These companies usually want to buy an off-the-shelf ‘bridge’ or a ‘router’ which will easily allow messages to pass between the ARCNET and Ethernet networks. Unfortunately in most cases it is not possible to buy an off-the-shelf product which meets the company’s requirements, and some custom engineering is often necessary. This article discusses the technical issues involved in specifying and designing devices that allow messages to be passed between ARCNET and Ethernet networks.

ARCNET and Ethernet Compared

ARCNET and Ethernet are network protocols which define how messages are transported between devices connected to a local area network (LAN). They are similar in that neither protocol defines the content of the messages; in terms of the OSI model of network protocols, both ARCNET and Ethernet function as data link and physical layer protocols. However, there are some important differences which are particularly relevant to the problem of transporting messages between ARCNET and Ethernet networks.

ARCNET Data packets and Ethernet packets are similar in their general structure. Both contain:

addresses of the sending and receiving devices
the message
error checking data

On the local area network, each device must have a unique address. ARCNET uses one-byte addresses which are typically selected when devices are installed. Ethernet uses six-byte addresses which are typically programmed into an EEPROM when a network interface card or device is manufactured. Both ARCNET and Ethernet define particular addresses which are used for broadcast packets.

ARCNET Data packets may carry messages of up to roughly 500 bytes, whereas Ethernet packets can accommodate messages of up to 1500 bytes. If longer messages must be transmitted, a technique known as ‘segmentation’ is often used to break the message into chunks that will each fit into a packet; note that segmentation is always performed by the network software, not the hardware.

ARCNET and Ethernet also differ in how packets are transmitted between devices. Ethernet is rather simple: basically an Ethernet controller can transmit a packet when no other device is transmitting. Ethernet does not define whether or how the receiving device should acknowledge receipt of a packet, so packets can be sent to non-existent or non-functioning devices without the Ethernet hardware being able to provide any indication of a problem to the network software. This simplicity makes it relatively easy to construct Ethernet bridges.

By contrast, an ARCNET Data packet cannot be transmitted until after a successful Free Buffer Enquiry handshake, a sequence of two short packets (one in each direction) whose purpose is to ensure that the eventual recipient of the Data packet has a buffer available in which it can store the Data packet. In addition, the recipient of a Data packet should acknowledge receipt of the Data packet by transmitting an Acknowledgement packet. These extra required packets can increase the reliability of the communication, but they also make it difficult for ARCNET bridges to be constructed.

There are numerous other differences between ARCNET and Ethernet that are relevant if you need to select a networking technology for an application, but they are beyond the scope of this article.

Hubs, Bridges and Routers

Hubs, bridges and routers are popular connectivity devices in corporate networks. Let’s examine how they could be used in order to solve ARCNET connectivity problems.

Hubs

Hubs (sometimes referred to as repeaters) are generally used to connect multiple network segments to form a single LAN. All of the network segments to be connected must use a particular networking technology, but some hubs will connect segments which use different types of cable. A LAN which includes one or more hubs must still meet the timing requirements of the networking technology, which may limit the maximum distance between devices. In addition, there may be limits on the number of hubs between any two devices on the network.

ARCNET hubs are commonly used to connect multiple ARCNET cable segments because a single cable segment is limited in both its length and the number of devices which it can connect. The use of hubs also allows more flexibility in the wiring topology. It is not possible to construct a hub which can connect ARCNET and Ethernet networks because of the fundamental differences between the two technologies.

Bridges

A bridge issued to connect multiple LANs which use the same data link networking technology. Bridges are typically used in order to overcome distance limitations of a networking technology. Intelligent bridges (sometimes referred to as switches) can be employed in order to increase the useful bandwidth of a network by providing some isolation between the LANs. Although the LANs connected by a bridge retain much of their individuality, no two devices in any of the LANs should have the same network address.

Ethernet bridges/switches are quite popular, but ARCNET bridges may not exist at all. The reason for this is primarily technical: it is very hard to build an ARCNET bridge because of the way the ARCNET protocol works. As discussed earlier, an ARCNET Data packet must be preceded by a Free Buffer Enquiry exchange and should be followed by an Acknowledgement packet transmitted by the recipient of the Data packet. An ARCNET bridge would have to mimic this handshaking behaviour for every possible destination device, which is probably impossible using ARCNET components available today; in addition, there would be the danger that the bridge could interfere with communication between ARCNET devices on the same LAN.

Likewise, you will probably never see an ARCNET-to-Ethernet bridge. As mentioned earlier in this section, bridges connect LANs which use the same networking technology. Although it is theoretically possible to construct a bridge that can connect LANs which use extremely similar networking technologies. ARCNET and Ethernet are probably not similar enough.

Routers

The purpose of a router is to selectively forward packets between different LANs. For a router to be able to do its job there must be a common higher-level device addressing scheme being used on all LANs which are to be connected, and the router must have some way of associating a particular higher-level address with a particular LAN address. A router decides where to send a message based on the higher-level address of the destination device contained in the message.

The behaviour of a router is typically defined in the protocol which has the functionality of the network layer in the OSI model. Examples of popular network-layer protocols include IP (used in TCP/IP networks) and IPX (developed by Novell). The higher-level addressing scheme used in IP is a four-byte address (e.g. 192.168.1.4), and every device on an IP internetwork must have a unique IP address. An IP router generally uses the Address Resolution Protocol (part of TCP/IP) to determine the LAN address of a particular device from its IP address.

The use of ARCNET-to-Ethernet routers are a solution to some connectivity problems. If both the ARCNET LAN and the Ethernet LAN which are to be connected use a common network-layer protocol such as IP, then will be possible to buy or construct a router which will forward messages between the two networks. However, not all ARCNET networks use a network-layer protocol, and in those cases it may be difficult or perhaps impossible to construct an appropriate router depending on the way that ARCNET is being used. In addition, the Ethernet devices must use the message format of the ARCNET network.

Gateways and Proxy Devices

In many case, it will be necessary to use some sort of a gateway to permit communication between ARCNET and Ethernet. A gateway is somewhat like an interpreter: it translates messages from one protocol to another protocol. The accuracy of the translation is limited by the differences between the particular protocols.

Gateways typically have very specific capabilities. There is no such thing as a generic ARCNET-to-Ethernet gateway, because neither ARCNET nor Ethernet define the syntax or semantics of the messages that are transmitted. It is possible to purchase off-the-shelf gateways for certain protocol pairs (e.g. Modbus and DeviceNet), but you should be aware that those gateways will typically only provide a subset of the capabilities of each of the individual protocols. Most off-the-shelf gateways require some specific configuration.

If you decide that you need a gateway to solve a particular connectivity problem because you have exhausted all other possibilities, be aware that developing a gateway can be an expensive undertaking. Most of that expense is in the development of a detailed specification and the software. One popular gateway hardware platform is an industrial PC with appropriate network interface cards, components which are available from many sources. You will probably want to budget for documentation! Naturally, the more you are able to limit the scope of the project, the more likely you are to get something at a price you can afford.

Network Proxy Devices

A relatively inexpensive alternative to a gateway is what we call a network proxy device. The function of a network proxy is to ‘represent’ a device on a different network by relaying messages to/from that device. For example, we can construct a network proxy device with one ARCNET and one Ethernet interface which allows one Ethernet device to communicate with any ARCNET devices connected to an ARCNET network. The network proxy device’s ARCNET interface has an ARCNET LAN address; any messages sent to that address will be forwarded to the designated Ethernet device. Likewise, messages transmitted from the Ethernet device to the network proxy device’s Ethernet interface will be transmitted on the ARCNET network to an ARCNET device specified in the message.

Unlike a gateway, a network proxy device for ARCNET does not have to be programmed to match the format of the messages travelling on the ARCNET network. However, it does have some restrictions:

a network proxy device must have one ARCNET interface for each Ethernet device which it will service. This effectively limits the use of these types of devices to situations in which one or a few Ethernet devices need to be able to communicate with the ARCNET device,
the Ethernet devices must be able to transmit and receive messages using a particular format which is derived from the format used on the ARCNET network.

Conclusion

There is no single solution to the problem of connecting ARCNET and Ethernet networks. When you talk to a systems integrator or consultant about a particular connectivity problem, be prepared for a lot of questions about your existing ARCNET system and your connectivity requirements. If you plan to design a new communication system which will be based on ARCNET and there is any possibility that someone might want to access the ARCNET devices from an Ethernet network (or some other network) in the future, we strongly recommend that the system also use some standard protocol that is routable so that routers can be used. Two such protocols which support ARCNET are TCP/IP and BACnet.

References

“ARCNET Local Area Network Standard” (ATA/ANSI 878.1), ARCNET Trade Association.
“BACnet: A Data Communication Protocol for Building Automation and Control Networks” (ASHRAE/ANSI Standard 135-1995), the American Society of Heating, Refrigerating, and Air-Conditioning Engineers, 1995.
“Guide to Configuring an ARCNET Network with Contemporary Control Systems,” Contemporary Controls.
Tanenbaum, Andrew S., “Computer Networks” (3rd ed.), Prentice Hall, 1996

Jim Butler is the director of software engineering at Cimetrics Technology, a company which specializes in network technology for distributed control and monitoring applications. Jim participated in the development of BACnet, an ANSI-standard network protocol which supports ARCNET and Ethernet. Jim received bachelor’s and master’s degrees in engineering from M.I.T.