If your business has an NBN fault that NBN has already attended and closed, and a cabler or electrician has already looked at, you are not alone. This is one of the most common situations we are called into — a commercial premises where everyone has been out, everyone has tested, and everyone has concluded the line is fine, while the business is still losing transactions and cloud access.

Location: A shop inside Eaton Fair Shopping Centre, Western Australia

Service type: NBN FTTN (VDSL2-17a)

Outcome: Line restored from 10.4 / 15.1 Mbps max attainable to 61.9 / 29.0 Mbps, with downstream SNR margin improving from 7.2 dB to 21.4 dB.

Figure 1. NBN's own line-rate report, captured pre-fix. Downstream actual is 5.2 Mbps against an assured 26.1 Mbps. Upstream actual is 11.8 Mbps against an assured 5.1 Mbps — the inversion that became our primary diagnostic clue.

The Complaint

Our Client reported the following symptoms:

• Download speeds of 5-10 Mbps on a plan with an assured line rate of 26Mbps / 5Mbps.

• Upload speeds of approximately 15 Mbps—faster than download

• Frequent dropouts disrupting EFTPOS transactions and cloud applications

• General instability across all connected devices

• nbn has already attended and confirmed their end is OK

The asymmetry was the key diagnostic clue. On any VDSL-based NBN connection (FTTN, FTTB), download speeds should always substantially exceed upload speeds. When the inverse occurs, it indicates line impairment—the physical medium is degrading the signal before it reaches the modem.

Figure 2. Pre-fix DSL statistics captured by our test meter while onsite. Note the remote FEC counter at 14,058 — the node was correcting a continuous torrent of errored frames arriving from the customer's line and EWL (Estimated Wire Length) reads 714 m.

The TechTV Diagnostic Process

Step 1: Line Health Assessment

We connected our EXFO MaxTester 635 to the line inside the shop's communications rack. The 635 gives us far more than a speed test — it reports per-bin channel response, DSL counters, SNR margin, attainable rate, and estimated loop length, all of which are required to distinguish a line fault from a CPE, plan, or provisioning issue.

Figure 3. Pre-fix connection summary from the MaxTester 635. Downstream SNR margin is 7.20 dB and upstream is 7.30 dB — both at the edge of VDSL2's minimum operating window. Maximum attainable bitrate is 10.374 Mbps downstream and 15.099 Mbps upstream — asymmetric in the wrong direction.

At this point the line was visibly in trouble. Downstream SNR margin of 7.2 dB is close to the point at which VDSL2 will drop sync entirely. Maximum attainable downstream was 10.374 Mbps — below even the assured rate, and lower than attainable upstream at 15.099 Mbps.

Step 2: Hlog Per Bin Analysis

The critical diagnostic tool in this case was the Hlog (Hybrid Log) per bin graph.

What is Hlog?

Hlog is the channel transfer function estimated by the VDSL2 modem during training. It measures the attenuation of the line across individual subcarriers (bins), each representing a narrow frequency band. The graph plots signal loss in dB against frequency in kHz. On a healthy line, Hlog is a smooth, monotonic curve that decays predictably as frequency rises, with a characteristic high-frequency roll-off. It is not linear — it is roughly logarithmic in shape — but it should be free of notches, plateaus, and discontinuities.

What we observed (pre-fix):

Figure 4. Pre-fix Hlog per bin, captured by the EXFO MaxTester 635. The line does not extend to the full VDSL2-17a band.

The pre-fix Hlog was not merely "noisy." It displayed three specific structural defects:

1. A non-monotonic plateau between roughly 3.5 and 5.2 MHz, where attenuation stops increasing with frequency as it should

2. A complete null between roughly 6 and 7 MHz, where no usable signal was returned

3. Notch structure between 8 and 11 MHz, with alternating peaks and dips

This signature is characteristic of an impedance discontinuity in the transmission line. The signal reaches a point where the electrical characteristics of the pair change abruptly, and part of the energy is reflected back. The reflected energy interferes with the forward signal at specific frequencies — producing exactly the plateaus, nulls, and notches visible in Figure 4. It is the frequency-domain fingerprint of:

• A split pair

• A bridge tap

• A poor termination or corroded joint

• Severe water ingress in the cable

The line was effectively "leaking" signal at specific frequencies, forcing the modem to reduce its sync rate to maintain stability. The downstream SNR margin of 7.2 dB was the modem's attempt to hold on.

The Root Cause: A Split Pair on a 10-Pair Cable

We traced the cabling from the MDF (Main Distribution Frame) through to the commercial premises' IDF (Intermediate Distribution Frame) and then to the communications rack. The building was wired with a 10-pair copper cable — standard for commercial installations where multiple services may be required.

Upon inspection of the terminations, we identified the fault: a split pair.

Figure 5. The incoming 10-pair cable after cutting and stripping back to identify the correct wires. Note the very subtle twist rate on the individual telephone pairs, particularly compared to Ethernet cabling — this is why split pairs occur, especially when a technician is unfamiliar with the cable type. In this case, the white wire from the white/brown pair was in use and that was the entire cause of the fault.

What Is a Split Pair?

In twisted pair cabling, data is transmitted across two wires that are twisted together. The twist in telephone cable above is very subtle compared to ethernet cabling - so it is possible to split the pairs especially if you are inexperienced. Check the photo above to see

• Provides common-mode rejection of external electromagnetic interference

• Maintains balanced transmission across the pair

• Prevents crosstalk between adjacent pairs

A split pair occurs when the termination incorrectly pairs wires from two different twisted pairs. For example:

Correct PairingSplit Pair (Fault)Pair 1: Blue / White-BlueBlue (Pair 1) / Orange (Pair 2)Pair 2: Orange / White-OrangeWhite-Blue (Pair 1) / White-Orange (Pair 2)

The result of a split pair is:

• Severe impedance mismatch at the termination point

• Increased crosstalk between the mis-paired conductors

• Signal reflection at the split

• Unbalanced transmission across the pair

This perfectly explains the pre-fix Hlog signature. A split pair is an impedance discontinuity — the point at which balance is lost becomes the reflection point, and the reflected signal produces the plateau, null, and notch structure we observed between 3.5 and 11 MHz.

It is worth noting that with the bad reputation FTTN rightly carries, the technology is surprisingly resilient. A VDSL2 modem will sometimes train on a severely compromised pair — occasionally even where only one conductor is effectively connected. This is precisely why a split pair can sit undetected for years: the modem syncs, the service appears to work, and the symptom presents only as poor speed or intermittent dropouts rather than a hard failure.

The Results

The transformation was immediate and definitive! Maximum attainable downstream bitrate rose from 10.374 Mbps to 61.852 Mbps — thats a 6x increase in download speed!

The Hlog graph shifted from a jagged, non-monotonic pattern with a null and plateau to a clean, monotonic decay extending to 17.5 MHz. The downstream SNR margin rose from 7.20 dB to 21.40 dB — a +14.2 dB improvement.

Connection Summary (Post-Fix)

Figure 6. Post-fix connection summary. Downstream SNR margin has risen to 21.40 dB and upstream to 19.90 dB. Maximum attainable bitrate is now 61.852 Mbps downstream and 28.999 Mbps upstream.

DSL Statistics (Post-Fix)

Figure 7. Post-fix DSL statistics. Remote FEC has collapsed from 14,058 to 0. KL0 has improved from 18.9 dB to 15.6 dB. EWL (wire length or distance to the node) has corrected from 714 m to 589 m

Hlog Per Bin Graph (Post-Fix)

Figure 8. Post-fix Hlog per bin. The curve is now smooth, monotonic, and free of notches, extending cleanly to the full VDSL2-17a band. Compare directly with Figure 4.

Conclusion

A commercial NBN service that had been misdiagnosed as a provisioning or CPE problem was traced to a single split pair on a 10-pair copper cable. The fault produced a reflection signature visible in the Hlog per-bin graph, elevated remote FEC counters at the node, and an inflated Estimated Wire Length — all of which resolved immediately once the pairing was corrected.

If your business is experiencing NBN issues that have survived multiple technician visits, speed tests, and modem swaps, the fault may be in the copper itself.

TechTV has been resolving complex NBN line faults for businesses and managed service providers across Perth, WA since 2015. If you need a specialist who can find what others miss, contact us at office@techtv.au.

Figure 9. TechTV's fully equipped work van — complete with Starlink — on the way back to Perth after fixing the fault described above. Yes, that's satellite internet on the roof of an NBN fault-fixing van. Yes, we're aware of the irony.