FAQ
TL;DR: Up to 65 % of home 100 Mbps link failures stem from mis-wired pairs [Fluke Networks, 2020]; “pairs must stay together” [Elektroda, Samuraj, post #2597911] A correctly crimped Cat 5e crossover (T568A↔T568B) should carry 100 Mbps over 12 m. Why it matters: This FAQ helps DIY users fix speed-limited Ethernet runs without tearing walls open.
Quick Facts
• 100Base-TX uses only pins 1-2 & 3-6; the other four conductors stay unused [IEEE 802.3u].
• Max run: 100 m at 100 Mbps; every 10 m of untwisted pair can add ≈3 dB loss [TI, App Note 2021].
• Certified Cat 5e cable supports 100 MHz signalling and 1000 Mbps to 100 m [TIA-568.2-D].
• Basic LED cable tester ≈ US$10–25; certifier (Fluke DSP) ≈ US$4 000 [Amazon Pricing, 2023].
• Auto-MDI/MDI-X NICs eliminate crossover needs since 2006 chipsets [Intel 82567 Datasheet].
Why does my Ethernet cable work at 10 Mbps but not 100 Mbps?
10 Mbps (10Base-T) tolerates untwisted or mis-paired conductors because it uses Manchester coding at 10 MHz. 100Base-TX moves data at 125 MHz across two balanced pairs; split pairs introduce crosstalk that kills the higher-frequency signal [Elektroda, Samuraj, post #2597911] If pairs 1-2 and 3-6 are not kept together, many cards down-shift to 10 Mbps or drop link altogether.
How can I test a hidden cable without removing it from the wall?
- Short pins 1-8 together at one outlet using a scrap RJ-45.
- At the far end, measure continuity for each pin with an ohmmeter; open or crossed readings reveal faults [Elektroda, tronic1, post #2591922]
- A US$20 LED tester speeds this by flashing each pair [Amazon Pricing, 2023].
What is the correct crossover pinout for 100Base-TX?
Plug A follows T568A; plug B follows T568B: 1↔3 (white-green↔white-orange) and 2↔6 (green↔orange). Pins 4-5 & 7-8 stay straight-through [IEEE 802.3u].
Do wire colours matter if the pairs stay intact?
Colour itself is cosmetic, but each twisted pair must stay together electrically. Swapping colours between pairs splits the twists, raising near-end crosstalk and causing 100 Mbps failure [Elektroda, menek, post #2597919]
Will Cat 3 cable lock me to 10 Mbps?
Yes. Cat 3 supports only 16 MHz bandwidth; 100Base-TX demands 100 MHz. The link may negotiate 10 Mbps but cannot hold 100 Mbps, regardless of perfect pinout [Category 3 Cable, TIA].
What length margin do I have on a 12 m run?
12 m sits at 12 % of the 100 m limit, leaving >7 dB insertion-loss headroom. If 100 Mbps still fails, focus on wiring errors or damaged conductors, not length [TI, App Note 2021].
Can I crimp RJ-45 plugs with a screwdriver like in the thread?
It works for 10 Mbps, but uneven force often leaves pins partially seated, raising resistance and reflection at 100 MHz. A US$15 ratchet crimper delivers uniform 2-ton pressure and reduces 100 Mbps failures by 40 % in Fluke lab tests [Fluke Networks, 2020].
What do the NIC LEDs tell me during troubleshooting?
A steady green often means link detected; blinking amber shows traffic. If green appears but the card negotiates 10 Mbps, suspect pair mis-wiring; if no green, look for open or shorted pins [Elektroda, creepingdeath, post #2597859]
Could external noise block 100 Mbps but allow 10 Mbps?
Yes. EMI from fluorescent ballasts or ungrounded cable shields can inject high-frequency noise. 100Base-TX fails first because its eye diagram margin is smaller. One user fixed this by grounding the foil shield on a 120 m run [Elektroda, bunkrowiec, post #2595181]
Does crossover direction matter—PC-A or PC-B?
No. A crossover is symmetrical: pin 1 maps to 3 on both ends, so either plug can sit at either PC [Elektroda, Samuraj, post #2599299]
Do modern PCs still need crossover cables?
Most NICs since 2006 include auto-MDI/MDI-X. They auto-swap pairs, letting you use straight cables even PC-to-PC. Check datasheets: Intel 82567 and Realtek 8111 chips support it [Intel 82567 Datasheet].
How can I spot counterfeit Cat 5e cable?
Look for missing UL or ETL markings, copper-clad aluminum conductors, and no “24 AWG SOLID” print. Cut a short piece: aluminum core is silver, true copper is reddish. Counterfeit cable often fails 100 MHz sweep tests by 30 dB [Fluke Networks, 2020].
