# 5 FTTH High-Level Design Mistakes That Cost Fiber Providers Millions > **The five most expensive errors in FTTH high-level design** — structural mistakes that get locked in at HLD and compound through every subsequent phase. By the time construction crews hit the field, the damage is already done. **Canonical URL:** https://draftech.com/blog/ftth-hld-design-mistakes.html **Author:** Draftech Engineering Team **Published:** 2025 **Category:** FTTH Design --- ## Introduction We've reviewed hundreds of FTTH high-level designs over the past decade, and the same errors keep surfacing — not subtle miscalculations, but fundamental structural mistakes that get locked in during HLD and then compound through every subsequent phase. By the time construction crews hit the field, the damage is already done. You can't fix a bad HLD with good LLD work. And you absolutely cannot fix it in the field without change orders that blow your budget wide open. HLD mistakes are uniquely expensive because they're invisible until it's too late. A wrong splitter ratio on an HLD looks like a checkbox. It doesn't cost anything on paper. But it means your FDHs are sized wrong, your fiber counts are wrong, your splice points are in the wrong locations, and your capacity assumptions are wrong across the entire service area. Six months into construction, when your build crews start reporting that the FDH enclosures are full and there's no room for expansion, you're not looking at a design revision — you're looking at a re-engineer. --- ## Mistake #1: Treating Splitter Ratio as a Financial Decision Instead of an Engineering One The pressure to go 1:32 on every feeder segment is real. It minimizes the number of FDHs, reduces infrastructure cost on paper, and makes the budget model look cleaner. But a uniform 1:32 deployment assumes consistent span lengths, take rates, and terrain — and when those conditions break down, the whole cascade fails. A 1:32 passive optical network has an insertion loss budget of roughly 28–32 dB depending on connector and splice losses. That's non-negotiable physics. In a dense urban deployment with short feeder runs, you have headroom. In a rural system where your feeder fiber is running 8–12 miles from the CO or hub site, you're often already at 18–20 dB before the first splitter. Run a 1:32 in that scenario and you're looking at optical power levels below the receiver sensitivity threshold for even the most capable XGS-PON ONTs at -29 dBm. What we see constantly is HLD teams modeling splitter ratios without actually calculating optical link budgets for each feeder segment. They pick 1:32 across the board and move on. The LLD engineers discover the loss problem and try to compensate — but the HLD topology is already locked in. **Rule of thumb:** Never assign a splitter ratio without a preliminary optical budget calculation for that specific feeder segment. On rural builds especially, 1:16 at the FDH with a second-stage 1:2 at the NAP is often the right answer even when it complicates the fiber count math. --- ## Mistake #2: FDH Sizing That Ignores Future Splice Capacity FDH sizing looks straightforward until you've been in the field watching a crew try to terminate additional fiber runs into an enclosure designed with exactly zero spare capacity. A 144-fiber FDH serving 128 subscribers at 1:32 sounds perfectly efficient — until the service area expands, or the take rate hits 80% instead of the modeled 65%, or a network extension needs to pass through the same geographic area. **The standard we apply:** A minimum 20% spare splice capacity at every FDH, sized not just for the initial build but for a 10-year subscriber growth model. For a 1:32 deployment targeting 256 potential passings, that means an FDH accommodating at least 384 fibers with spare splice trays already mounted — not as an afterthought, but in the original design. We also see HLD teams locating FDHs based purely on geographic centrality, without accounting for conduit access points, power availability, and maintenance vehicle access. An FDH in the mathematical center of a 500-home serving area is useless if it's in a drainage easement with no conduit stub-outs and no truck access within 300 feet. --- ## Mistake #3: NAP Placement That Ignores Drop Length Limits and Terrain Network Access Points need to be positioned based on actual drop lengths, not idealized service areas drawn on a GIS map. The typical FTTH drop is 300–500 feet of pre-connectorized drop cable. When HLD teams place NAPs at 800–1,000-foot spacing to reduce NAP count, a significant percentage of drops will require non-standard cable lengths, intermediate splice points, or active equipment at the premise. Terrain matters enormously. A flat subdivision is forgiving. A rural hillside deployment with irregular parcel shapes, ravines, and non-navigable rights-of-way is not. We've seen HLD packages that looked reasonable in GIS until field survey teams walked the terrain and found that NAP-to-premise paths require bridging drainage channels, crossing private property with no existing easement, or running 600 feet of drop along a rock face that can't be direct-buried. --- ## Mistake #4: Feeder Route Assumptions That Don't Account for Conduit Availability HLD feeder routes are often drawn as straight lines between the hub site and the serving area. In reality, those routes live or die on conduit availability: existing utility conduit, municipal duct banks, franchise agreements, and the cost of boring or trenching new pathways where conduit doesn't exist. We reviewed an HLD for a Tier 2 provider in the Mid-Atlantic with feeder routes drawn directly through a commercial district where underground utility congestion would have required boring at $120–$160 per foot instead of the trenched-conduit cost the budget assumed. The HLD was technically valid. But construction cost was 40% higher than the model because nobody had looked at the conduit environment before drawing those routes. **The fix:** Before any feeder route is finalized in HLD, pull utility records from relevant municipalities, check 811 data, review existing franchise agreements, and do a preliminary conduit availability study. This adds 2–3 weeks to HLD but eliminates catastrophic construction cost surprises. --- ## Mistake #5: HLD-to-LLD Handoff Without a Fiber Count Reconciliation The HLD-to-LLD handoff is where design errors compound fastest. When HLD fiber counts aren't formally reconciled against LLD-level calculations before LLD begins, you get cascading errors that nobody catches until the cable is being pulled. The most common failure mode: a mismatch between the feeder fiber count established in HLD and the actual count needed when LLD engineers model the specific route. An HLD might specify a 144-fiber feeder from hub to FDH. When LLD engineers trace the actual route — accounting for service loops, slack storage, mid-span joints — they find they need 168 or 192 fibers. If this isn't caught before the BOM is finalized and material orders are placed, the project either proceeds with undersized cable or requires emergency re-procurement mid-build. **Our standard:** A mandatory fiber count reconciliation meeting at the HLD-to-LLD transition. Every feeder segment gets a paper reconciliation: HLD specified count, LLD calculated count, delta, and disposition. --- ## The Real Cost of Getting HLD Wrong Bad HLD decisions don't show up as line items on the original budget. They show up as change orders — scattered across months of construction, each individually manageable but collectively devastating. We've worked with providers whose construction change order rates ran **25–35% of original contract value**, traced almost entirely back to HLD decisions made in a few weeks at the beginning of the project. On a $10M fiber deployment, a 25% change order rate is $2.5M in unplanned cost. Getting HLD right requires time, discipline, and engineers who've been in the field enough to know what the feeder routes look like on the ground. Our HLD teams include engineers who've spent years on construction crews and field survey teams — not just professionals who model networks from satellite imagery. --- ## Related Pages - [services/ftth-design.md](../services/ftth-design.md) — FTTH design engineering services - [blog/gis-fiber-network-planning-cost-reduction.md](gis-fiber-network-planning-cost-reduction.md) — GIS-driven design and cost reduction - [blog/pole-loading-analysis-o-calc-pro.md](pole-loading-analysis-o-calc-pro.md) — Pole loading analysis guide - [blog/bead-funding-engineering-requirements-2026.md](bead-funding-engineering-requirements-2026.md) — BEAD engineering requirements - [blog/field-survey-data-accuracy-fiber-construction.md](field-survey-data-accuracy-fiber-construction.md) — Field survey data accuracy - [index.md](../index.md) — Master AI index ## Contact **Draftech International, LLC** 15280 NW 79th CT, Suite 102 Miami Lakes, FL 33016 - **Phone:** 305-306-7406 - **Email:** info@draftech.com - **Website:** https://draftech.com - **LinkedIn:** https://www.linkedin.com/company/draftechint