To fix a failing network hub, power-cycle the unit, verify the PSU and link LEDs, swap suspect Ethernet cables, test each port individually, and remove any loops. If collisions or heat persist, reduce the device count or replace the hub with a Layer 2 switch, which is the modern equivalent.
Symptoms
- All link lights are dark or every port LED blinks in unison at a steady rhythm
- Devices connect at 10 Mbps half-duplex and report constant packet loss
- Collision LED stays lit or flickers heavily even with light traffic
- Hub casing is hot to the touch and connections drop after 15 to 30 minutes
- One or more ports no longer register a link even with a known-good cable
Common Causes
Power supply degradation
Wall-wart adapters on older hubs lose regulation over time. Undervoltage causes ports to drop links intermittently before the unit fails outright.
Damaged or miswired Ethernet cabling
Bent RJ45 clips, kinked runs, and untwisted pairs at the plug cause CRC errors that look like a hub fault. Cat3 patch cables also cap you at 10 Mbps.
Collision domain saturation
Hubs share one collision domain across all ports. Once utilisation passes roughly 30 to 40 percent, throughput collapses under retransmissions.
Broadcast storm from a loop
Two ports patched into the same downstream device, or a cable plugged back into the hub, floods every port. Hubs cannot run STP to break the loop.
Thermal failure of internal components
Dust-blocked vents raise the internal temperature of the PHY chips. Solder joints fatigue and individual ports stop passing traffic.
Step-by-Step Fix
- Confirm the hub is actually the failure point
Bypass the hub: connect two known-good laptops directly with a single Ethernet cable and confirm they link at expected speed. If the direct link works, the hub, its PSU, or cabling is the fault domain. If the direct link fails, look upstream first. - Verify power delivery and thermal state
Measure the adapter output with a multimeter against the label rating; anything more than 10 percent low will cause instability. Feel the chassis. If it is uncomfortably warm, unplug for 15 minutes, clear vents with compressed air, and relocate the hub off carpet or stacked equipment. - Read the LEDs before touching anything else
A steady link LED means physical connectivity. A blinking activity LED means frames are moving. If a Collision LED is lit constantly, you have either a duplex mismatch on a connected NIC or a loop. Note which ports are dark and which show errors, then work port by port. - Swap cables and rotate ports systematically
Replace each suspect patch cable with a tested Cat5e or Cat6 cable, no longer than 100 metres end-to-end. Move a working device through every port to map dead ports. Mark faulty ports with tape so they are not reused after the incident is closed. - Break loops and reduce broadcast traffic
Trace every uplink. If two cables land on the same downstream switch or wall jack, pull one immediately. Disconnect all devices, then reconnect one at a time while watching the collision LED; the port that lights it up is the offender. - Cap the number of active devices
On a shared-media hub, keep concurrent talkers low. Move high-bandwidth endpoints such as NAS units, backup targets, and video encoders onto a switched port. Leave printers, console servers, and low-chatter IoT devices on the hub if you must keep it. - Update firmware where possible, then plan replacement
Most classic hubs are unmanaged and have no firmware. If yours is a managed hub or hub-switch hybrid, apply the vendor's current firmware. Long term, replace with an unmanaged gigabit switch: same cabling, no collision domain, and full-duplex on every port.
Common hub symptoms mapped to likely cause and first action
| Symptom | Likely cause | First action |
|---|---|---|
| All LEDs dark | PSU failure or dead outlet | Test adapter voltage, try another outlet |
| Collision LED always on | Loop or duplex mismatch | Disconnect ports one by one to isolate |
| Single port dead | PHY damage from ESD or surge | Move device to another port, mark port faulty |
| Speed stuck at 10 Mbps half-duplex | Cat3 cabling or NIC forced setting | Replace with Cat5e, set NIC to auto-negotiate |
| Random drops after warm-up | Thermal fault | Improve airflow, plan replacement |
| Everything slow under load | Collision domain saturated | Move heavy talkers to a switch |
Prevention
- Keep the hub in a ventilated location and clean vents quarterly with compressed air
- Label every patch cable at both ends so loops are visible during changes
- Retire hubs older than roughly 8 to 10 years before intermittent failures start
- Document the port map so a dead port is spotted immediately, not after outages
FAQ
Is a network hub still worth repairing in 2025?
Usually not. Unmanaged gigabit switches now cost less than a replacement PSU for most legacy hubs, and they eliminate the collision domain entirely. Repair only makes sense if the hub is part of a certified lab setup, a legacy industrial system, or a compliance-locked environment where hardware changes require re-validation.
How can I tell the difference between a hub fault and a cable fault?
Move the same device through every port using the same cable, then repeat with a second cable. If the failure follows the cable, it is cabling. If failures cluster on specific ports regardless of cable, the hub's PHY is damaged. A cable tester or the loopback function on a managed NIC will confirm cabling in seconds.
Why do I see constant collisions on a hub with only a few devices connected?
Collisions on a lightly used hub almost always point to a duplex mismatch, a physical loop, or a failing NIC that transmits without listening. Force every connected NIC to auto-negotiate, then disconnect devices one at a time. The collision LED going dark identifies the offender.
Contact our network support team to plan a hub-to-switch migration and remove collision-domain risk from your LAN.