HK1091561B - Method, interface and network for the cyclic transmission of ethernet telegrams - Google Patents

Description

Method, device and network for periodically sending Ethernet telegrams

Technical Field

The invention relates to a method, an interface and a network for periodically sending Ethernet information.

Background

Ethernet (Ethernet) is today the most common technology used to transmit data at a rate of up to 100 megabits per second (mbps) in a regional communication network, a "local area network" (LAN). LANs are local communication networks that are confined to a geographic area and are constructed from one of a plurality of servers and workstations "" nodes "", which are connected via a transmission link, such as a coaxial glass fiber or twisted pair cable. There are many network topologies that can be used for LANs, the most well known of which are bus, ring, star, or tree architectures.

LANs operate with a network operating system and a standard network protocol. Ethernet is one possible network protocol and in this example supports a number of communication protocols, such as the TCP/IP protocol or the IPX protocol. The international reference model for network data transmission, the OSI layer model, consists of a stack of layers comprising seven layers, each layer having a set of protocols defined for that layer, each protocol providing its services to a second, higher layer, and ethernet is associated with that second layer, the "data link layer". This data link layer aggregates the data to be transmitted to form information with specific information added due to the respective communication protocol. The data link layer has the responsibility of transporting data information between nodes and error detection within the network.

In the case of the ethernet concept, the data link layer is divided into two levels, where the first level adds a header section to the data, the "start identifier", which contains the information required by the receiver protocol to modify the data transmission. Within the second level of the ethernet protocol, the data information is then encapsulated with an additional preamble and a final section, the "checksum", for transport between nodes. With such ethernet messages it is possible to transmit data of a length of up to 1500 bit groups, which is necessary to follow a fixed break time between the individual ethernet messages.

Sending and receiving ethernet messages over an ethernet transport link is the responsibility of an ethernet controller, also known as a Media Access Controller (MAC), which switches between the node and the ethernet transport link and is connected to the node via a bus system. Such ethernet controllers are typically controlled by a software driver, which is incorporated into the node's corresponding operating system. The ethernet controller typically includes a transmit shift register and a receive shift register to decouple the ethernet transmission link from the physical memory of the node. Modern ethernet controllers also typically have an opportunity to directly access the physical memory of the node, the "direct memory access" (DMA), for which reason software drivers within the node's operating system have time-saving aspects of storing ethernet information to be sent and received directly in or retrieving such information from the node's memory.

The ethernet protocol is mainly used in office communication networks. The benefits of the ethernet concept when using standard hardware and software components also the opportunity to achieve high data transmission rates with simple networking technologies means that ethernet communication networks are increasingly required to be used in industrial production for data exchange between stations. However, when the ethernet protocol is used in automation technology, additional complex hardware and/or software techniques must be used to ensure the real-time capability of ethernet data transmission. In controlling a machine, it is generally necessary to perform a periodic execution of a control job without any substantial time fluctuation (i.e., "chattering"), with a predictable response time to a response to a control request.

If the ethernet message is to be sent periodically, for example as part of a real-time application executed by a control computer designed as a node within the ethernet network, the control computer sends the appropriate ethernet message to its ethernet controller in each control cycle using a software driver integrated into the operating system in order to address the sensors and actuators linked via the ethernet transmission link. In this case, the software driver automatically adds the break time, start identifier, preamble and checksum defined by the ethernet transmission standard (IEEE 802.3) to the real-time data to be transmitted before transmitting the ethernet information to the ethernet controller. The ethernet controller then loads the corresponding ethernet information into its transmit shift register, preferably using a direct memory access transfer operation, and begins sending the ethernet information forward over the ethernet transport link from a particular fill level of the transmit shift register.

Such a transmission sequence of the control computer connected to the ethernet controller contains a plurality of operations which are affected by jitter, which in the worst case accumulates and then exceeds a maximum permissible value for real-time applications, which is generally in the range of a few microseconds. In this case, jitter is interrupted by fluctuations that occur at the node while the node generates data to be transmitted, by waiting times, and is maintained by routine fluctuations within the program code until before transmission of the ethernet message. In modern control computers with a cache, the routines of the program code also fluctuate additionally, since depending on the cache capacity it is necessary to wait for different lengths of time until the memory at the node has the required capacity.

The ethernet controller is typically connected to the nodes via a bus system, of which a PCI bus is commonly used. Since this bus is usually also used by other system components, there may be different waiting time lengths due to the bus configuration. This is encountered both when the ethernet controller utilizes direct memory access transfers to access the physical memory of the controlling computer and when real-time data is transferred over the bus system under the control of software drivers. Similar chattering occurs at all times during bus configuration. Furthermore, the ethernet controller always sends ethernet messages onwards on the ethernet transport link starting from a specific fill level of the transmission shift buffer. In this case, the transmission operation of the ethernet information may then be delayed for different lengths of time depending on the fill level of the transmission shift buffer, which causes additional jitter.

If the total jitter obtained in the transmission operation is higher than the maximum allowable jitter of the corresponding real-time application, this conflict requires the use of a sophisticated method such as IEEE 1588 (IEEE standard for a precision clock synchronization protocol for network measurement and control systems) to provide a sufficiently accurate time base among all the communicating users on the Ethernet transmission link, which can then be used to compensate for the jitter.

Disclosure of Invention

It is an object of the invention to provide a method for transmitting data in the form of ethernet messages over an ethernet transmission link, an interface for linking a node to an ethernet transmission link and an ethernet network which can be used in a simple manner for jitter-free and periodic transmission of ethernet messages, in particular messages containing real-time data.

This object is achieved by a method according to claim 1, an interface according to claim 9 and an ethernet network according to claim 14. Preferred developments are specified in the dependent claims.

According to the invention, in order to transmit data in the form of ethernet messages over an ethernet transmission link using an interface for linking a node to the ethernet transmission link, the data to be transmitted are converted using a conversion unit which conforms to a transmission standard of the ethernet protocol in order to provide the ethernet messages, and then a transmission operation of the provided ethernet messages is carried out using a transmission unit, wherein successive ethernet messages are output onto the ethernet transmission link.

The continuous transmission of Ethernet information of the present invention allows the transmission operation and jitter-free transmission of Ethernet information to be truly repeatable. The fact that the interface for linking the nodes to the ethernet network transmits the next ethernet message directly after an ethernet message has been transmitted ensures that all jitter-affected operations in the transmission sequence from the operation of converting the data to be transmitted into ethernet messages up to the operation of outputting the messages onto the ethernet transmission link are compensated. This is because the timing of the transmission operation is determined entirely by the node-to-ethernet interface, ensuring by the continuous transmission of information that the latter is completely free of jitter.

According to a preferred embodiment of the invention, for carrying out the transmission of the provided ethernet messages, the length of the cycle time is modified for a specified length of the ethernet messages to be within the framework of the maximum permissible duration of the cycle, so that the ethernet messages are output continuously onto the ethernet transmission link over the entire cycle time. This embodiment ensures continuous transmission operation of ethernet messages within an architecture that specifies a maximum transmission cycle duration while optimizing the use of cycle lengths, and jitter that occurs during ethernet message generation is completely eliminated.

According to a further preferred embodiment of the invention, after the transmission of the ethernet messages is performed by means of the interface, the number and/or length of the ethernet messages to be transmitted in a cycle is adapted to a specified cycle time, so that the ethernet messages can be output continuously onto the ethernet transmission link during the entire specified cycle time. This embodiment allows continuous transmission operation of ethernet information within a framework of transmission cycles while optimizing the use of the data length available in the cycle, and jitter occurring during generation of ethernet information is completely eliminated.

According to another preferred embodiment, the modification of the ethernet information takes into account: the baud rate of the ethernet transmission link, the length of the start identifier, the preamble and checksum embedded in the corresponding ethernet message during the data conversion according to the transmission standard of the ethernet protocol, and the length of the break time to be followed between the ethernet messages to be transmitted. This embodiment provides a simple way of determining the optimum length of ethernet information to be continuously transmitted over the ethernet transport link. Another advantage of this example is that it allows for the number and length of the ethernet messages to be sent to be chosen such that in one cycle a plurality of ethernet messages (preferably of the same length) are sent, the total bit length corresponding to the cycle time, if the calculated length of the ethernet messages is greater than the maximum possible length of the ethernet messages. This ensures that it is a simple matter to determine an optimal length of the ethernet information to be transmitted.

According to another preferred embodiment, the provided Ethernet information is stored in a buffer memory, the interface starting the transmission operation with reference to a specified filling level of the buffer memory. This embodiment ensures that there is always enough ethernet information to be sent within the interface to ensure a continuous transmission operation. This prevents the transmission operation from idling and consequently causing a delay in the information that would subsequently cause a violation of the cycle.

Another insight of the present invention is that if the data is real-time data, a real-time application generating the real-time data to be transmitted is synchronized at the node to the transmission of the ethernet message. This embodiment prevents the overflow of ethernet information within the interface during the transmission operation, which would result in the ethernet information no longer being able to be sent fast enough. The fact that the real-time application program operated at the node is tuned to the time base of the interface performing the transmission operation ensures that the node only transmits ethernet messages in a state coordinated with the transmission operation of the interface, and therefore does not overflow the ethernet messages.

According to a further preferred embodiment, an ethernet network with an ethernet transport network linked to a plurality of nodes is designed such that ethernet messages can be transmitted without collision on a transport channel. This ensures that transmissions over the ethernet can continue without interruption of the transmission operation and thus violation of the loop due to collisions on the transmission link.

Another advantage of this embodiment is that the ethernet transmission link has a ring topology arrangement in which ethernet messages sent out by the transmission node are forwarded from one node to the next. This embodiment allows the ethernet information to be transmitted from one node to the next with little delay and without collision.

Drawings

The invention is described in more detail below with reference to the attached drawing figures, wherein:

FIG. 1A illustrates an Ethernet network;

FIG. 1B shows a design of an Ethernet node connection of the present invention;

FIG. 2A shows an Ethernet message, an

Fig. 2B illustrates a transfer operation of the present invention.

Detailed Description

The ethernet concept is the most common communication standard in area-restricted communication networks (LANs), which allows the use of data resources between workstations (generally referred to as computers or machines, hereinafter also referred to as nodes) to be linked in a simple manner. In this context, ethernet is based on a LAN design in which a plurality of nodes are interconnected via a common transmission medium, with the ethernet concept encapsulating data to be transmitted in a predetermined format into data packets (also referred to as ethernet messages in the following). In this case, the ethernet includes three parts, namely the transmission link and the network interface (i.e., hardware), the set of protocols that control access to the ethernet transmission link, and the ethernet information format.

Fig. 1A schematically shows an ethernet network, in which a plurality of nodes 1 are connected to one another via an ethernet transmission link 2. In this arrangement the nodes are linked to the ethernet transmission link by an ethernet controller 3, which is preferably integrated in the relevant node. A node 1 for linking to an ethernet transmission link 2 provided with a neighboring ethernet controller 3 according to the invention is shown in detail in fig. 1B. It is the responsibility of an encoding unit 31 to send ethernet messages in the ethernet controller and it is the responsibility of a decoding unit 32 to receive ethernet messages from the transmission link 2. The encoding unit 31 and the decoding unit 32 each have a respective buffer memory 33, 34 in the form of a shift register connected thereto for buffer-storing the ethernet messages to be transmitted and received. These transmit and receive shift registers 33, 34 are thus preferably designed such that they can directly access a physical memory 11 at the node 1 using a "direct memory access" (DMA) mode. Alternatively, it is possible for data exchange operations to occur between the sending transpose memory 33 or the receiving transpose memory 34 and the physical memory 11 via a Central Processing Unit (CPU)12 at the node 1. But direct access using DMA mode allows faster data exchange.

The exchange of data between the physical memory 11 of the node 1 and the ethernet transmission link 2 is typically controlled by a CPU 12 at the node 1. In addition, the node CPU 12 also manages all operations required to operate the ethernet, i.e., manages sending and receiving operations, and ensures that data to be sent by the node is encapsulated into ethernet information and that data is decapsulated (unpacking) from the received ethernet information. The operating system executed by the CPU 12 at the node 1 typically has a layered software architecture to separate protocol-specific processing operations from information-specific and hardware-specific processing operations. It is thus possible to use different communication protocols in the ethernet standard without any changes to the hardware-specific drivers. At the same time, it is also possible to change the hardware of the node without making a protocol-specific software change at the same time.

An ethernet message 5 can hold up to 1500 bit sets and consists of a header part with a start identifier 51, a preamble 52 (which identifies the destination and source addresses and the type of data packet), a central part 53 containing data, and an end part 54 containing a checksum and used as an error recognition mechanism, the structure of which is schematically shown in fig. 2A.

An ethernet message transfer operation using the ethernet transmission link 2 is performed such that the software driver used by the CPU 12 converts the data to be transmitted into ethernet messages which are stored in the physical memory 11 of the node 1 when the ethernet controller 3 operates in DMA mode. These stored ethernet messages are then accessed by the transmit shift buffer 33 of the ethernet controller 3 to load the ethernet messages into the shift buffer. If there is already enough ethernet information transferred from the physical memory 11 to the transmission shift register 33 under the control of the software driver of the CPU 12, and thus a sufficient filling level has been reached, the transmission shift register 33 outputs the buffered ethernet information via the coding unit 31 onto the ethernet transmission link 2. In this case, the ethernet data transmission only takes place while the ethernet network is in a rest state. In addition, there is also a collision avoidance mechanism in the ethernet transmission link 2.

When an ethernet message is received, the received ethernet message is buffered in the reception shift buffer 34 by the decoding unit 32, and an interrupt is initiated at the node 1 by the ethernet controller 3. This interrupt prompts the software driver within the CPU 12 of the node 1 to transfer the received information to the physical memory 11 using DMA mode and then forward the information to the operating system of the node for processing.

The ethernet concept is used as a communication protocol for network systems, in particular because standard hardware and software components can be used, and in addition a high data transmission rate is possible. However, when the ethernet standard is used in an industrial environment, particularly for automation, the ethernet protocol must ensure real-time data transmission. In order to be able to reliably execute a real-time application, such as a machine controller, using an ethernet network, a cycle time of 50 microseconds and an acceptable jitter time (i.e. an error from the desired cycle time) of 10 microseconds are necessary.

If a control computer, which is a node 1 in an ethernet network, has to control in real time the sensors or actuators linked to the ethernet transmission link 2 as other nodes, the control computer, in each control cycle, transfers the ethernet message 5 to the relevant ethernet controller 3 for transmission by means of a software driver stored in its CPU 12. The ethernet controller 3 will then load the corresponding ethernet message 5 (preferably in DMA mode) into its transmit shift buffer 33 and start transmitting the ethernet message over the ethernet transmission link 2 from a certain filling level in the transmit buffer.

However, such transmission sequences contain a number of operations that are affected by jitter, which accumulates in the worst case. A first jitter is derived only from fluctuating outage latencies in controlling the computer operating system and software drivers to convert Ethernet messages. Furthermore, routine fluctuations may occur within the data code that passes before the ethernet information is transmitted. In modern cache-based control computers, the run time of the same passcode also fluctuates, since depending on the cache size it is necessary to wait for different lengths of time to wait for the desired memory. Further jitter may also occur when transferring ethernet information to an ethernet controller. The ethernet controller is linked to the control computer via a bus system, such as a PCI bus. Since the bus is also used by other system components of the control computer, different lengths of latency may be incurred during bus configuration when the ethernet controller accesses physical memory to transfer ethernet messages to the transmit shift registers. Even when the ethernet controller does not operate in DMA transfer mode but transfers data from physical memory to the transmit shift registers of the ethernet controller via the CPU, a similar jitter is induced during bus configuration. In addition, the transmission of Ethernet messages is delayed for different lengths of time depending on the fill level of the transmit shift register. If all of the operations of the dithered image are accumulated, there is a risk that the total dither is larger than the dither permitted by the corresponding real-time application and real-time control is no longer ensured.

In order not to perform complex methods to bring the timebase into the line between the individual nodes and thus compensate for communication jitter as is conventional, the present invention involves programming the ethernet controller 3 with a software driver of the CPU 12 at the computer node 1 so that the ethernet message 5 is sent out from the transmit shift buffer 33 without a break in time. In this case, the transmission shift buffer 33 and the associated coding unit 31 of the ethernet controller 3 are controlled such that the next ethernet message is transmitted immediately after an ethernet message has been transmitted, following the break time defined in the ethernet transmission standard.

To ensure that ethernet messages are sent continuously during a specified cycle time in a real-time application to be executed, the software driver of CPU 12 calculates how much and how long ethernet messages need to be sent in order to actually follow the specified cycle time. The software driver compiles the data 53 to be transmitted into ethernet information 5 of appropriate length with a start identifier 51, a preamble 52 and a checksum 54 according to the ethernet transmission standard and stores it in the physical memory 11 of the node 1. The transmission shift buffer 33 in the ethernet controller 3 then accesses these ethernet messages 5 and buffers them. From the transmission of a particular fill level in the shift buffer 33, the transmission operation begins and the ethernet message is continuously transmitted, as shown in fig. 2B. In this case, a transmission operation is shown in which two messages of the same length are sent in a given cycle time, following a specified break time.

With the transmission shift buffer 33 integrated in the ethernet controller 3, the operation of providing ethernet messages in the physical memory 11 of the node 1 by the software driver in the CPU 12 is decoupled from the transmission time for these ethernet messages, so that jitter due to real-time applications and jitter caused when ethernet messages are transferred to the ethernet controller 3 are compensated. Since the timing of the transmission operation is entirely dependent on the ethernet controller 3 and the downstream transport entity of the ethernet transmission link 2, and the ethernet controller continuously issues the ethernet messages 5 from its transmission shift buffer 33, a true repetitive and jitter-free transmission is possible.

To allow continuous transmission of ethernet messages, real-time applications are synchronized at the node with the ethernet controller 3 using software drivers within the CPU 12. The ethernet controller 3 defines a time base from which real-time applications are synchronized on the control computer 1. This ensures that the software driver in the CPU 12 will always transfer enough transport ethernet information to the ethernet controller 3 to prevent the transmission shift register 33 in the ethernet controller 3 from idling and thus prevent the occurrence of message delays that would violate cycle time. Furthermore, the synchronization of the real-time application at the node 1 with the time base of the ethernet controller 3 ensures that not too much ethernet information is transferred to the ethernet controller 3, so that the transmission shift buffer 33 does not overflow and the ethernet information can thus be transmitted quickly.

When counting the amount and length of ethernet messages to be transmitted in a cycle of the real-time application, the software driver in the CPU 12 of the node 1 will keep track of the baud rate used on the ethernet transmission link 2 and the additional data (i.e. the start identifier 51, the preamble 52 and the checksum 54) automatically embedded after the data to be transmitted has been encapsulated, as well as the break time between ethernet messages. These additional signals are defined in the ethernet standard IEEE 802.3 and in the case of a 100-Base-Tx ethernet, i.e. a 100Mbaud fast ethernet, the start identifier is 8 bits, the preamble is 56 bits, the checksum is 32 bits and the break time is 69 bits.

If a cycle time of x microseconds for a real-time application must now be achieved, the following equation applies: (L is the maximum bit length of the Ethernet information)

L＝(x.100)—(8+56+32+69)

In the case where the cycle time is 100 microseconds, one would get:

set of 1226 bits L9808 bits

The software driver in the CPU 12 of node 1 is thus able to send one or more messages of the same length in a 100 microsecond cycle time, the total length of which, including the break time, is 1226 bit groups. For example, if two messages are sent, the length of each message bit group is 613 bit groups (including the off-time). If the length of the obtained information of a cycle time is larger than the maximum ethernet information of 1500 bit groups, it is absolutely necessary to divide the information into a plurality of pieces. It has then been necessary to properly transmit multiple ethernet messages to use the entire cycle time. For example, if a 500 microsecond cycle time is specified, five messages each containing a 1226-bit group (100 microseconds) may be sent.

As an alternative to specifying a cycle time, it is also possible to achieve continuous transmission of ethernet messages by specifying the length of the ethernet message to be transmitted and then deriving the required cycle time therefrom. In this case, the software driver within the CPU calculates the optimum cycle time to ensure continuous delivery of the Ethernet message from the specified length of the Ethernet message and from the maximum allowable duration of the control cycle, so as to be able to execute a real-time application such as a machine controller using the Ethernet network. The software driver then compiles the data to be transmitted into ethernet information of appropriate length according to the ethernet transmission standard, containing a start identifier 51, a preamble 52 and a checksum 54, and stores it in the physical memory 11 of the node 1. The transmission shift register 33 of the ethernet controller 3 then accesses these ethernet messages 5 and stores them in a buffer. From a certain filling level in the shift buffer 33, the transmission operation starts and the ethernet information is continuously transmitted during the calculated cycle time, following a defined break time.

In the transmission operation of the present invention, in which the transmission operation is carried out substantially continuously, it is also necessary to ensure that no collisions occur on the transmission channel, since in this case the ethernet controller must interrupt the transmission and start the transmission again at a later time. In this case, an appropriate Ethernet topology for collision free transmission operation is a point-to-point connection between nodes. It is also possible to activate multiple users with a crash-proof switch. A multi-node ring network is also possible in which case ethernet messages can be transmitted between nodes and then returned to the original transmitting node with little delay.

Generally, real-time applications also require feedback from the user that has been activated. In this case, the ethernet transmission link 2 is in the form of a full duplex transmission link with separate transmission and reception channels, so that the ethernet information to be transmitted is not affected by the reception information containing feedback. In this case, the amount of return data must also not exceed the amount originally sent out, since the latter is equivalent to the maximum transmission capacity.

Claims (13)

1. A method of transmitting data in the form of ethernet messages over an ethernet transport link, having the steps of:

converting data to be transmitted according to an Ethernet transmission standard to provide Ethernet information, and

periodically transmitting the provided Ethernet information using the Ethernet transmission standard, wherein

The transmission operation for the provided ethernet information is controlled such that after an ethernet information has been transmitted, the next ethernet information is transmitted directly while following the break time defined in the ethernet transmission standard, so as to continuously output ethernet information onto the ethernet transmission link during the entire cycle time;

for the transmission of the provided ethernet messages, the length of the cycle time is adapted to a specified length of the ethernet messages within the framework of a maximum permissible duration of a cycle, in order to continuously output ethernet messages onto the ethernet transmission link during the entire cycle time.

2. A method of transmitting data in the form of ethernet messages over an ethernet transport link, having the steps of:

converting data to be transmitted according to an Ethernet transmission standard to provide Ethernet information, and

periodically transmitting the provided Ethernet information using the Ethernet transmission standard, wherein

The transmission operation for the provided ethernet information is controlled such that after an ethernet information has been transmitted, the next ethernet information is transmitted directly while following the break time defined in the ethernet transmission standard, so as to continuously output ethernet information onto the ethernet transmission link during the entire cycle time;

for the transmission of the provided ethernet messages, the amount and/or length of the ethernet messages to be sent within a cycle is adapted to a specified cycle time, so that the ethernet messages are continuously output onto the ethernet transmission link during the entire specified cycle time.

3. The method according to claim 2, wherein for calculating the number and/or length of the ethernet messages to be transmitted in the cycle, the baud rate used on the ethernet transmission link, the length of the start identifier, the preamble and the checksum inserted into the ethernet message when the data is converted according to the ethernet transmission standard, and the length of the break time to be followed between the ethernet messages to be transmitted are taken into account.

4. The method of claim 3, wherein the maximum bit length L of the Ethernet information to be transmitted in the loop is calculated as follows:

L＝(ba×zy)—(st+pr+ch+pa)

wherein the baud rate used on the ethernet transmission link is ba Mbaud, the cycle time is zy microseconds, the length of the start identifier is st bits, the length of the preamble is pr bits, the length of the checksum is ch bits and the length of the break time is pa bits.

5. The method of claim 4, wherein when the maximum bit length L is greater than the maximum possible bit length of Ethernet messages, the number and length of Ethernet messages to be transmitted are selected such that multiple Ethernet messages are transmitted in the cycle, the total bit length of the multiple Ethernet messages corresponding to the cycle time.

6. A method as claimed in any one of claims 1 to 5, characterized in that for the transmission of the provided Ethernet information, the provided information is stored in a buffer memory and the transmission operation is initiated as soon as the buffer memory has reached a specified filling level.

7. The method according to any of claims 1 to 5, wherein the data to be transmitted is real-time data, and a real-time application generating the real-time data to be transmitted is synchronized with the Ethernet messaging operation.

8. An ethernet network node having:

a control unit (1; 11, 12) which converts the data to be transmitted in accordance with an Ethernet transmission standard in order to provide Ethernet information, and

a transmission unit (3; 31, 33) which periodically transmits the provided Ethernet information over an Ethernet transmission link (2) using the Ethernet transmission standard,

wherein the control unit (1; 11, 12) controls the provided Ethernet information transmission operation by the transmission unit (3; 31, 33) such that the next Ethernet information is sent directly after a transmitted Ethernet information while following the break time defined in the Ethernet transmission standard, so as to continuously output Ethernet information onto the Ethernet transmission link (2) during the entire cycle time;

the control unit (1; 11, 12) is designed to adapt the length of the cycle time to a specified length of the Ethernet message to be transmitted in the framework of a maximum permissible duration of a cycle, in order to continuously output Ethernet messages onto the Ethernet transmission link (2) over the entire cycle time.

9. An ethernet network node having:

a control unit (1; 11, 12) which converts the data to be transmitted in accordance with an Ethernet transmission standard in order to provide Ethernet information, and

a transmission unit (3; 31, 33) which periodically transmits the provided Ethernet information over an Ethernet transmission link (2) using the Ethernet transmission standard,

wherein the control unit (1; 11, 12) controls the provided Ethernet information transmission operation by the transmission unit (3; 31, 33) such that the next Ethernet information is sent directly after a transmitted Ethernet information while following the break time defined in the Ethernet transmission standard, so as to continuously output Ethernet information onto the Ethernet transmission link (2) during the entire cycle time;

the control unit (1; 11, 12) is designed to adapt the quantity and/or length of the Ethernet messages sent in a cycle to a specified cycle time in order to continuously output Ethernet messages onto the Ethernet transmission link (2) during the entire cycle time.

10. A node according to any one of claims 8 to 9, characterized in that the transmission unit (3; 31, 33) has a buffer store (33) for buffer-storing the provided ethernet information, the control unit (1; 11, 12) being designed to start the transmission operation in dependence on a specified filling level in the buffer store (33).

11. A node as claimed in any one of claims 8 to 9, characterized in that the control unit (1; 11, 12) is designed to synchronize a real-time application generating real-time data to be transmitted with the ethernet messaging operation.

12. An ethernet network having an ethernet transmission link (2) and a plurality of nodes (1) according to any one of claims 8 to 11 connected to the ethernet transmission link, wherein the transmission channels of the ethernet transmission link (2) are designed to transmit the ethernet messages without collisions.

13. An ethernet network according to claim 12, characterized in that the ethernet transmission link (2) has a ring topology arrangement and the transmitted ethernet information is forwarded from one node to the next.