FTTH Low-Level Design: Where to Place Splice Points Without Wrecking Your Build

Splice point placement is one of those LLD decisions that looks like a detail and behaves like a foundation. Get it right and your construction crews move efficiently, your optical budgets hold, and your future maintenance windows don't turn into all-day events. Get it wrong and you're watching a crew try to re-enter a buried enclosure that was placed in the wrong conduit section, in the wrong access vault, in the wrong part of the road — six weeks after the concrete poured. Bad FTTH low-level design splice point placement creates problems that no amount of good field work can fully fix.

This isn't a theoretical exercise. The engineering logic behind splice point placement in LLD directly determines construction sequence, optical link performance, and long-term network maintainability. We're going to walk through how to think about this — not the textbook version, but the version that holds up when a backhoe cuts the wrong segment in a rural county with no cell service and a two-hour drive to the nearest supply house.

Why FTTH LLD Splice Point Decisions Live or Die at Low-Level Design

Splice point placement in FTTH low-level design (LLD) determines access feasibility, optical loss budgets, construction labor cost, and future maintenance accessibility — all of which are impossible to validate at the HLD stage. HLD splice representations are schematic; LLD must specify the exact access type, closure model, terrain constraints, and fiber count at every splice location before the construction package is issued.

High-level design establishes your fiber topology — how many FDHs, where the splitter stages go, what the feeder routes look like in broad strokes. But HLD splice point representations are almost always schematic. They show that a splice will occur at a given segment junction; they don't specify whether that splice happens at a handheld closure strapped to a strand, a buried vault accessible from a 20-inch ring, or a pedestal at grade in a utility easement that the county owns and won't let you disturb.

That specificity is an LLD job. And the decisions made at LLD have direct cost consequences at construction. A mid-span aerial splice that could have been placed at an existing pole tap gets buried because the LLD engineer modeled segment lengths based on CAD measurements alone, without checking pole spacing in the field. Now a crew needs a bucket truck and a 45-minute setup to access what should have been a 15-minute pedestal pull. That happens dozens of times on a large build and it adds weeks to construction.

The governing principle we apply: every splice point must have a defined access method before the LLD is released for construction. Not assumed. Defined.

The Basic Math: Segment Length vs. Splice Loss Budget

Before you can place splice points intelligently, you need to know your optical budget for each segment. This is arithmetic, not art.

On a typical XGS-PON deployment, you're working with a class B+ link budget of 28 dB between OLT and ONT. Break that down for a real feeder segment: your feeder fiber contributes roughly 0.35 dB/km of attenuation (for G.652.D compliant single-mode), each fusion splice adds 0.07–0.12 dB depending on equipment and technician quality, and each connector pair contributes 0.3–0.5 dB. Add a 2.5 dB safety margin for aging, connector contamination, and temperature cycling, and you're working with a net budget somewhere around 25.5 dB before passive splitter insertion loss comes into the calculation.

A 1:32 splitter introduces 17.9 dB of insertion loss. Alone. That leaves you 7.6 dB to cover fiber attenuation, splices, and connectors on both the feeder and distribution sides. On a feeder run of 6.2 km with four fusion splices and two connector pairs, you've already used approximately 5.3 dB of that remaining budget before a single distribution splice is accounted for. Tight — but workable if the LLD controls splice count tightly and keeps connector pairs to the minimum necessary.

Where this breaks down is when LLD engineers add splice points without tracking cumulative loss. Every unnecessary mid-span splice — placed because the engineer was trying to balance segment lengths for construction convenience — chips away at a budget that was already thin. On a 10-mile rural feeder in a project we worked in eastern Tennessee, we counted 11 mid-span splices that were the result of segment-length balancing decisions made without a running optical budget calculation. The feeder came in at 26.4 dB before the splitter. Two segments were marginal at best.

Terrain-Driven FTTH LLD Splice Placement Logic: What the GIS Map Doesn't Show

Terrain-driven splice point placement accounts for vertical elevation changes, access road proximity, subsurface conditions, and overhead obstructions that GIS data doesn't capture. Engineers placing splice points from GIS data alone work with accurate horizontal measurements but no information about access feasibility in the field. Terrain factors — hillsides, creek corridors, private easements — frequently require moving planned splice locations by 500–2,000 feet from the desktop-optimized position.

This is where a lot of LLD packages fall apart. Engineers placing splice points from GIS data are working with accurate horizontal measurements but essentially no information about vertical terrain, access obstructions, or subsurface conditions. Those three factors control everything about how practical a splice location actually is.

Aerial Deployments: Pole-Anchored Logic

On aerial plant, splice points want to live at poles. Full stop. Not at mid-span locations, not strapped to a messenger wire somewhere between structures — at poles, where a bucket truck has a stable work platform, where a closure can be properly lashed to the strand with support, and where future access doesn't require approximating the location from a GPS coordinate and hoping the enclosure didn't migrate 18 inches over three winters of ice loading.

The practical rule: place aerial splice closures at poles with a minimum 8-foot clearance above the ground wire level per NESC Section 235, and anchor the closure to the strand with a minimum of two lashing supports within 18 inches of each closure end. We've seen closures placed with single-support lashing on mid-span locations that developed a persistent 0.4 dB increase in splice loss over two years — entirely from mechanical stress on the fiber pigtails as the closure swayed. That's 0.4 dB you didn't budget for and can't easily fix.

On aerial builds, try to align splice point placement with naturally occurring pole clusters where possible — areas where pole spacing tightens up at road crossings, near substations, or at joint-use attachment clusters. These locations give you multiple access points within short distances, which matters when you need to reroute around a failed segment.

Underground Deployments: Access Vault Location Is Everything

Underground splice points in direct-buried or conduit systems need to be at accessible vault locations — and "accessible" means vehicle access within 40 feet, a vault large enough to accommodate a splice trailer or at minimum a two-technician enclosure entry, and a surface cover that isn't going to be buried under two inches of asphalt when the county repaves. That last one is not hypothetical. On a project outside Columbus, Ohio, a developer paved over 14 of our originally accessible splice vaults during a subdivision buildout phase that started after our fiber was already in the ground. We got them back, but it took 11 months of negotiation and a surveyor-documented as-built to locate them all.

Place underground splice vaults at lot corners, road intersections, or conduit transition points where multiple runs converge. These locations serve double duty — they're natural access points that other utilities also service, which means they're less likely to be built over, and they're GPS-locatable reference points that hold up in the field without sub-centimeter survey accuracy.

Fiber Count at Each Splice Point: Capacity Planning for Segments You Haven't Built Yet

One of the most persistent LLD mistakes is designing splice points that have exactly the capacity needed for today's fiber count — and nothing else. No spare tubes in the tray. No additional tray slots in the closure. No room for a future express cable. Just the minimum.

Our standard on any splice enclosure is a minimum 25% spare capacity at time of initial splice. On a 96-fiber feeder where 72 fibers are active, that means at minimum 24 fibers are spliced through to spare pigtails, heat-shrink protected, and stored in a spare tray. The incremental cost of this at construction — probably $340 in additional splice labor and materials per closure — is trivially small compared to the cost of opening a buried closure, adding tray hardware, and re-entering the enclosure to add capacity when the network expands.

Future capacity also means physical space. Specify closures with at least two additional tray slot positions beyond what's needed for the initial build. For an inline aerial closure serving a single mid-span joint on a 144-fiber feeder, a 288-fiber dome closure with half the trays populated is the right answer — not a snug 144-fiber closure that's full from day one. The price difference on a single closure is $87. The cost of swapping it out in year four is a minimum $1,400 in truck rolls, labor, and traffic control.

Splice Point Spacing: Real Numbers From Real Builds

There's no universal rule for splice point spacing — it depends on plant type, terrain, fiber count, and access environment. But we can give you the ranges we actually use and why.

On aerial feeder plant in flat or rolling terrain, we typically target splice points every 2,000–4,500 feet along the feeder route. That's based on standard reel lengths (typically 3,000-foot reels for 144-fiber and above), pole spacing that allows for convenient access, and the goal of keeping individual cable segment handling weights reasonable for construction crews. In urban areas with dense pole spacing, you can push closer to 4,500 feet without losing practical access. In rural areas where pole spacing opens up to 300–400 feet between structures, 2,500 feet is often the limit before you're creating mid-span splice situations.

Underground feeder runs in conduit give you more flexibility because you're not constrained by pole spacing — but you're constrained by pull distance. For a 2-inch HDPE conduit with standard 144-fiber armored cable, maximum pull lengths without intermediate pull boxes run approximately 800–1,200 feet in straight conduit, significantly less on curved routes. Splice vaults at pull points are a construction necessity regardless of optical budget, so the question becomes whether those pull-point vaults are also optimally located for optical access or just for construction convenience.

On distribution plant — the cable segments between the FDH and the NAPs — splice point spacing is usually determined by NAP serving radius rather than cable segment logic. We try to keep distribution cable segments at 1,500 feet or under between FDH and the furthest NAP, which limits splice count to one or two per distribution branch and keeps distribution splice losses well within budget even on tight optical plans.

Access Requirements Your LLD Documentation Must Specify

This is the part that gets skipped. An LLD splice plan that shows splice point locations without specifying access method is an incomplete document — and construction estimators know it. They'll build in contingency for access unknowns, and that contingency is always higher than the actual cost of doing the access work properly if it had been designed upfront.

Every splice point in a complete LLD should have the following attributes documented:

• Access type: Aerial (pole-mounted bucket), underground vault (ring-and-cover at surface), handheld closure (no special access required), pedestal (at-grade locked box)
• Vehicle access clearance: Yes/no, with distance to nearest accessible road if no direct access
• Closure type and manufacturer: Not just "dome closure" — specify whether it's a Corning CFUL, a CommScope FOSC-H, or equivalent, with tray count and fiber capacity
• Splice count at this location: Active splices, spare splices, express (pass-through) fibers
• Utility conflict flag: Any known crossing, shared vault, or proximity to high-voltage infrastructure within 10 feet

That attribute set takes maybe 20 minutes per splice point to document during LLD. It saves hours of field estimation and prevents the category of change order that reads "relocated splice vault due to access conflict not identified in design package." We see those change orders constantly — and they're almost entirely preventable.

How Bad Splice Placement Cascades Through Construction

There's a failure chain that's worth tracing explicitly, because it shows up on builds in every region and the root cause always traces back to the same LLD decisions.

It starts with a splice point placed at a convenient segment-length boundary without field verification. Construction crews discover an access problem — the vault location is in a drainage easement, or the pole is double-deadend and can't accommodate a closure without a make-ready application, or the subsurface is solid rock and the vault excavation is $2,200 instead of $400. The field foreman makes a judgment call and relocates the splice point 200 feet in one direction. That changes the cable segment length on two sides of the splice. Which changes the optical budget margin on one of those segments. Which may or may not get caught in the field, depending on whether OTDR testing is done before or after the distribution drops are connected.

Three months later, a technician runs an OTDR on a problematic subscriber drop and finds a 0.8 dB elevated loss event at a splice nobody documented as relocated. They can't correlate it to any as-built record because the field change never made it back to the LLD documentation. Now you have a real cable plant with a shadow LLD — and every future maintenance event on that segment starts with a 45-minute document reconciliation exercise.

Getting the LLD splice placement right prevents all of it. Our FTTH design services cover LLD from route engineering through splice point documentation to as-built reconciliation, specifically to close this loop. And if you want to understand how these LLD decisions flow from upstream HLD choices, our breakdown of common FTTH HLD design mistakes covers the five most expensive HLD errors we catch on incoming design reviews.

Terrain-Specific FTTH LLD Splice Placement Guidance: Where the Standard Rules Break

Flat suburban terrain is forgiving. Most of the guidance above holds without significant modification. But three terrain types consistently break the standard rules and require explicit LLD treatment.

Rocky or Mountainous Terrain

Direct-buried conduit in rock is expensive — we've seen boring costs hit $127/foot in hard granite outside of Asheville, NC, compared to $18–22/foot in soft Carolina soil. That cost forces longer conduit segments between splice vaults, which pushes optical budgets harder and creates fewer access points. In rocky terrain, plan for maximum conduit segment lengths of 600 feet between access points, budget for sealed rock-entry vaults rather than standard polymer boxes, and seriously consider aerial routing alternatives wherever an existing pole line is within 500 feet of the planned underground route.

High Water Table and Coastal Areas

Buried splice vaults in coastal Georgia, the Florida panhandle, and Louisiana delta areas face a persistent challenge: the vault floods. Not occasionally — seasonally. Standard inline buried closures that work fine in Tennessee will fail within 18 months in a coastal environment where the vault is below the water table during wet season. Either go to aerial plant where possible, specify pressure-rated inline closures rated for continuous submersion (there are only a handful that actually hold up — the Corning FOSC-450 is one we've used successfully in these environments), or build your access vaults with drainage provisions that connect to the municipal storm system.

Bridge and Railroad Crossings

Never place a splice point within 150 feet of a bridge or railroad crossing. This is a practical rule, not a code requirement. Splice points near crossings end up being the points where permitting authority access restrictions make maintenance nearly impossible — railroads won't let you open a fusion splice closure within their ROW during any flagged possession window, which typically costs $1,800–3,200 in flagman fees for a 4-hour window. Put the splice on the near side of the crossing with plenty of margin, and run a continuous express cable through the crossing itself.

For more on how GIS planning tools can help identify these terrain constraints before LLD begins, our piece on GIS-driven fiber network planning and cost reduction covers the specific workflow we use to flag splice placement risks during early design.

The FTTH LLD Review Gate That Catches Splice Point Placement Problems

We run a mandatory splice placement review before any LLD is released for construction. It's a one-day exercise for a standard-size project — maybe two days for a 500-mile rural build — and it catches an average of 17 splice point issues per 100 miles of plant on projects where we weren't involved in initial LLD.

The review checks:

• Running optical budget from OLT through each FDH to the worst-case ONT on each segment
• Field-verified access for every splice point
• Closure sizing versus fiber count at each location, with spare capacity calculation
• Segment length consistency with cable reel lengths (eliminate reel-change joints where possible)
• Splice point proximity to crossing structures, drainage features, and utility conflicts
• Make-ready implications for any aerial splice points on poles with existing attachments

That last item connects directly to the broader make-ready and pole loading picture. A splice point that looks like a simple aerial closure placement can trigger a full pole loading analysis and make-ready application if the pole is already loaded near NESC limits. We've seen LLDs that added 37 splice closures to poles that needed make-ready before any new attachment could go on — a fact that was entirely invisible until the field survey team walked the route. At that point, the construction schedule was already committed.

If you're planning a new FTTH build or reviewing an existing LLD for construction readiness, our engineering team can run a splice placement audit and optical budget reconciliation on your design package. Reach out at info@draftech.com — this is work we do regularly, and it consistently prevents the worst construction surprises before they happen.