Microchip has announced three new Ethernet switches targeted at industrial applications: the LAN9694, LAN9696, and LAN9698.
The key difference between these three is the total bandwidth capability of each switch: the LAN9698 is a 102 G Ethernet switch, while the LAN9696 and LAN9694 are 66 G and 46 G, respectively.
The LAN969x switches target industrial applications. Image used courtesy of Microchip
All are equipped with a single 1 GHz Arm Cortex-A53 CPU. Arm touts the 64-bit A53 as the “most widely used low-power processor” capable of high-performance tasks in a power-constrained environment. The Cortex-A53 uses the Arm Neon SIMD architecture for vector processing. In addition to using the Cortex-A53, all three LAN969x switches can support a combination of 1 G, 2.5 G, 5 G, and 10 G Ethernet speeds.
Time-Sensitive Networking for Industrial Applications
Microchip designed the LAN969x devices for industrial automation because they leverage deterministic communication and time-sensitive networking (TSN) technology in addition to redundancy techniques for increased dependability.
TSN is the IEEE 802.1Q standard that enables deterministic, timely communication in Ethernet through a time-scheduling algorithm. According to Cisco, a TSN solution involves two main types of devices: end devices that run customer applications and bridge devices, which are Ethernet switches such as the LAN969x.
Time-sensitive network topology. Image used courtesy of Cisco
TSN uses a central network controller (CNC) application running on a bridge device to discover the physical network topology and determine if a user-defined time schedule for Ethernet frame transmission is feasible. If it is, TSN-Ethernet frames are transmitted between devices in a TSN flow, which must meet strict time requirements. Non-TSN frames are transmitted on a best-effort basis.
Configurable for Two Redundancy Protocols
The LAN969x family can also be configured for two types of redundancy protocols: high-availability seamless redundancy (HSR) and parallel redundancy protocol (PRP).
HSR is a cost-effective redundancy protocol that uses a ring topology to send duplicate Ethernet frames on two different, opposite-direction paths from a source node to a destination node. The destination node accepts the first of the duplicate frames it receives. The destination nodes are referred to as doubly attached nodes (DANH). The ring requires a redundancy box to connect it to outside, non-DANH devices such as a desktop.
Duplicating frames ensures that if one path fails, the destination node will still receive the frame. The HSR protocol provides instant recovery from any failure of a single network component.
HSR uses a ring topology to achieve redundancy. Image used courtesy of iGrid T&D
PRP does not use a ring topology. It transmits duplicate frames on two separate paths to two different local area networks (LANs). Attached to each LAN are single-attached end nodes (SANs). As a result, each end node does not need to be connected to the topology through a redundancy box. Both LANs can then be interconnected through a double-attached node.
PRP sends duplicate frames to two separate LANs to achieve redundancy. Image used courtesy of iGrid T&D
Because the LAN969x family supports either type of redundancy, customers have more flexible design options for a wide variety of industrial applications.