The MicroTCA specification was designed to provide industry with off-the-shelf, modular, and hot-swappable multi-device systems. MicroTCA is intended to be more affordable and less bulky than the previous AdvancedTCA standard. In this blog post, I describe the main elements of a typical MicroTCA system.
The typical MicroTCA chassis offers space for multiple advanced mezzanine cards (AMC), MicroTCA carrier hubs (MCH), power modules, and cooling units (fans).
Advanced mezzanine card (AMC)
AMCs are electronic devices that follow the PCI Industrial Computers Manufacturers Group (PICMG) specification, which define the card’s form factor, power supply, and data interfaces. The specification defines card widths of 73.5 mm (single-width) and 148.5 mm (double-width), which makes them compatible with MicroTCA.
AMC devices range from FPGA cards (populated with FGPA mezzanine card (FMC) sites) to complete CPU cards and SATA hard disk drive storage devices. The MicroTCA chassis provides 12V and 3.3V to each cards.
The AMC specification defines the following interfaces on the AMC connector:
MicroTCA carrier hub (MCH)
A MicroTCA chassis has space for one or two MCH modules. The MCH’s main purpose is fabric switching and platform management:
- Fabric switching: The MCH handles the switching of the AMC’s high-speed ports. An MCH can handle one of the two fat-pipe port blocks (ports 4–7 or 8–11), connecting all AMCs together in a star pattern. For example, an MCH with a PCI Express switch enables PCIe interconnection between all the AMCs within the chassis. The MCH can support PCI Express, Serial RapidIO, and XAUI protocols. It also handles the switching of ports 0 or 1, which provide Gigabit Ethernet connections to all AMCs.
The MCH typically enables external equipment to connect to the AMCs within the chassis by providing connectors on its front-panels (e.g. RJ-45 or SFP+).
- Platform management: The MCH provides an Intelligent Platform Management Interface (IPMI). The AMCs and other chassis devices like the cooling units and alarm panels auto-negotiate their power requirements and functionalities with the MCH. The MCH in turn allocates power to the units and monitors their sensors to protect the system.
The MicroTCA power module, controlled by the MCH, can allocate power independently to all the MicroTCA chassis slots. Two power modules can be used within a chassis for redundancy.
The following diagram illustrates the interconnections within a typical MicroTCA system. For the sake of simplicity, the described system is comprised of 3 FPGA cards, 1 CPU card, 1 storage device, and 1 PCI Express MCH.
In this system, the MCH provides both Gigabit Ethernet and PCI Express switching between the CPU and the FPGA AMCs, as well as the external world. Since the MCH is in the MCH1 slot, only fat pipes 4–7 are used for PCI Express. The chassis backplane directly routes the SATA port between AMC slots 1 and 2 to connect the SSD to the CPU card.
In my next blog post, I will complete my overview of MicroTCA systems by discussing the MicroTCA.4 specification, which is specifically designed for physics applications.