How a $15 MoCA Filter Saved My 1 Gbps Internet from a Hidden Frequency War

March 07, 2026

I switched from Cogeco to oxio internet last week, and what should have been a straightforward modem swap turned into one of the most satisfying troubleshooting sessions I’ve had in a while. My shiny new Gemtek GT-DUO31 DOCSIS 3.1 modem was delivering painfully slow speeds despite paying for a 1 Gbps plan. Speed tests were coming back at a fraction of what I was paying for, pages were timing out, and video calls were dropping constantly. The old Cogeco modem had worked flawlessly for years, so I knew something was fundamentally wrong with the new setup.

The Symptoms That Didn’t Add Up

I logged into the modem’s admin page at 192.168.100.1 and immediately saw trouble. The signal levels were way out of spec:

Metric	Actual Value	Acceptable Range
Downstream OFDM Power	-14.7 to -16.0 dBmV	-7 to +7 dBmV
Upstream Transmit Power	50+ dBmV	Max 47 dBmV

The event logs were even worse. SYNC Timing Synchronization failures every 3-6 seconds, Lost MDD Timeouts, and constant channel loss/degradation events. The modem was stuck in a perpetual cycle of losing and re-establishing its connection. It was fighting to stay online and losing.

Critical (3) - SYNC Timing Synchronization failure
Warning (5) - Lost MDD Timeout
Critical (3) - 16 consecutive T3 timeouts
Warning (5) - OFDM Profile Change - Channel degraded

The Investigation

My first instinct was to blame the coax cabling. I had two splitters in the chain adding about 7 dB of total signal loss, and I suspected some of the cable runs might be older RG-59 instead of RG-6. Weak signal from the street seemed like the obvious culprit.

I swapped cables. Checked connections. Tightened every F-connector in the house. Nothing changed.

Then I had the kind of breakthrough that only comes from systematically eliminating variables. I powered off my ScreenBeam MoCA adapter, and full speeds came back instantly. One Gbps, clean signal, no errors.

That was the moment everything clicked.

Understanding the Frequency Conflict

To understand why this happened, you need to know how DOCSIS 3.1 and MoCA share the coax spectrum. They were never designed to coexist without proper isolation.

graph LR subgraph "Coax Frequency Spectrum" A["5-42 MHz\nDOCSIS Upstream"] ~~~ B["54-1002 MHz\nDOCSIS 3.0\nDownstream"] B ~~~ C["1002-1218 MHz\nDOCSIS 3.1\nOFDM Extension"] C ~~~ D["1125-1675 MHz\nMoCA 2.0/2.5"] end style C fill:#e74c3c,stroke:#333,color:#fff style D fill:#e74c3c,stroke:#333,color:#fff

See the problem? The DOCSIS 3.1 OFDM channels extend up to 1218 MHz. MoCA 2.0 operates at 1125-1675 MHz. There’s a direct frequency overlap between 1125 and 1218 MHz where both technologies are trying to use the same spectrum simultaneously.

My old Cogeco modem had MoCA built right into the chipset. It managed both DOCSIS and MoCA internally on a single chip, so they never conflicted - the modem knew which frequencies belonged to which protocol and kept them separated. The new GT-DUO31 is a pure DOCSIS modem with zero MoCA awareness. It has no idea that a ScreenBeam adapter is screaming MoCA signals directly into its OFDM receive path through the shared coax splitter.

The MoCA signals were bleeding through the splitter and corrupting the modem’s OFDM channels, which explains the terrible signal levels and constant synchronization failures. The modem was trying to interpret MoCA traffic as DOCSIS data and choking on it.

What I Was Trying to Protect

This wasn’t a simple modem-and-router situation where I could just ditch MoCA. The MoCA backbone was critical infrastructure for my entire home network:

graph TD ISP["ISP Coax\nFrom Street"] --> S1["Splitter #1"] S1 --> S2["Splitter #2"] S2 --> MODEM["GT-DUO31\nDOCSIS 3.1 Modem"] S1 --> SB1["ScreenBeam #1\nMoCA Adapter"] MODEM --> XT9A["ASUS XT9\nMain Router - WAN"] XT9A -->|LAN| XT9B["ASUS XT9\nRouter Node"] XT9B --> SW1["Office Switch"] SW1 --> PROX["Proxmox Server\nRyzen 9 5950X\n64GB RAM"] SW1 --> PI["Raspberry Pi\nPi-hole + Homebridge"] SW1 --> SB1 SB1 -->|"MoCA over\nBasement Coax"| SB2["ScreenBeam #2\nLiving Room"] SB2 --> SW2["Living Room Switch"] SW2 --> SONOS["Sonos Arc"] SW2 --> NSW["Nintendo Switch 2"] SW2 --> ATV["Apple TV 4K"] SW2 --> XT9C["ASUS XT9\nSatellite Node\nWired Backhaul"] style MODEM fill:#3498db,stroke:#333,color:#fff style SB1 fill:#e67e22,stroke:#333,color:#fff style SB2 fill:#e67e22,stroke:#333,color:#fff style PROX fill:#2ecc71,stroke:#333,color:#fff

The MoCA link through the basement coax carries wired backhaul from the office to the living room, providing hardwired connections to the Sonos Arc, Nintendo Switch 2, Apple TV 4K, and the satellite XT9 mesh node. Running new Ethernet through the basement was not something I wanted to do. The MoCA backbone had to stay.

The Fix: A $15 Frequency Filter

The solution was a PPC Belden MoCA POE (Point of Entry) filter placed on the coax line between splitter #2 and the GT-DUO31 modem. This is a small barrel-style filter that does one simple thing: it passes DOCSIS frequencies (5-1002 MHz) cleanly while blocking MoCA frequencies (1125-1675 MHz) from reaching the modem.

graph TD ISP["ISP Coax"] --> S1["Splitter #1"] S1 --> S2["Splitter #2"] S2 --> FILTER["PPC Belden\nMoCA POE Filter\n$15"] FILTER -->|"Passes 5-1002 MHz\nBlocks 1125-1675 MHz"| MODEM["GT-DUO31 Modem\nClean DOCSIS Signal"] S1 -->|"MoCA flows freely\nthrough this leg"| SB1["ScreenBeam #1"] style FILTER fill:#27ae60,stroke:#333,color:#fff style MODEM fill:#3498db,stroke:#333,color:#fff style SB1 fill:#e67e22,stroke:#333,color:#fff

The key insight is where you place the filter. It goes between the splitter and the modem only. The MoCA signals still flow freely through the other splitter leg to the ScreenBeam adapter. The modem gets clean DOCSIS, the ScreenBeam gets its MoCA, and everybody’s happy.

The Mental Model: Signal Isolation on Shared Media

This problem boils down to a fundamental networking concept: when multiple protocols share the same physical medium, you need proper isolation between them. This is the same principle behind VLANs on Ethernet, channel separation in Wi-Fi, and frequency coordination in cellular networks.

graph TD A["Shared Physical Medium\n(Coax Cable)"] --> B{"Are protocols\nfrequency-aware\nof each other?"} B -->|"Yes\n(Built-in MoCA modem)"| C["Internal Coordination\nNo conflict"] B -->|"No\n(Separate devices)"| D{"Do frequencies\noverlap?"} D -->|"No overlap"| E["Safe to coexist\nNo filter needed"] D -->|"Overlap exists"| F["Install frequency filter\nto isolate signals"] F --> G["Filter placement:\nProtect the device that\nCAN'T handle the interference"] style F fill:#27ae60,stroke:#333,color:#fff style G fill:#27ae60,stroke:#333,color:#fff

The mental model here is simple: whenever you swap a device on shared media, ask yourself what frequencies or protocols the old device managed internally that the new device might not. The old Cogeco modem was a Swiss Army knife that handled both DOCSIS and MoCA on one board. The GT-DUO31 is a scalpel - excellent at DOCSIS, completely blind to MoCA. That blindness is what created the conflict.

The Result

Full 1 Gbps speeds restored. ScreenBeam MoCA running perfectly. Wired backhaul to the living room intact. All hardwired devices - the Proxmox server, Raspberry Pi, Sonos, Apple TV, Nintendo Switch 2 - completely unaffected. Modem signal levels back in spec. Total cost: $15 and a couple hours of troubleshooting.

Key Learnings

DOCSIS 3.1 and MoCA 2.0 have a frequency overlap at 1125-1218 MHz - This is the root cause of interference when both technologies share the same coax without isolation
Modems with built-in MoCA handle the conflict internally - When you replace one with a DOCSIS-only modem, that internal coordination disappears and you inherit the problem
MoCA POE filters are cheap and effective - A $15 PPC Belden filter placed between the splitter and the modem blocks MoCA frequencies while passing DOCSIS cleanly
Filter placement matters - The filter goes on the modem’s coax input only, leaving the MoCA path through the other splitter leg unobstructed
Always check modem signal levels at 192.168.100.1 - Downstream power outside -7 to +7 dBmV and upstream power above 47 dBmV are red flags that something is actively degrading your signal
Systematically eliminate variables - Cable swaps told me nothing, but powering off the MoCA adapter instantly revealed the root cause
When swapping devices on shared media, audit what the old device managed that the new one doesn’t - This is the single most transferable lesson from this experience