Token Bus Network: IEEE 802.4 Definition

A token bus network is a LAN topology where stations are logically arranged in a ring but physically connected to a shared bus, and a special frame called a token controls which station may transmit. IEEE 802.4 is the standard that defines this token-passing bus access method and its physical layer specifications. This article explains how it works, why it existed, and how it compares to Ethernet and token ring.

Formula

The token bus protocol uses a token that circulates among active stations in a logical ring. The key performance metric is the token rotation time (TTRT), which is the time taken for the token to complete one full cycle. The overhead per cycle is the time spent on token passing and control frames, not on data. A simplified formula for the maximum usable bandwidth is:

Max throughput (%) = 100 × (1 − (n × overhead_per_station) / TTRT)

Where:

• n = number of active stations on the logical ring
• overhead_per_station = time for token transmission and processing per station (assumed 0.05 ms in the calculator)
• TTRT = target token rotation time in milliseconds

This is a simplified model; real 802.4 networks have additional timing parameters like slot time and ring maintenance overhead.

Worked Example

Using the default values: TTRT = 10 ms, n = 20 stations.

1. Overhead per cycle = 10 ms × 0.05 = 0.5 ms
2. Max throughput = 100 × (1 − (20 × 0.05) / 10) = 100 × (1 − 1/10) = 90%

So with 20 stations and a 10 ms target rotation time, the theoretical maximum data throughput is 90% of the raw bandwidth. The calculator shows this as 90.00%.

Background: Why Token Bus Was Created

In the early 1980s, Ethernet (IEEE 802.3) used a bus topology with CSMA/CD (Carrier Sense Multiple Access with Collision Detection). This method is simple but has a critical flaw: it is probabilistic — stations must compete for the medium, and collisions can occur, causing delays that are not predictable. For factory automation and process control, engineers needed a network with deterministic timing: they had to know the maximum time before a station could transmit.

Token bus was developed by the IEEE 802.4 working group to provide this determinism while still using the inexpensive and robust coaxial cable bus that factories already used. The standard was published in 1985 and later became part of the MAP (Manufacturing Automation Protocol) initiative by General Motors, which aimed to standardize factory floor communications.

How Token Bus Works

The physical topology is a bus (a single cable with terminators at both ends). All stations are connected to this bus, but they are organized into a logical ring — an ordered list of station addresses. The token, a special frame, is passed from one station to the next in this logical order, regardless of their physical positions on the cable.

When a station receives the token, it may transmit data frames for a limited time (the token holding time). After that, it passes the token to its logical successor. If a station has no data, it immediately forwards the token. This guarantees that each station gets a chance to transmit within a bounded time.

Key features of the protocol:

• Token passing: only the token holder can transmit data frames.
• Logical ring maintenance: stations can join or leave the ring dynamically, which requires a complex set of control frames (solicit-successor, set-successor, etc.).
• Priorities: 802.4 defined up to 8 priority classes, allowing time-critical traffic to be sent first.
• Physical layers: the standard specified several options, including 1 Mbps, 5 Mbps, and 10 Mbps over 75-ohm coaxial cable (broadband or carrierband).

Common Problems and Why It Failed

Token bus was technically sound but commercially unsuccessful. The most significant problems were:

• Complexity: The ring maintenance protocol was far more complex than Ethernet’s simple collision detection. This made hardware expensive and debugging difficult.
• Cost: Token bus controllers required more logic and memory, making them significantly pricier than Ethernet chips.
• Slower speeds: While Ethernet quickly moved to 10 Mbps and beyond, token bus was stuck at 10 Mbps maximum, and the overhead of token passing reduced effective throughput.
• Physical limitations: The broadband variant required a headend and frequency translators, complicating installation.

By the early 1990s, Ethernet’s cost dropped dramatically, and its speed increased. The deterministic timing issue was addressed by switches and later by standards like Time-Sensitive Networking (TSN). Token bus became obsolete and is no longer used in new installations.

Comparison: Token Bus vs. Token Ring vs. Ethernet

It’s easy to confuse token bus (802.4) with token ring (802.5). Here’s a clear comparison:

• Physical topology: Token bus uses a bus (all stations share a cable); token ring uses a physical ring (or star with a hub that simulates a ring).
• Token passing logic: In token ring, the token passes to the next station in physical order. In token bus, the token passes to the next station in logical order, which may not match physical positions.
• Failure handling: Token ring had a monitor station to detect lost tokens. Token bus had a more complex distributed ring maintenance.
• Ethernet: Uses CSMA/CD (or switched full-duplex) and has no token. It is probabilistic but much simpler and faster.

In practice, token ring was more successful in office environments (IBM’s choice), while token bus was promoted for factories. Both lost to Ethernet.

Costs and Implementation Considerations

If you are researching historical systems, the costs were significant. Token bus interface cards cost several times more than Ethernet cards. The cabling (75-ohm coaxial) was cheaper than twisted-pair, but the active components (headend, modems) added cost. For a factory with hundreds of nodes, the total cost was often prohibitive.

Today, you would not implement a token bus network for a new project. If you encounter legacy equipment, you would likely need to replace it with Ethernet-based industrial protocols like PROFINET, EtherNet/IP, or EtherCAT, which offer deterministic behavior over standard Ethernet hardware.

Alternatives and Legacy

The deterministic timing requirement that motivated 802.4 is now met by:

• Switched Ethernet with full-duplex links eliminates collisions.
• Time-Sensitive Networking (IEEE 802.1Qbv) provides scheduled traffic and bounded latency.
• Industrial Ethernet protocols like PROFINET IRT and EtherCAT use special scheduling to achieve microsecond-level determinism.

Thus, the goal of 802.4 lives on, but the technology is gone.

Frequently Asked Questions

Is token bus the same as token ring?

No. They share the token-passing concept but differ in topology and standard. Token bus is IEEE 802.4 and uses a physical bus with a logical ring. Token ring is IEEE 802.5 and uses a physical ring (or star-wired ring).

Why was token bus used in factories?

Factories needed a network with predictable maximum delay for control messages. Token bus guaranteed a station would transmit within a known time, unlike Ethernet’s CSMA/CD, which could suffer collisions and indefinite delays under heavy load.

What speeds did IEEE 802.4 support?

The standard defined 1 Mbps, 5 Mbps, and 10 Mbps over 75-ohm coaxial cable. The 10 Mbps option was the most common for MAP.

Is token bus still used today?

No. It is obsolete. New industrial networks use Ethernet-based solutions with deterministic scheduling. You might find token bus in very old factory equipment, but it is no longer installed.

What is the difference between a logical ring and a physical ring?

A physical ring has each station connected to the next in a closed loop. A logical ring is an ordered list of addresses; the token passes from one address to the next, but the physical connections can be a bus or star. This allows the physical layout to be simpler while maintaining the token order.

Related

• Fix Steam Cloud Sync Conflicts Without Losing Saves
• Big Walk Platforms Pricing & Crossplay Guide
• Mac Can’t Connect to Apple Media Services? Fix

in t | Token rotation time | ms | 10 | 1 | 10000
in n | Active stations | count | 20 | 2 | 1000
out overhead | Overhead per cycle | ms | t * 0.05 | 2
out max_throughput | Max throughput | % | 100 * (1 - (n * 0.05) / t) | 2