Out-of-spec downstream power + documented 48hr degradation — requesting line technician

Category: Your Home Network Topic/sub-category: Internet (or "Connection Issues" if available)

Subject: Out-of-spec downstream power + documented 48hr degradation — requesting line technician

Body:

Service plan: Gigabit Extra (~941/43 Mbps) Modem: Arris Surfboard S33v2 (DOCSIS 3.1), firmware TB01.03.001.11 Location: Mount Laurel, NJ

I have 48 hours of continuous documented evidence showing intermittent service degradation with root cause on the plant side. Summary below; full diagnostic logs available via DM.

Modem signal readings (current):

Downstream Signal Power: -11 dBmV (DOCSIS spec: -7 to +7 dBmV — out of spec by 4 dB)
Downstream Signal SNR: 35 dB (minimum for QAM256: 36 dB — below spec)
Downstream Frequency: 567 MHz

Throughput delivery (48hr continuous speedtest sampling, wired host):

Measured: 512–704 Mbps down, 32–38 Mbps up
Delivered: 54–75% of provisioned plan
Consistent under-delivery across all sampling windows, not transient

Latency analysis (continuous ping monitoring, 5-second interval to 1.1.1.1 and 8.8.8.8):

Idle baseline: 16–21ms
Spikes to 244ms under load
9 of 10 recent 5-minute windows flagged with elevated latency (20–46ms average)

Traceroute analysis (every 2 hours):

Hop 2 (Comcast CMTS): 14–29ms latency variance — modem-to-node segment
Hop 9 (Newark peering): occasional packet loss
Latency variance originates at Hop 2 = plant-side, not in-home

Bufferbloat testing (Waveform.com, wired):

Current grade: A under controlled conditions (SQM on home router)
Peak loaded latency observed: 244ms during degradation events
Pattern: ~80% of time within spec, ~20% degraded — consistent with intermittent plant issue

In-home equipment ruled out:

All tests run from wired host behind professional-grade gateway (Ubiquiti UDM Pro)
Smart Queue Management (CAKE) enabled on WAN with confirmed correct algorithm
DNS, switching, and WiFi all verified clean
Throughput and latency degradation occur regardless of in-home load

Diagnosis: Downstream power 4 dB below DOCSIS spec, combined with SNR below QAM256 minimum, is consistent with attenuation on the drop or tap side. The intermittent nature of latency spikes and throughput drops, combined with Hop 2 variance, indicates plant-side instability — likely corroded F-connector, damaged drop, water-damaged tap port, or attenuation from upstream amplifier.

Request:

Please dispatch a line technician for plant-side signal investigation at the tap and amplifier serving my address.

I am specifically not requesting:

A modem swap (modem is reporting accurately and functioning correctly given signal received)
An in-home technician visit (issue is documented upstream of demarc)
A modem reboot (already verified; signal levels persist)

Happy to provide full diagnostic logs (PDF report with continuous ping data, MTR traces, speedtest history, and bufferbloat measurements) via DM upon request.

Responses

Oldest First

Selected Oldest First

XfinityPeterH

Official Employee

•

2.8K Messages

1 day ago

user_cvxm8s This is an incredibly thorough breakdown, and honestly it is awesome to see how tools like AI can help pull together such detailed analysis and point toward a likely root cause. You’ve clearly put a lot of effort into isolating variables and narrowing this down.

That said, while your findings do suggest a potential issue outside the home, we do still need to follow the standard troubleshooting path step by step. Even when things appear to point upstream, our process requires starting with the basics and working forward so we can properly document everything and ensure nothing is missed along the way.

Because of that, we aren’t able to directly request a line technician right away. The next step would be to go through the standard troubleshooting flow, which typically begins with verifying signal levels, connections, and account diagnostics on our side. Once those steps are completed and documented, we can absolutely look at escalating this further, including getting the right type of technician involved if the issue continues to point to the network or plant.

I definitely understand the goal here and appreciate how much detail you’ve provided. If you’re open to it, we can start with that initial process together and work toward getting this escalated appropriately.

XfinityDilary

Official Employee

•

3.3K Messages

I appreciate the detailed response, user_cvxm8s. Let’s move this to a private message so I can check the signal levels on my end. Could you please send our team a Direct Message with your name and service address? Our team can take a further look.

To send a Direct Message:

Click "Sign In" if necessary

• Click the "Direct Messaging chat" icon

• Click the "New message" (pencil and paper) icon

• Type "Xfinity Support" in the "To:" line and select "Xfinity Support" from the drop-down list that appears. The "Xfinity Support" graphic replaces the "To:" line

• Type your message in the text area near the bottom of the window

• Press Enter to send it

I am an Official Xfinity Employee.
Official Employees are from multiple teams within Xfinity: CARE, Product, Leadership.
We ask that you post publicly so people with similar questions may benefit from the conversation.
Was your question answered? Please, mark a reply as the Accepted Answer. tick

15 hours ago

user_cvxm8s

Visitor

•

2 Messages

Quick correction on plan tier before I share the rest — I’m subscribed to Gigabit X2 (2,000 Mbps down / 200 Mbps up), not the standard Gigabit. The percentages in my earlier post understated the under-delivery. Updated numbers below.

Understood on the troubleshooting flow, happy to continue working through it with you.

A clarification on the framing of the earlier reply: the analysis is standard network diagnostics — modem-reported DOCSIS signal levels measured against published specification, continuous ICMP monitoring at 5-second intervals against multiple anchor hosts, periodic MTR traces to identify the hop at which latency variance originates, and bufferbloat testing under controlled load. The tooling automates collection across a 48-hour window; the methodology and conclusions are what any network engineer would apply when triaging intermittent service degradation. The -11 dBmV downstream power and 35 dB SNR are direct readings from the modem’s diagnostic interface, not interpretations.

For additional context, here is the statistical summary from continuous monitoring, with percentages corrected against the 2 Gbps plan.

SERVICE DEGRADATION EVIDENCE — Wired Testing, 14 Samples Over 16 Hours

Plan: Gigabit X2 (2,000 Mbps down / 200 Mbps up)
Equipment: Enterprise gateway, Cat6 to Xfinity modem (bridge mode)
Test method: Wired host with 2.5GbE NIC, speedtest-cli, continuous ping, traceroute every 2h

In-home capability to deliver 2 Gbps confirmed:

• Modem (Arris S33v2): 2 Gbps capable, DOCSIS 3.1, OFDM channels active
• Gateway WAN port: 2.5GbE / 10GbE capable
• Modem-to-gateway link: Cat6, 2.5GbE negotiated, link clean
• Test host NIC: 2.5GbE confirmed, eliminates client-side bottleneck

SPEED TEST RESULTS (14 tests, 7 AM – 9 PM):

• Plan: 2,000 down / 200 up
• Best: 805 Mbps down (40% of plan)
• Median: 683 Mbps down (34% of plan)
• Worst: 149 Mbps down (7.5% of plan)
• Average: 603 Mbps down (30% of plan)
• Standard deviation: 204 Mbps
• 100% of tests delivered below 80% of plan
• 93% of tests delivered below 50% of plan
• 50% of tests delivered below 30% of plan
• Morning average (7–8 AM): 682 Mbps (34% of plan)
• Evening average (5–9 PM): 403 Mbps (20% of plan) — 41% worse than morning

Not a single test came within 1,000 Mbps of provisioned plan, on a wired connection through known 2 Gbps-capable equipment.

LATENCY UNDER LOAD (Bufferbloat):

• Idle: 18.8 ms average, 28.7 ms peak
• Under load: 39.5 ms average, 263.5 ms peak
• Bloat added: +20.7 ms average, +244.7 ms peak

Grade: B (borderline C). SQM/CAKE is active on my side — without it, this would be worse.

TRACEROUTE HOP ANALYSIS (13 samples):

• Hop 1 — Gateway (local): 0.4 ms — clean
• Hop 2 — 10.27.227.130/131: 2–29 ms (std dev 10.1 ms) — FIRST PROBLEM HOP
• Hop 3 — po-58-rur301/302.hillsboro.nj: 10–19 ms — Hillsboro NJ node
• Hops 7–8 — Newark backbone: 11–23 ms — normal
• Hop 9 — Newark peering: packet loss observed
• Hop 10 — Cloudflare peering: 22–171 ms — spike at handoff

Hop 2 is the CMTS or the first node after my modem. Latency variance of 10.1 ms standard deviation at this hop — with a clean 0.4 ms hop 1 — points to plant-side congestion or signal issues between my drop and the node.

WHAT I’M ASKING FOR:

1. CMTS-side diagnostics on the node serving my address — specifically upstream SNR, uncorrectable codewords, and T3/T4 timeout counts on my modem’s upstream channels.
2. Node utilization data for the Hillsboro, NJ CMTS (10.27.227.x) — is this node oversubscribed during evening hours?
3. Plant maintenance history — any recent work, known ingress issues, or scheduled node splits in my service area?
4. Confirmation that my account is provisioned for the 2 Gbps tier I’m being billed for. The -11 dBmV downstream power and 35 dB SNR are well below what’s required to sustain 2 Gbps service reliably; this signal level is barely adequate for 1 Gbps.

I’ve eliminated my home network and equipment as the cause:

• Wired testing throughout (no WiFi variable)
• 2.5GbE end-to-end (no client-side bottleneck)
• Local hop is 0.4 ms with zero loss
• SQM/CAKE active on my gateway
• Modem and gateway both confirmed 2 Gbps capable
• Problem appears at hop 2 — your first equipment after my modem

The 41% evening degradation pattern, combined with downstream signal levels insufficient to support 2 Gbps, indicates this is both a plant signal issue and likely a node capacity / utilization issue. A modem swap will not fix either — the signal arriving at my drop and the CMTS-side provisioning both need investigation.

Happy to provide raw logs or specific measurement detail. Account details available via DM.

Equipment: customer-owned modem, enterprise gateway, 2.5GbE wired test environment
Service tier: Gigabit X2

14 hours ago

Your Home Network

user_cvxm8s

Out-of-spec downstream power + documented 48hr degradation — requesting line technician

XfinityPeterH

XfinityDilary

user_cvxm8s

New to the Community?