ARCNET Network

Arcnet is a token-passing local area network ( LAN ) especially designed for embedded technologies. This standard offers up to 10 Mbps data rate, 255 nodes or stations, variable packet length of unicast or broadcast messages, over a linear bus or over a distributed star wiring, with deterministic performances and a low cost per node. With this real-time networking characteristic, it was ideally suited for industrial networks and for process control applications.

Like the Ethernet networks and the Controller Area Network bus ( CAN bus ), Arcnet is a data-link layer technology which does not define any application layer ( Arcnet standard defines the data link layer and the physical layer which are the two lowest layers in the OSI model ). So, when a « Lambda » company searched to develop a new network for answering new usages, its designers used Arcnet to build the data layer, then they added their own programs to build the application layer according to their specific needs. Frequently, it was never mentioned that Arcnet was being used in their communicating solutions.

The « Arcnet » term means something like « Attached Resources Computer Network ». At the beginning in 1976 ( officially in 1977, that is during the « glorious years » of the digital economy ) this system has been created as an office network by John MURPHY, a « Data Point Corporation » scientist.

Right at the start, the technique was developed with the objective to build heterogeneous networks by connecting together some terminals talking with any floppy disk system, while usually the objective was to build homogeneous networks by connecting similar systems together. In other words, Arcnet did not make any assertions about the type of hardware that would be connected.

Hey ! Do you not remember the 5" ¼ floppy disks and the 3" ½ floppy disks ? We are sorry, but we should still remember we are talking about the « glorious years » of the digital technologies, indeed ?! Before Ethernet, it was the first clustering solution based on local networks.

Then, the system has become an embedded networking technology that is still frequently found in applications such as industrial control, robotics, power generation, telecommunications, security systems, transportation, building automation ( as a data link layer for BACnet for instance ) and gaming. In the same time - getting in mind we are talking about I/O device-sharing network - the token-passing ( or token-ring, or token-bus ) protocol was chosen as an interconnection method.

In the early 80s, Ethernet started and the IBM PC microcomputers were proposed. While Arcnet was more flexible, more robust, more deterministic, less expensive and more resources friendly than the first Ethernet networks, the Arcnet market share was approximately equal to the Ethernet one's at the end of the decade.

In the early 90s, Ethernet switches to an easier cabling, that is from co-axial cabling to twisted-pair cabling over a star topology based on hubs, improving its performances ( 10 Mbps compared to the 2.5 Mbps data rate for Arcnet ) with less expensive technologies. Finally, the Ethernet demand increased and the Arcnet demand decreased.

In response - 1992 - the company started to develop « Arcnet Plus » ( 20 Mbps ), but it was too late. Ethernet had definitively got the leadership.

To conclude, Arcnet has been a very popular network, widely used all around the world in control systems and in embedded applications. Over 22 million nodes were sold - for a large part of them still operating - including those from many major PLC's makers. But by the time now, this technology is rarely used for new networks installations or developments.

Originally introduced just before Ethernet, Arcnet incorporates a « token-passing » protocol - which is also named « token ring » or « token-bus » - where media access is determined by the station with the token. Very quickly, this technique consists in the activation of a single token for all one network, then in the circulation of the token from one node to another. This way, only the owner of the token can send data at any time before passing this token to its next neighbour.

The result is that no station can consume all the bandwidth since only one packet can be sent each token pass for a maximum message length of 507 bytes. This scheme >avoids collisions ( because there are not possible ) and gives Arcnet its greatest advantage in real time applications : Arcnet is a deterministic network, while standard Ethernet is not.

Thus, some modern industrial Ethernet control networks have got deterministic features by using specific and well adapted mechanisms, as the Ethercat and Powerlink networks have done for instance. They can even supply a very higher level of accuracy for real time synchronization in data exchanges.

Arcnet systems supports various media, and two topologies. All these media and all these topologies can be mixed together in a whole network, resulting in a very flexible communication solution. In fine, if you cut one ARCNET network … you just get two ARCNET networks within milliseconds !

Media

• Coaxial cable.
• Twisted pair ( RS-485 ).
• Optic fibres.

Topologies

• Hub-less bus topology.
• Distributed star topology requiring hubs.

A designer can select popular and lower cost RS-485 transceivers when distances are relatively short. One of the complaint regarding Ethernet is the need to operate in a star topology which requires the use of hubs or switches. Not only does these devices require a source of power, the designer needs to find a place to mount them and it represents a source of failure. For embedded applications, bus topology is much more convenient since hubs and switches are not required.

The data exchange protocol is based on the deterministic token-passing technology. It supports packet flow control, error detection ( CRC 16 ), auto-reconfiguration, variable packet size up to 507 bytes and various software drivers and it needs only five simple commands to do all that. As for the performances, the minimum message time takes 141 µs over a 2.5 Mbps Arcnet network.

Token-passing technology

This technology assumes that every node has a guaranteed response time to transmit, offering a deterministic behaviour.

Short resume ( for more details, please read the token bus description ) :

• Transmitting on the network is only permitted when a node has the token.
• Every node can transmit once during each token rotation.
• Users can compute the maximum of point to point transmission time, and they can compute the worst case.

Auto-configuration

The network is automatically reconfigured when a node joins or leaves the network.

• Token pass is automatically reconfigured. It takes a typical time of 20 to 30 milliseconds.
• The network supports live node insertion and deletion.

Packet-flow control

The transmitter checks receiver to make sure it is ready to receive a packet.

• Reduced software overhead.
• Increased bandwidth.
• No lost packets due to input buffer overruns.

Commands

• ITT : Invitation to transmit.
• FBE : Free buffer enquiry.

• PAC : Packet.
• ACK : Acknowledgement.

• NACK : Negative acknowledgement.

• Deterministic performances.
• Logical ring.

  Nodes automatically find their neighbour to create a ring.

• Automatic reconfiguration.

  A new node joins the ring automatically without software intervention.

• Broadcast and directed messages.
• Multi-master with automatic token generation.
• Multiple cabling options.
• High speed ( relatively to the 80s ).

  Standard 2.5 Mbps, optionally 19 kbps to 10 Mbps.

• Low cost chips.
• Low protocol overhead.

  3 or 4 « bytes-good » usage of available bandwidth.

• Variable packet size.

  0 to 507 bytes.

• High noise immunity and EMC standards compliant.
• Easy and simple manageable technology.

• Used in embedded design to communicate between controllers ( although it is at a higher level and in a less efficiency manner than for transputers ).
• A multi-master network of nodes allows equal access to all nodes.
• Simple low cost hardware design, that is one Chip plus transceiver.
• Robustness.
• The Arcnet worst cases performance are largely better than the Ethernet one's.
• No special development tools required.

  ---

• Used for industrial high speed applications ( relatively to the 80s and to the earlier 90s ).

CAN was originally designed for the automotive electronics market and has experienced success as a device level network when used in conjunction with higher layer protocols such as DeviceNet and CANopen. As Arcnet is best considered a controller level network because of its higher performances compared to the CAN ones.

Basically, CAN communication is limited to 1 Mbps and it can only send eight data bytes per frame. Arcnet chips can transmit up to 10 Mbps and data packets can be up to 507 bytes in length. Not only is Arcnet faster than CAN, Arcnet can send more data per transmission.

Nevertheless, some modern control networks such as Ethernet Powerlink integrate CAN frames into Ethernet frames, and then improve the low performance of the basic CAN network, while keeping some benefits of the CAN architecture.

This tutorial presents a clear summary about the main characteristics of this digital communication standard. This presentation was written by « Arcnet » organization members. You will learn what it is, where it is used, how does it work and what about the protocol features. More, you will learn some things about logical ring, Arcnet frames and token technology, Arcnet messages timing, topology, signals, controllers and processors in usage.

This standard specifies all the functionalities of the system as a local area network token bus.

• The first chapters present the capabilities, the fundamentals, the MAC sub-layer, the physical layer and the supported media.
• Then, you will find the specifications and the descriptions for the formats and the facilities : the frame formats, the field description, the timers, the flags and the registers.
• You will read the specifications about the protocol, the services, the physical layer, the electric cabling, the optic fibre wiring, and the hubs.
• Finally, you can read the specifications to build a balanced multipoint system.

This document is a draft for the 878.2 standard, specifying a mechanism of packet fragmentation. These specifications were written in the early 90's. It seems they were a proposition similar to the TCP standard. The major part of this draft specifies the protocol, that is the transmission and the frames formats. The protocol described here was originally developed under the sponsorship of the companies « Apple Computer », « Novell », « Actinet », « Standard Microsystems » and « Pure Data Search ». It was published as RFC 1201.