A 100-meter conveyor line with 30 PROFINET drives requires 30 Ethernet cables. In a star topology, each cable runs from a central switch to a single drive — that is 30 home runs, averaging 50 meters each, for 1,500 meters of installed Ethernet cable. In a daisy-chain topology, the same line runs one cable from the switch to Drive 1, a short jumper from Drive 1 to Drive 2, and so on down the line — roughly 200 meters total, a 7:1 reduction. The cable savings are obvious. What is less obvious is what that loop of drives costs you in latency, fault tolerance, and commissioning complexity.
This article compares the two topologies for real conveyor installations and explains where hybrids make more sense than either extreme.
Star Topology: Deterministic Latency, Maximum Copper
In a star topology, each drive connects to a dedicated port on an industrial Ethernet switch. The electrical path from PLC to any drive is exactly two cable segments: PLC to switch, switch to drive. Every drive experiences the same transmission latency — one switch forwarding delay of 5–10 µs plus the cable propagation delay of roughly 5 ns per meter. For a 100-meter cable run, the total latency from PLC to drive is under 15 µs, and it is identical for every drive on the line.
The advantage is predictability. When the PLC issues a speed change to 30 drives simultaneously, all 30 PROFINET telegrams arrive within the same update cycle. No drive waits longer than any other. For coordinated multi-axis conveyors — where cartons transfer from one zone to the next at a precise handoff point — this determinism ensures that Drive 17 does not receive its speed command 100 µs after Drive 16, which would create a momentary speed mismatch at the transfer point.
The disadvantage is the cable plant. Thirty drives need 30 switch ports, which means either one large 32-port managed switch or multiple smaller switches distributed along the line. Each switch needs power, an enclosure, and fiber or copper uplinks back to the main control panel. On a long outdoor conveyor — a mining stacker, a ship-loading boom — running 30 individual Ethernet cables back to a climate-controlled switch cabinet is often more expensive than the drives themselves.
Daisy-Chain: Minimum Copper, Cumulative Risks
The G115D PROFINET drives include a built-in 2-port switch. Port 1 receives the incoming Ethernet connection, Port 2 forwards it to the next drive. Each drive becomes a network node that passes traffic through — the drive's own PROFINET telegrams terminate at the drive, while traffic addressed to downstream drives passes through at wire speed with less than 2 µs of added latency per hop.
A 30-drive daisy-chain creates 30 hop points. The last drive in the chain is 30 hops from the PLC, accumulating roughly 60 µs of switching latency plus cable propagation delay. PROFINET RT (Real-Time) handles this gracefully — the protocol is designed for line topologies and tolerates up to 100 drives on a single chain with update cycles as low as 1 ms. The 60 µs of cumulative switching latency is a fraction of the 1 ms cycle, so it does not degrade control performance. What it does degrade is fault tolerance.
If Drive 12 loses power — or if a forklift snags the cable between Drive 12 and Drive 13 — every drive downstream of the break (Drives 13 through 30) loses communication with the PLC. The conveyor stops. In a star topology, losing one cable kills one drive. In a daisy-chain, losing one cable can kill an entire downstream segment. The G115D switch ASIC has a bypass mode for power-loss scenarios — if Drive 12 loses 400 V input but the switch ASIC still receives power from the upstream drive's Port 2 — but a severed cable has no bypass. The downstream segment is dark until the cable is repaired.
The Hybrid: Daisy-Chain Segments with Star Uplinks
The practical compromise for most conveyor lines is neither pure star nor pure daisy-chain. Divide the line into logical segments of 5–8 drives each. Within a segment, daisy-chain the drives. Connect the first drive of each segment back to a central switch via a dedicated uplink cable. The result: each segment behaves like a small daisy-chain with low latency accumulation (under 15 µs within the segment), and a cable failure inside one segment takes down only that segment — not the entire line.
For the 3 kW fanless PROFINET model on a 5-drive accumulation segment, this means the drives at the infeed, three accumulation zones, and the discharge share one uplink cable. If a pallet crushes the cable between Zone 3 and Zone 4, only Zone 4 and the discharge stop — Zone 1 through Zone 3 continue running on their own daisy-chain segment, feeding product up to the break point.
Hybrid topology also simplifies commissioning. Each segment can be brought online independently — commission the infeed segment, verify all 5 drives respond, then move to the next segment. Debugging a star network with 30 drives means finding which of 30 cables is faulty. Debugging a segmented daisy-chain means finding which segment is down (the switch tells you immediately which uplink port lost link), then walking that 5-drive segment to find the break.
Network Recovery Time: The Metric That Matters
When a cable breaks in a daisy-chain, the PROFINET controller detects the link-down event on the last reachable drive and marks the downstream devices as faulted. The recovery time — from cable break to the PLC recognizing lost devices — is typically 3–6 PROFINET update cycles, or 3–12 ms at a 2 ms cycle. The drives themselves coast or brake depending on their configured fault response.
With media redundancy protocol (MRP) configured on a ring topology — a special case of daisy-chain where the last drive connects back to the switch, forming a loop — a single cable break causes zero packet loss. The ring reconfigures in under 200 ms (MRP standard) or under 30 ms (MRPD, the PROFINET-specific fast variant). The G115D supports MRP on all PROFINET models, enabling true ring redundancy when the last drive's Port 2 returns to a second port on the same managed switch. For critical conveyor sections — a high-speed sorter, a merge unit — the ring topology eliminates the single-point-of-failure risk of a pure daisy-chain at the cost of one additional cable.
How many drives can I daisy-chain before latency becomes a problem?
For PROFINET RT with a 2 ms update cycle, 50 drives on a single chain is the practical limit — each hop adds under 2 µs, totaling under 100 µs of cumulative switching latency, which is 5% of the cycle time. Beyond 50 drives, consider splitting into two chains or stepping up to PROFINET IRT (Isochronous Real-Time) for sub-millisecond determinism. For conveyor applications running 4–8 ms update cycles, 50 drives are well within the safe margin.
Does daisy-chaining reduce the total cost that much?
On a 100-meter line with 30 drives, replacing 30 home-run cables (averaging 50 m each, total 1,500 m) with a single chain cable plus 29 short jumpers (averaging 3 m each, total roughly 200 m) saves roughly 1,300 meters of installed Ethernet cable and 24 switch ports. At industrial prices — roughly $2/m for PROFINET-rated cable and $50–100 per managed switch port — the cable savings alone cover the cost of 2–3 additional drives. Add the labor savings from pulling and terminating 29 fewer long cables, and daisy-chain is not just a convenience — it is a capital cost decision.
Daisy-chain topology saves copper. Star topology saves fault domains. A hybrid of short daisy-chain segments with dedicated star uplinks saves both — and gives you a network you can troubleshoot with a cable tester and a walkie-talkie instead of a network analyzer and three hours of downtime.
