Manufacturing & Quality
4.0 factories, automation, validation
Partner Portal
Access tools, pricing, ESG data
There is a generation of fibre access infrastructure across Europe that was built correctly, works reliably, and is now running into a supply chain that has moved on without it. Point-to-point Ethernet FTTH networks deployed between roughly 2008 and 2018 — across the Nordics, Benelux, Germany, UK, and progressively across Southern and Eastern Europe as fibre rollouts accelerated - were built on platforms from Nokia, Keymile, Huawei and others that used CSFP transceivers to achieve the port density required for subscriber-scale deployment. The networks are still running. The platforms are largely still in service. The OEM transceiver programmes that originally supplied them are not.
We supply CSFP transceivers. We have done so for long enough to recognise the pattern: first the OEM supply gets unreliable — longer lead times, minimum order quantities that do not make sense for a 20-port top-up, part numbers that quietly go end-of-life. Then someone tries a generic alternative from a catalogue distributor. Then the compatibility problem surfaces at installation. Then they call us.
A CSFP - Compact Small Form-Factor Pluggable - delivers two independent bidirectional optical channels in the physical footprint of a single SFP port. Each channel operates as a BiDi link: the central office end transmits at one wavelength and receives at another, and the CPE end does the reverse. Two subscribers, one port, one module. On a line card that would otherwise carry 80 SFP ports, CSFP doubles the subscriber capacity to 160 without changing the chassis, the backplane, or the cabling infrastructure.
CSFP operates on a specific wavelength plan: the central office module transmits at 1490nm on both channels and receives at 1310nm on both channels. The CPE end uses a standard BiDi SFP - cheaper, widely available, no compatibility issues - transmitting at 1310nm and receiving at 1490nm. The asymmetry is intentional: put the more complex dual-channel component where there are fewer of them (the central office), and the simpler single-channel component at the subscriber premises.
| Parameter | CSFP (Central Office) | BiDi SFP (CPE end) |
| Channels per module | 2 independent BiDi channels | 1 BiDi channel |
| TX wavelengths | 1490nm (both channels) | 1310nm |
| RX wavelengths | 1310nm (both channels) | 1490nm |
| Port density effect | 2× subscriber connections per port | 1:1 |
| Availability | Specialist market — thinning | Commodity — widely available |
| OEM supply status | Largely end-of-life or EOL pending | Stable through multiple vendors |
The choice between point-to-point Ethernet FTTH and passive optical networks was not simply technical. In Northern Europe particularly, regulatory frameworks that required open access to fibre infrastructure pushed operators toward P2P architectures, because a dedicated fibre to each subscriber is structurally simpler to offer as a wholesale product to multiple service providers. P2P also offered operational simplicity: each subscriber has their own dedicated link, fault isolation is immediate, and bandwidth guarantees are absolute. For operators whose primary market was business services, MDU buildings, or public sector connectivity, P2P was the natural choice.
The result was a significant deployment wave across Europe during the 2008–2018 period, with different markets at different stages - earlier builds in the Nordics and Benelux, later rollouts accelerating through Central, Southern and Eastern Europe as national broadband programmes came online. These networks are now at various points of maturity, but share the same characteristic: the infrastructure is paid down, the subscriber relationships are established, and the operators have no particular reason to replace them. The capex argument for a full architecture migration to GPON or XGS-PON is weak when the existing P2P plant is delivering what was promised.
Field observation: The service providers and integrators who contact us about CSFP are not planning migrations. They are managing working networks that need routine component replenishment and occasional expansion. The conversation is never 'should we replace the architecture' — it is 'we need 40 modules for this site and the lead time from our usual supplier just went to 16 weeks.'
The platforms that used CSFP are not in active development. Several OEMs has EOS/EOL announced, while these systems remains in service across thousands of sites but is not the focus of most NEMs (Network Equipments Manufacturers) since the focus for FTTH development is focused on PON. When platforms reach this stage, the OEM transceiver programme follows the same trajectory: part numbers are maintained for a period, then moved to end-of-life status, then discontinued. Lead times extend before they disappear entirely. The problem for operators is that this process is rarely announced clearly - part numbers that were available with a two-week lead time one year arrive with a 16-week lead time the next, without any formal notification that supply is tightening.
The natural response when OEM supply becomes unreliable is to look for alternatives in the broader catalogue market. CSFP is a defined MSA standard — which in principle means any MSA-compliant module should be interchangeable. In practice, this is where the compatibility problem surfaces.
CSFP compatibility on the platforms used in P2P FTTH deployments depends on more than physical and optical specification. The platforms use vendor-specific management interfaces and firmware that interrogate the module's EEPROM data on insertion. A module that carries incorrect or generic identifiers may be rejected by the platform entirely, accepted but flagged as unsupported, or accepted without DDM functionality. The third outcome is the most dangerous: the module appears to work, but the operator has lost the diagnostic visibility that would alert them to a marginal link before it fails.
Field observation: We see two distinct failure modes from generic CSFP modules on these platforms. The first is outright rejection - the line card will not bring the port up. The second is silent degradation - the port comes up, traffic flows, but DDM is non-functional or reporting incorrect values. The operator has no visibility of TX power, RX power, or temperature, and the first sign of a problem is a subscriber complaint or a link failure. The second failure mode causes real operational damage because it takes longer to diagnose and the root cause is less obvious.
CSFP coding for a specific platform is a set of EEPROM data values - vendor ID strings, part number identifiers, serial number format, calibration coefficients, and DDM thresholds - that the host platform interrogates when a module is inserted. Getting this right requires actual test access to the platform and a systematic approach to validating the module behaviour under the platform's management interface. This means checking that the port comes up cleanly, that DDM values are reported correctly and calibrated accurately, that the module is not flagged in the platform's alarm system, and that optical performance matches specification under the platform's operating conditions.
| Coded module | Uncoded or incorrectly coded module | |
| Port behaviour | Comes up cleanly on insertion | May be rejected, flagged, or come up without DDM |
| DDM visibility | TX power, RX power, temperature, voltage and bias current reported accurately | DDM absent, incorrect, or reports fixed placeholder values |
| Alarm behaviour | Module operates within alarm thresholds configured for the platform | May trigger false alarms or suppress valid alarms |
| Maintenance visibility | Operator can monitor link health proactively via DDM trending | Operator is flying blind — first indication of degrading link is often a subscriber fault report |
Field observation: The integrators who now come to us directly have usually been through the generic module experience and private-label compatible solution. The conversation starts with 'we need CSFP for this specific platform - and it needs to actually work, not just physically fit.' quickly followed by 'and can you ensure ongoing availability of the exact part we qualify?'
The P2P FTTH CSFP situation is a specific instance of a broader market dynamic that plays out across multiple optical product categories. When a technology reaches installed-base maturity - when the platform is no longer in active development but the network it powers is still in active service - the supply chain reorganises around newer, higher-volume markets. The transition is gradual and often invisible until it is not.
The distinction that matters is between a supplier who stocks a product category and a supplier who has qualified it against the specific platforms in your network. That distinction applies across the optical market wherever a product is niche enough that doing the qualification work is a genuine investment rather than a catalogue listing.