2Gbps now officially available in HFC areas
07-11-2026, 10:19
|
#16
|
|
cf.addict
Join Date: Mar 2007
Posts: 453
|
Re: 2Gbps now officially available in HFC areas
Shame really, as I noted before the "small" fibre add on cabs were installed around our area (ex-NTL co-ax) about 3 years ago. Looks like they will never be used. That leaves OR as the only FTTP supplier at this point in time.
|
|
|
11-11-2026, 15:53
|
#17
|
|
Mr Gary Pearce
Join Date: Nov 2026
Location: United Kingdom
Services: TV
INternet
CCTV
Home security
Alarms
Hikvision
Dahua
Ajax
Posts: 14
|
Re: 2Gbps now officially available in HFC areas
Having worked alongside cable and telecoms distribution plant for decades, the NetCo restructuring and the Mustang timeline make complete sense once you look at the physical RF reality of the legacy coaxial network:
**1. The 2Gbps HFC Reality: Heavy Downstream Bandwidth, Asymmetric Upstream Trap**
Virgin can push 2Gbps downstream over DOCSIS 3.1 today by bonding two 192MHz OFDM blocks alongside 24 to 32 legacy SC-QAM downstream carriers. The Hub 5 Broadcom BCM3390 chipset handles that downstream throughput with ease.
However, the crippling bottleneck on legacy HFC remains the **upstream return path**:
- The vast majority of UK HFC nodes are still running the legacy **Sub-Split (5 to 65 MHz)**.
- That tiny 60MHz slice of spectrum only accommodates four 6.4MHz ATDMA channels (plus one 50MHz OFDMA upstream block where enabled), capping real-world aggregate upstream bandwidth across an entire node segment.
- Delivering true symmetric multi-gigabit speeds on HFC would require retrofitting **Mid-Split (85 MHz)** or **High-Split (204 MHz)** diplex filters across every street cabinet, line extender, and distribution amplifier in the cascade—which costs nearly as much in physical labour as blowing micro-fibre down the existing ducts.
**2. The NetCo / Nexfibre Shift**
Liberty Global's comments about not putting further FTTP CAPEX directly on the VM02 balance sheet are driven by the corporate spin-off of the network into **NetCo**:
- By ring-fencing the physical network into a wholesale NetCo (mirroring the Openreach / BT Retail split), they can merge or co-build directly with **Nexfibre** (backed by InfraVia's £4.5B external financing) without burdening VM02's retail balance sheet with debt.
- Project Mustang isn't dead—it's being shifted to NetCo wholesale financing where it can compete for third-party wholesale ISP traffic (Sky, Vodafone, TalkTalk) against Openreach.
**3. OPEX Always Kills HFC in the End**
Regardless of how long they sweat the DOCSIS 3.1 assets, the operational cost of active coaxial plant is staggering compared to passive optical:
- HFC requires powered street nodes, line amplifiers, and power supply units continuously drawing unmetered mains electricity 24/7/365, all vulnerable to summer/winter temperature RF tilt and seasonal balancing.
- XGS-PON is completely passive between the Headend OLT and the home ONT. Zero street electricity, zero RF amplifiers to replace after lightning strikes or power surges, and an immediate 10Gbps symmetrical path. HFC will be sweated where cash flow demands it, but every operator knows full fibre is the only viable endgame.
|
|
|
11-11-2026, 18:00
|
#18
|
|
Roger_Gooner
Join Date: Nov 2008
Location: London
Services: Virgin Media M350, Hub 5, Mega TV with TNT Sports and Sky Sports & Cinema HD, TV 360 box + landline
Posts: 221
|
Re: 2Gbps now officially available in HFC areas
Quote:
Originally Posted by homesecurity
Having worked alongside cable and telecoms distribution plant for decades, the NetCo restructuring and the Mustang timeline make complete sense once you look at the physical RF reality of the legacy coaxial network:
**1. The 2Gbps HFC Reality: Heavy Downstream Bandwidth, Asymmetric Upstream Trap**
Virgin can push 2Gbps downstream over DOCSIS 3.1 today by bonding two 192MHz OFDM blocks alongside 24 to 32 legacy SC-QAM downstream carriers. The Hub 5 Broadcom BCM3390 chipset handles that downstream throughput with ease.
However, the crippling bottleneck on legacy HFC remains the **upstream return path**:
- The vast majority of UK HFC nodes are still running the legacy **Sub-Split (5 to 65 MHz)**.
- That tiny 60MHz slice of spectrum only accommodates four 6.4MHz ATDMA channels (plus one 50MHz OFDMA upstream block where enabled), capping real-world aggregate upstream bandwidth across an entire node segment.
- Delivering true symmetric multi-gigabit speeds on HFC would require retrofitting **Mid-Split (85 MHz)** or **High-Split (204 MHz)** diplex filters across every street cabinet, line extender, and distribution amplifier in the cascade—which costs nearly as much in physical labour as blowing micro-fibre down the existing ducts.
**2. The NetCo / Nexfibre Shift**
Liberty Global's comments about not putting further FTTP CAPEX directly on the VM02 balance sheet are driven by the corporate spin-off of the network into **NetCo**:
- By ring-fencing the physical network into a wholesale NetCo (mirroring the Openreach / BT Retail split), they can merge or co-build directly with **Nexfibre** (backed by InfraVia's £4.5B external financing) without burdening VM02's retail balance sheet with debt.
- Project Mustang isn't dead—it's being shifted to NetCo wholesale financing where it can compete for third-party wholesale ISP traffic (Sky, Vodafone, TalkTalk) against Openreach.
**3. OPEX Always Kills HFC in the End**
Regardless of how long they sweat the DOCSIS 3.1 assets, the operational cost of active coaxial plant is staggering compared to passive optical:
- HFC requires powered street nodes, line amplifiers, and power supply units continuously drawing unmetered mains electricity 24/7/365, all vulnerable to summer/winter temperature RF tilt and seasonal balancing.
- XGS-PON is completely passive between the Headend OLT and the home ONT. Zero street electricity, zero RF amplifiers to replace after lightning strikes or power surges, and an immediate 10Gbps symmetrical path. HFC will be sweated where cash flow demands it, but every operator knows full fibre is the only viable endgame.
|
Much of what you posted technically about HFC RF spectrum is accurate, but the narrative on NetCo and cable CAPEX relies on an outdated structural premise.
1. The NetCo Reality
The standalone corporate NetCo proposal is dead. The actual timeline is:
- February 2024: NetCo proposed.
- July 2025: NetCo cancelled after Telefónica withdrew support.
- September 2025: Dedicated in-house Fixed Wholesale unit launched.
That internal Fixed Wholesale operation is very much alive today, acting as VMO2's wholesale vehicle for selling access across the Virgin Media and nexfibre footprints through a common commercial interface. It serves wholesale customers including ISPs and other network operators.
2. Project Mustang & CAPEX
Liberty Global's decision not to put further FTTP CAPEX directly on VMO2's balance sheet has nothing to do with NetCo or wholesale mechanics. It is simple financial arithmetic: VMO2 has £22.4bn of debt and lease obligations and cannot justify the massive upfront capital burden of Project Mustang while waiting years to realise the financial returns from migrating customers and services from HFC and RFoG to XGS-PON.
3. High-Split vs. R-PHY / DAA Economics
The cost comparison for HFC is actually potentially much worse than your post suggests. Delivering true symmetrical multi-gigabit DOCSIS isn't just an exercise in swapping street diplex filters. A serious DOCSIS 4.0 upgrade would involve extensive modification of the RF distribution plant and, in a modern DAA-based deployment, substantial investment in Remote PHY nodes, digital fibre transport, CCAP infrastructure and compatible active street equipment.
As recently as July 2020, Virgin Media had started deploying R-PHY in Coventry, with a much wider rollout planned. At that time, the strategy was to double down on the DOCSIS network. DOCSIS 4.0 would have been a logical subsequent development. However, VM subsequently pivoted towards full-fibre XGS-PON and Project Mustang, correctly in my view.
The point is that upgrading HFC to deliver symmetrical multi-gigabit speeds must involve substantial additional capital expenditure, while retaining the ongoing power consumption and active maintenance costs of the coaxial network. Depending on the condition of the existing plant and the extent of the upgrade, this could cost more than installing fibre through existing ducts.
With VMO2's current financial constraints and little commercial justification for symmetrical multi-gigabit HFC, I see no prospect of a comprehensive High-Split/DOCSIS 4.0 programme.
In any case, why would VM want symmetrical speeds on its legacy network? The recently-announced 2Gbps download and 100Mbps upload service is cheap to deliver as it requires no civil engineering and will be more than sufficient for the vast majority of customers, and it's reported that 2Gbps would be available on almost 70% of the HFC network next year.
4. The Mustang Pause
I agree Project Mustang isn't dead, but not for the reason you put forward. The holy grail for VMO2 is an all-IP, passive access platform, bringing substantial long-term benefits through lower electricity consumption, reduced maintenance and the eventual elimination of all of the active HFC street infrastructure. (There are even bigger savings to be made also from headend consolidation, but that's a somewhat complicated subject requiring its own thread.)
Mustang's wider expansion has been halted because VMO2's balance sheet cannot support the required investment at present. It can continue using the fibre already built and migrating customers where commercially sensible. A broader programme could resume when its finances permit, even if that means waiting several years or, potentially, a decade.
__________________
I'm always right... except when I'm wrong
|
|
|
12-11-2026, 01:51
|
#19
|
|
Mr Gary Pearce
Join Date: Nov 2026
Location: United Kingdom
Services: TV
INternet
CCTV
Home security
Alarms
Hikvision
Dahua
Ajax
Posts: 14
|
Re: 2Gbps now officially available in HFC areas
Spot on regarding the CAPEX reality. Coming from the physical cabling and field installation side, squeezing 2Gbps out of legacy HFC rather than completing the full Project Mustang XGS-PON overbuild makes total balance-sheet sense for Liberty Global, but on the ground it introduces some very interesting physical plant challenges:
1. The Physical Coax Bottleneck (Drop Cables & Isolators)
To push 2Gbps downstream on DOCSIS 3.1, VM has to push OFDM carrier spectrum right up to 1.0GHz / 1.2GHz. While modern WF100/RG6 can handle that fine over short runs, out in the street cabinets and domestic brown omni-boxes you regularly encounter 20-to-30-year-old Telewest/NTL drop leads, corroded galvanized f-connectors, and legacy Technetix ground isolators. At 800MHz–1GHz, the high-frequency attenuation curve steepens drastically. We see properties where 1Gbps was rock solid, but negotiating full 2Gbps profiles triggers forward error correction (FEC) errors because the physical drop cable simply doesn't have the SNR margin at the top of the band without a repull.
2. The Upstream Asymmetry Gap
Downlink marketing numbers look great at 2Gbps, but on HFC the elephant in the room remains the upstream. Without high-split (204MHz) or full DOCSIS 4.0 upgrades to street taps and line amplifiers, HFC is still choked on sub-split/mid-split OFDMA. Getting 2000Mbps down with ~100-150Mbps up is a huge disparity compared to native XGS-PON altnets offering symmetrical 2Gbps/2Gbps. For domestic multi-user homes with cloud backups, remote camera sync, and NAS transfers, that upstream bottleneck becomes obvious very quickly.
3. The Internal 1Gbps 'Bottleneck Wall'
As Carth rightly noted earlier, delivering 2Gbps to the Hub 5/5x port is only half the battle. Because the Hub only has a single 2.5Gbps port (with the remaining three capped at standard 1Gbps), 95% of domestic setups hit a brick wall. Unless the homeowner runs Cat6/Cat6a into a multi-gig 2.5GbE managed switch and has 2.5G NICs or clean 160MHz Wi-Fi 6E/7 channels, they literally cannot pull more than 940Mbps on any single device.
If Mustang really is paused on the VM balance sheet, DOCSIS 3.1 will have to carry the heavy water for several years to come. It’s remarkable how much bandwidth they've managed to wring out of copper coax originally pulled in the 1990s!
|
|
|
12-11-2026, 05:34
|
#20
|
|
Mr Gary Pearce
Join Date: Nov 2026
Location: United Kingdom
Services: TV
INternet
CCTV
Home security
Alarms
Hikvision
Dahua
Ajax
Posts: 14
|
Re: 2Gbps now officially available in HFC areas
From the physical cabling and RF infrastructure side, there are two distinct technical realities behind this rollout that explain both the 100Mbps upload bottleneck and the headaches subscribers will hit inside the home:
1. The Physical Return Path Spectrum Barrier (Why Upload is Capped at 100Mbps):
On XGS-PON (pure full-fibre), symmetrical 2Gbps is simple because optical transmission uses separate 1577nm downstream and 1270nm upstream wavelengths with zero RF diplex constraints.
On legacy Hybrid Fibre-Coaxial (HFC), the return path has historically been severely restricted by the physical diplex filters built into street cabinet amplifiers and distribution taps. Most UK cable plants were built on sub-split (5–65MHz) or mid-split (5–85MHz) frequency allocations.
To deliver 2Gbps downstream, VM bonds 31–32 legacy SC-QAM channels alongside a wide DOCSIS 3.1 OFDM block in the upper spectrum (stretching towards 1000–1218MHz). However, the upstream is still crowded into 4 or 5 legacy ATDMA channels plus a narrow OFDMA channel. Until Virgin physically replaces every legacy line amplifier, directional coupler, and street tap across an entire distribution node to High-Split (5–204MHz) or full DOCSIS 4.0 FDX/ESD, squeezing more than 100Mbps upload out of HFC is an RF impossibility without risking severe ingress noise swamping the node.
2. The 940Mbps Local LAN Bottleneck (What Users Actually Experience):
Paul and Carth hit the nail on the head regarding internal home networking. We see this constantly on multi-gigabit upgrades: a customer pays for 2Gbps, runs an online speed test from a PC plugged into a standard wall port or unmanaged gigabit switch, and logs a fault because it tests at "only 930–940Mbps".
Standard Gigabit Ethernet (1000BASE-T) has a hard physical throughput ceiling of roughly 940–948Mbps once TCP/IP framing, inter-packet gaps, and preamble overhead are accounted for.
To actually see 2000Mbps on a single client device:
- The device must connect directly to Port 4 (the dedicated 2.5Gbps port) on the Hub 5.
- The client PC needs a 2.5GbE PCIe network card or USB-C multi-gig adapter.
- Any intermediate switch between the Hub and the PC must be a 2.5Gbps multi-gig switch.
- The interconnecting patch leads and building runs must be solid-copper Cat6 (not cheap flat CCA / copper-clad aluminium cables that suffer from packet loss under heavy multi-gigabit throughput).
3. Latency & Jitter vs Raw Bandwidth:
For gaming, video calling, and interactive tasks, raw bandwidth above 100Mbps is largely irrelevant—packet round-trip time and bufferbloat are what matter. Even with 2Gbps downstream, HFC cable still relies on time-division multiple access on the return path, which inherently introduces more jitter under load than dedicated point-to-point fibre. Putting the Hub into Modem Mode and running a third-party router with SQM (Smart Queue Management such as CAKE or fq_codel) will do infinitely more for real-world responsiveness and game latency than jumping from 1Gbps to 2Gbps on an asymmetrical coax drop.
|
|
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT. The time now is 03:21.
|