Lesson #5: Exploring ARCnet Specifications
ARCnet is an acronym for Attached Resource Computer
NETwork, which was founded by the Datapoint
Corporation. Novell used the term RX-Net (2.5 Mbps)
to denote its form of this architecture and TRX-Net for
the Turbo (25 Mbps) version. ARCnet uses a
token-bus frame/packet passing scheme.
ARCnet operates at 2.5 Mbps throughput,
or 25 Mbps for the "Turbo" version and can
be connected using (standard IBM SNA) RG-62 A/U coax cable or unshielded twisted
pair (UTP) wiring. Although ARCnet can support up to
255 nodes (host computers)
on a single network, systems of this size are not practical.
Typical ARCnet coax cable specifications:
RG-62 A/U, solid conductor, CL2, 90%+ braided
shield, PVC jacket, nominal 93 ohm impedance, 15.5
RG-62 A/U, solid conductor, CL2P, 90%+ braided
shield, plenum jacket, nominall 93 ohm impedance,
12.5 nominal capacitance/ft.
ARCnet was one of the early topologies used in networking
but is rarely used as the topology of choice in current LAN
environments. ARCnet, however, still is a solid, functional
and cost-effective means of networking.
The diagram above shows the components of a typical ARCnet NIC.
This card is configured for use with twisted pair (UTP) wiring. ARCnet cards
for use with coax cable have a BNC -Bayonne Neil Councelman
connector (also commonly called British Naval Connector). You will find
jumpers or DIP switches for setting the following characteristics:
- Node number
- Base I/O address
- Memory address
- Network timeout
node number and network timeout
are characteristics unique to ARCnet.
Each network interface card (NIC) on an ARCnet network is
manually assigned a node number. This number-ID
is an 8-bit hardware address (insted of a MAC address) that must be unique
for each host on the network and be within the range of 1 to 255.
ARCnet manages network access with a token passing
bus mechanism. The token (permission to speak on the LAN) is
passed from the lowest node number to higher number nodes in ascending order.
Lower numbered addresses get the token before the higher numbered addresses.
Network traffic is made more efficient by assigning sequential numbers to
nodes using the same order in which they are cabled. Choosing random numbers
can create a situation in which a node numbered 23 can be a whole building
away from the next number, 46, but in the same room as numbers 112 and 142.
the token has to travel in a haphazard manner that is less effective than if
you numbered the three workstations in the same office sequentially, 46, 47,
and 48, and the workstation in the other building 112. With this configuration,
the frame stays within the office before venturing on to other stations.
Depending on the vendor design specifications of your ARCnet card, most
base I/O addresses, node addresses,
and memory addresses are set by DIP switches. these
addresses are set using a binary mode calculation with an on or off setting
in the required switch block.
Interrupt settings are made by jumper
combinations at marked locations on the network interface card (NIC). Many
types and brands of ARCnet cards are on the market today. Refer to the
documentation for the NIC or call the technical support group for the
specific vendor of the component.
Most ARCnet cards require a shared memory address.
Many manufacturers use the area of D000:0 to DFFF:0 as the default. Standard DOS
memory is limited to the first 640 KB of a PC's memory. However, extended or
expanded memory managers can be used to make memory above 640 KB available
for use by DOS programs. If memory above the 640 KB line (above hex address
A000:0) is used for network boards, however, the memory becomes unavailable
for use by a memory manager. this makes it difficult to optimize a PC's upper
memory area. If a network card requires a memory area such as D000:0-DFFF:0,
the memory available for use by DOS memory managers is reduced by 64 KB. this
is usually accomplished with an optional switch setting such as X=D000-DFFF
to exclude the shared memory address range.
the ARCnet topology uses coax, twisted pair, or fiber optic cabling to
connect network devices. An ARCnet network is used primarily with either
coax or twisted pair cable. Most ARCnet installations are coax and use
RG-62 A/U coax cable terminated with 93 Ohm
terminators. Twisted pair (UTP) installations are newer
and use stranded 24 or 26 gauge wire, or solid core 22, 24, or 26 gauge type cable
terminated with 100 Ohm terminators. Many ARCnet networks use a
mix of both coax and UTP cabling. UTP cable is simple to install and provides a reliable
connection to the workstations, whereas coax provides a means to span longer distances.
Typical ARCnet installations are wired as a physical star.
ARCnet can run off a linear bus topology using coax or
twisted pair as long as the cards specifically support BUS. the most popular
star-wired installations of ARCnet run off two types of hubs:
Passive hubs cannot amplify
signals. Each hub has four connectors. Because of the characteristics of
passive hubs, unused ports must be equipped with a terminator, a
connector containing a resistor that matches the ARCnet cabling
characteristics. A port on a passive hub can only connect to an active
device (an active hub or an NIC). Passive hubs can never be connected to
Active hubs have active electronics
that amplify signals and split them to multiple ports. the number of ports
on an active hub varies with the manufacturer, but eight is typical. A
port on an active hub can be connected to a port on another active device
(such as another active hub or an NIC) or to a passive hub.
One of the greatest flexibilities of ARCnet is that you can integrate
connections from active hubs to a linear bus connection as long as you
terminate at the last connection point.
A maximum time limit of 31 microseconds is allotted for
an ARCnet signal frame. This is also called a network
timeout. Signals on an ARCnet can travel up to 20,000
feet during the 31 microsecond default time-out period. You can
sometimes extend the range of an ARCnet by increasing the time out value. However,
20,000 feet is the distance at which ARCnet signals begin to seriously degrade.
Extending the network beyond that distance can result in unreliable or failed
communication. Therefore, the time-out parameter and cabling distance recommendations
should be increased only with great caution.
The maximum cable distances between individual components in an ARCnet
network are dependent on how the components are connected (see the
table at the end of this lesson).
In cabling ARCnet networks with coax cable, you must
follow several rules:
- Never connect a passive hub to another passive hub directly.
- Passive hubs should never be used to connect two active hubs.
- Passive hubs are only used to connect an active hub and a node.
- Unused connectors on active hubs do not need to be terminated.
- Unused connectors on passive hubs must be terminated using a 93 Ohm terminator.
The figure above shows an ARCnet configuration using active and passive
hubs. Active hubs are required to extend the network for long distances
and to configure networks that have more than four nodes. Passive hubs are
used as an economical means of splitting a port on an active hub to support three
Common sources of problems on ARCnet networks are as follows:
Duplicate addresses. No more than one node can have a given node
address on the same network. If two or more nodes share an address, one
of the two workstations will either lose its network connection or will
not be able to find a network.
Missing terminators may not present visible problems on a small network.
Missing terminators cause data retransmits on smaller systems, eventually
appearing as transmit time out errors or network errors.
Using a terminator with an incorrect rating. Coax uses 93 Ohm; UTP
must use 100 Ohm terminators.
A terminator's value in ohms depends on the impedance of the cable. the
cable's impedance and the terminator's value should always match.
Failed active hubs (or a port on that hub)
Cable lengths that exceed specifications (see
table). Twisted pair, cabled in a bus rather than a star, cannot
have more than ten NICs per segment. this number varies with different
ARCnet UTP installed in a bus configuration is generally used only in very
small networks of six nodes or less. this configuration has the major
drawback of halting the network if a single cable is disconnected. In an
ARCnet bus configuration, the network must be brought down to make any
changes or service to the ARCnet cards.
Coax connector not built and/or crimped correctly.
Twist-on connectors are responsible for more intermittent errors on
a network than most other failures because of their design.
Twist-on coax connectors became popular in the IBM 3XXX systems. these
systems used RG-62 coax cable and operated at 1.5 Mbps throughput. the
twist-on connectors are not recommended for use on any modern LAN cable
system because of the higher data rates employed.
Maximum ARCnet Cable Distances|
The primary characteristics of ARCnet are:
ARCnet operates at 2.5 Mbps throughput using a token-passing-bus
ARCnet is cabled with RG-62 A/U coax with 93 Ohm terminators or the
newer UTP with 100 Ohm terminators.
the maximum time it takes for the ARCnet signal to travel the length
of the network is 31 microseconds.
the maximum distance an ARCnet signal can travel between the two nodes
farthest away from each other is 20,000 feet.
the absolute maximum number of ARCnet nodes that can occupy a given
network segment is 255. An ARCnet segment consists of all cabling and
nodes that share a given network address.
We will be building a typical ARCnet coax cable for our
hands-on lab project for this lesson.
From your LAN-Wire materials kit locate the:
- 6' length of RG-62 A/U coax cable
- RG-62 A/U crimp-on BNC connectors (Qty. 3)
Set the Coax wire stripper to the RG-59 setting.
(RG-59 & RG-62 are the same diameter)
the dual-set blades of the coax wire stripper will strip the end of
the coax cable with a precise two-stage cut. Strip the ends of your RG-62
cable using a rotary motion with the coax stripper.
Crimp the center pin of the BNC connector onto the end of the center
Slide the crimp ferrule over the coax cable.
Slip the body of the BNC connector under the braided shield as far as
it will go.
Slide the crimp ferrule over the braided shield up to the BNC body
Use the ratchet crimping tool to evenly and
completely crimp the ferrule to the BNC body.
Repeat for the other end of the cable.
Use a digital multimeter to test your cable for
shorts. (any Ohm reading other than infinity)
Using a BNC butt-splice and terminating resistor,
also test for a perfect 93 Ohm reading.