Does the industrial world really need another Ethernet-based bus? The SERCOS vendor community thinks so, and has responded to the increasing market demands for Ethernet by developing a third generation of the SERCOS interface, dubbed SERCOS-III, combining the best features of Ethernet and SERCOS.
Since its inception in the 1980s, the SERCOS interface has built support as a standardized (IEC61491) means of communication between motion controllers and digital drives, I/O and sensors. More than 50 control manufacturers and 30 drive manufacturers worldwide offer SERCOS interface products, with hundreds of thousands of systems installed, including digital servo/vector/stepping drives, I/O modules and, more recently, hydraulic drives. The current generation of this highly deterministic interface operates at 8/16 Mbit/sec. with very low overhead and low jitter, more than adequate for most of today's needs.
Why go to Ethernet, then? The answer is cost and speed. The first and second-generation SERCOS interface has been most effective in drives 1 kW and above. The cost of its optical components has limited its adoption in the large segment of the market involving lower power drives. And there is a market demand,sometimes more perceived than real,for even higher communication speeds.
The problem was the differing requirements for networking information and motion control. Industrial Ethernet is characterized by the use of high bandwidth to move large amounts of data at low hardware costs, but it is not deterministic. On the other hand, the SERCOS interface is optimized for high-speed, deterministic motion control involving the exact synchronization of multiple drives. It offers a set of more than 500 standardized parameters that describe the interplay of drives and controls in terms independent of manufacturer. It is currently based on a higher-cost fiberoptic ring topology.
The challenge then was to create the best of both worlds and combine the high bandwidth and low cost of the non-deterministic Ethernet with the deterministic performance and standardized parameters of SERCOS interface.
Many technologies have been added to the Ethernet stack simply by adding a new protocol. However, putting motion under TCP/IP (transmission control protocol/Internet protocol), results in less determinism and the added expense of switching devices.
Instead, the SERCOS-III solution places the standard Ethernet TCP/IP under control of the motion bus and uses the lower-cost Ethernet hardware and twisted-pair copper cable. This maintains the deterministic motion control of SERCOS interface, allows links to the existing manufacturing communications infrastructure, provides for the possibility of new features and lowers hardware costs. Minimum cycle time is halved from the current 62.5 μsec. to 31.25 μsec.
SERCOS-III offers the added benefits of providing integration of IP protocols, new cross communication between slaves, synchronization of several motion controls, line or ring topology, and fault tolerance in case of a break in the ring. Ethernet's twisted-pair cable, combined with its differential driver and receiver, allows SERCOS-III to retain its notable noise immunity. Ethernet fiberoptics will, however, be supported by SERCOS-III.
Instead of the current SERCON816 ASIC, a lower-cost standard module such as a field-programmable gate array (FPGA) communication controller is used to handle all sequencing and synchronization tasks, relieving the host microprocessor of these time-intensive functions. Because a more-flexible hardware solution is implemented in SERCOS-III, the development of a SERCOS-III core (SERCOS-III IP) is necessary. With this, manufacturers of components and systems can combine the SERCOS-III hardware and their own logic components in one common FPGA.
In addition, there are plans to integrate the SERCOS core into a general-purpose communication controller (GPCC), which can support various industrial Ethernet protocols. Thus, it is possible to implement control and drive devices that can easily be adjusted to the respective Ethernet protocol simply by using the appropriate driver software. This is a huge advantage, because industrial OEMs now need to handle only one type of wiring. The OEM's customers profit from not needing different hardware configurations if they use several different industrial Ethernet protocols in his plant.
Thus, SERCOS-III lowers cost, maintains determinism and provides transmission speed to 100 Mbit/sec. It also protects vendor/user investment due to high compatibility with the previous SERCOS interface (topology, profiles, telegram structures, synchronization and standardized parameters).
So, does the industrial world need another Ethernet-based bus? When it's this much of a step forward, we think the answer is clearly "Yes!"
The first prototypes of the SERCOS-III hardware will be available this year, with first products envisioned for market in 2005.
