtex-7 and digital converters are also above the threshold for the open standard SFF.
Now, purpose-built SFF systems are available that can use some 3rd Generation Core i7 processors, but you are losing the many advantages of an open standard architecture and its vast ecosystem.
Pico Chassis Versions of MicroTCA
Although MicroTCA is typically 19” rackmount, there are Pico style and other shorter width enclosures. Pico shelves are a standard option in the core MicroTCA. 0 specification. Fig 3a shows an example of a 2-slot Pico chassis with two AMCs (chassis courtesy of partner for this product, Schroff). This chassis is approximately 10" wide for dimensions of 44.5 mm high × 250 mm × 320 mm deep. Often, a MicroTCA Carrier Hub (MCH) is not required in Pico units. The chassis utilizes an active backplane with fan speed control circuitry that is triggered by temperature sensors in the chassis. The IPMI connections are routed to both AMC slots.
Many of the SFF applications require a conduction cooled enclosure. Similar to the previous enclosure, the conduction- cooled version in Fig 3b also can hold 2 slots.
One of the advantages of utilizing a smaller MicroTCA system is the ability to leverage hundreds of standard AdvancedMCs (AMCs) that are readily available in the marketplace. This includes dozens of options of modules that are geared for:
- Processing, FPGAs
- A/D & D/A Conversion, Transceivers
- I/O, Network Interface
- Carriers (FMC, XMC, PMC, etc)
MIL/Aero applications can be a driver for leading-edge performance, but often the long design cycles and conservative approach can slow down the innovation. An advantage of a truly COTS architecture serving multiple markets is that the advances contrived from other markets such as communications and High Energy Particle Physics can be leveraged. For example, the physics community requires very high performance digitizers, with more channel options, and various RF boards for LLRF, beam position monitoring, etc. The technology can be leveraged to highend radar and radar jamming applications drive powerful DAQ, A/D and D/A converters. Today, there are AMCs (and FMCs) that offer a DAC 14-bit @ 5.7 GSPS and ADC 14-bit @ 4.0 GSPS. In communications applications, the push for faster networking has driven 100G processors and FPGAs.
An example of a high-performance density system is a 1U rugged chassis that holds 6 AMCs. Figure 4 shows the chassis layout with an integrated MCH that provides GPS receiver/IEEE 1588/SyncE capabilities. The precision timing and time stamping features provide deadly accuracy for target acquisition. Ten years ago, a weapon system would have the RADAR as part of the system or a short distance away. In today’s network-centric warfare, the RADAR and weapon system can be many miles away. In sending and receiving data, by the time the system receives the input and is ready to analyze the data, the target is in a different location. With SyncE and IEEE 1588, the packet is time-stamped to the exact Grand Master Clock (GMC) time when it is sent. When the packet is received, the system can adjust the time differential and allow the algorithms to predict the location of the target. Therefore, the precision-timing of the system can help make targeting more accurate.
In just a 1U height one can have a powerful core i7 quad-core processor, four Virtex-7 FPGAs with removable 5.7 GSPS DACs and 4.0 GSPS ADCs, and a 12 Gbps SAS 800GB storage module with Advanced RAID and external expansion. Or, for more storage, a JBOD module can hold up to 8 mSATA chips at up to 1 TB each (for a total of 8 TB in one bay). The chassis is designed to meet MIL-STD-810F for shock and vibration, and MIL-STD-461E for EMI.
MicroTCA also offers a wealth of high-end storage, graphics, I/O, FPGA and other boards which make the ecosystem rich and diverse. Other specialty products in the MicroTCA form factor include a 72-core Tilera Processor, 40GbE chassis platforms, a PCIe Gen3 carrier with the ability to use the highest performance commercial NVIDIA or other graphics cards. MicroTCA also provides multiple ruggedization levels including MTCA.0 (Core specification), MTCA.1 (Rugged Air Cooled), MTCA.2 (Hardened Air/Conduction Cooled Hybrid, MTCA.3 (Hardened Conduction Cooled) and MTCA.4 (with RTMs for Physics, other applications).
SFF systems can be a great fit for applications requiring some specific functionality in a small space. But small doesn’t have to mean low performance or limited features/capabilities. As an “SFF-like” architecture, MicroTCA can provide massive performance in small spaces. There are a wealth of configurations and options to allow for SWaP performance across a wide spectrum of application requirements.
This article was written by Luis Teruel, Hardware Engineering Manager, VadaTech (Henderson, NV). For more information, Click Here .