Faster cycle time for PROFINET

PROFINET has always been a very fast communication system. PROFINET real-time communication (RT) uses the prioritisation methods of Ethernet and can therefore be implemented on standard Ethernet controllers. With a data rate of 100Mbps and full-duplex transmission, bus update times that are faster by several factors compared with today's fieldbuses are possible. As a result, RT is usually more than sufficient for typical factory automation applications. But, once an application's requirements include the need to synchronise nodes to within 10-100us or to form a highly dynamic control loop via the bus, additional measures become necessary. The highly accurate isochronous real-time (IRT) synchronisation process was developed for this purpose. This process enables Ethernet transmission delay times of differing and fluctuating lengths to be eliminated.

In addition, there are also specific applications that will place even more stringent requirements on communication performance in the future. That is why mechanisms were defined in PROFINET Specification V2.3 that make communication with PROFINET even faster. With Fast Forwarding, Dynamic Frame Packing, and Fragmentation, it is possible to achieve short cycle times as fast as 31.25us, together with high-precision isochronous operation. Thanks to this feature and its openness, PROFINET is further cementing its position as the leading Industrial Ethernet standard.

So just how fast is a cycle time of 31.25us? To grasp just how fast this cycle time is, a special model was developed and was presented at Hanover Fair (April 2012, Germany). LEDs react too slowly, and only an oscilloscope is able to visualise the cycle time of 31.25us as well the slight jitter over the entire system. This cycle time makes PROFINET an interesting candidate for highly dynamic measuring equipment applications, since sampling rates up to 32kHz over the network are possible. In the model, an analogue signal was also sampled, transmitted via PROFINET, and evaluated at another station. In terms of real-world applications, such cycle times will be used in the future, for example, to record torque characteristics in test stands.

Isochronous operation was demonstrated with a traditional stroboscope test. For this, a stroboscope is aimed at a variable-speed disk in such a way that a permanent image of a written text is produced. This functionality will also be needed in the measuring equipment industry, as well as in other industries like the printing industry, where it will be used for multi-axis closed-loop control in printing machines. In the latter case, it will not only increase the productivity of the overall printing machine but will also allow production of printed products with higher-resolution - and thus sharper - images.

Undisturbed transmission of data

Openness is another highlight of PROFINET. Even with the short cycle time of 31.25us, standard data can be transmitted without limitation via TCP/IP. This was demonstrated in a full HD video that shows undisturbed transmission of these data all the way through the PROFINET system. The ability to transmit standard data is necessary in order, for example, to transfer new parameters, quality assurance data, or images for production monitoring.

Several different measures were implemented in PROFINET V2.3 to increase performance. The decision whether to forward a frame in the integrated switch of a device requires address information in the frame header. With Fast Forwarding, the FrameID (FID) address information is integrated once at the start of the frame header so that instead of having to wait for a large number of bytes it is possible for forwarding to take place early on. As a result, the typical delay times of 3-6us per node of today can be reduced to 1.2us.

The summation frame method for optimising the ratio of frame to user data is already known from the fieldbus world. In this case, the I/O data for several nodes on the network are integrated in one frame so that the frame header and trailer (FCS) only occupy usable bandwidth once. For nodes with only a few I/O bytes in particular, the usable bandwidth increases noticeably, because a minimum frame of 64 bytes is prescribed for Ethernet. Like other Ethernet systems, PROFINET also uses this method and, in so doing, opens up further potential for optimisation. In contrast to ring bus systems, PROFINET relies on the performance advantages of a full duplex system, i.e. input and output data are sent simultaneously on the 2-pair cable. When a single summation frame is used, this would have to be sent, received, and checked completely down to the last node, including the checksums. This is where Dynamic Frame Packing comes in. Because the data of the first nodes in the line are not relevant for the nodes placed further at the end, these are removed during the passage. This shorts the frame in its passage through the network. The time-determining arrival of the frame at the last node is thus must sooner, thereby significantly reducing the overall update time for all nodes.

Open TCP/IP communication included in 31.25us cycle time

A proven and important advantage of PROFINET is its unlimited TCP/IP communication even when IRT communication is occurring simultaneously. For this, the architecture of PROFINET provides for time scheduling in addition to synchronisation. The network is not loaded with I/O frames during a defined time phase but instead is free for any TCP/IP frames, which can take up a duration of up to 125us with Fast Ethernet and thus define the minimum cycle time.

The Fragmentation function defined in PROFINET V2.3 takes large TCP/IP frames in the individual nodes and divides them into smaller individual parts prior to sending. These fragments are then sent in consecutive cycles. The counterpart then re-assembles them so that the upper-level application layer receives an unaltered TCP/IP frame. This allows users to realise bus cycles of 31.25us with shared I/O and TCP/IP communication, without having to reduce the available bandwidth for the TCP/IP communication.

In contrast to other Ethernet systems, the integration of fragmentation in the switch module of a device means that PROFINET does not require additional special device modules, which would increase costs and cause an inflexible location specification. Likewise, PROFINET does not require any additional computing power or network controller availability.

Future ready

By necessity, these performance optimisations in the microseconds range require integration in hardware. New generations of PROFINET modules will implement all these new functions in hardware. No other standard allows such high IO communication performance while simultaneously ensuring high availability of bandwidth for standard communication. Accordingly, various technology suppliers will offer easy-to-integrate systems in the form of ASICs, network controllers or FPGAs, and thus provide device manufacturers with the basis for producing high-performance approaches that meet customer requirements.

If more stringent requirements for automation system components arise in the future and require implementation in products, PROFINET has already made the necessary advancements and will provide the reserved performance for these future innovations. The user can thus choose a system today, without any consideration of the performance that may be needed in the future.

Performance isn't the only thing that counts, however. Without a convincing overall package that includes real-world diagnostics, integration, safety, and wireless systems, the best performance is of no use. Thanks to its widespread use around the world, the dedicated commitment of its users, and its outstanding underlying quality control and certification activities, PROFINET provides both maximum performance and a future-proof concept.

For further details about PROFINET go to www.uk.profibus.com.