FAT vs FDT: Fiber Cabinets in FTTH Networks

FAT vs FDT is not a choice between two interchangeable boxes — it is a question about which boundary each one sits on. The FDT (fiber distribution terminal, also called a fiber distribution hub) is where the feeder cable from the central office ends and breaks out into distribution cables, usually through first-level splitters. The FAT (fiber access terminal) is the last enclosure before the subscriber, where distribution fibers terminate and drop cables leave. Put simply: the FDT changes how many cables you are dealing with, the FAT changes which house you are connecting. Everything else — port counts, enclosure ratings, splitter placement — follows from that split of duties. This guide is written for the people who have to specify and install both, and it sits inside the wider FTTH and PON deployment guides for last-mile access.

The boundary is the whole difference

In a PON, every enclosure sits on one of three fiber segments: the feeder from the OLT, the distribution cable through the streets, and the drop to the premises. There are exactly two boundaries between those segments, and the two cabinets in this comparison own one each.

The row that matters most is the last one. A technician standing at an FDT is working on a network; a technician standing at a FAT is working on a customer. That difference sets the enclosure, the access requirements, the IP rating and the way you record the plant — far more than the splitter inside either box.

Where the boundary sits inside a two-level split

The most common misunderstanding in this comparison is that the FDT/FAT distinction is about where the splitter lives. It is not. Splitter location is a design choice; the feeder/distribution and distribution/drop boundaries are physical facts. The three classic recipes — home run, centralized split and distributed split — are cataloged in the FOA's FTTH architecture reference, and only the last two involve a FAT at all.

A total split ratio of 1:32 can be built three ways, and only the third puts a splitter in both boxes:

• Single-level, splitter at the FDT. One 1:32 splitter at the cabinet, and every FAT becomes a passive splice-and-patch box. Clean and cheap in hardware, but each subscriber then needs a dedicated distribution fiber — 352 subscribers means 352 fibers leaving one cabinet. Rarely practical.
• Single-level, splitter at the FAT. One 1:32 splitter in each access box. Cheap on distribution fiber, but a 32-port box weight and cable bundle at a pole is awkward, and a single splitter fault takes 32 homes down.
• Two-level. 1:4 at the FDT, 1:8 at the FAT. This is the shape most European and Asian FTTH builds settle on, and it is the one worth sizing properly.

In the two-level case, the ratios are not interchangeable. At a fixed total split, the first-level ratio decides how many distribution fibers you pull, and the second-level ratio decides how many drops each FAT carries:

All three rows deliver the same 32-way split and the same optical budget penalty (within a few tenths of a dB). What changes is the cost, and it moves in one direction only: heavier level-1 splitting multiplies both the distribution fiber count and the number of boxes on the street. Going from 1:2/1:16 to 1:8/1:4 takes distribution fibers from two per OLT port to eight and FATs from 24 to 96, because a four-drop box serves a quarter as many homes as a sixteen-drop box. What the extra boxes buy is small enclosures and a small outage footprint — four homes behind each door instead of sixteen. So heavy splitting at the FDT is the right call when distribution cable and duct are cheap to build, when the FAT positions are tight (a pole mounting or a narrow wall that will not take a sixteen-drop box), and when a single fault must not take sixteen homes down; the price is four times the boxes, four times the distribution fibers and four times the visits. It is the wrong call when the goal is fewer boxes or fewer visits, and it never reduces truck rolls, because every new connection is still a FAT visit. Split lighter at the FDT (1:2/1:16) when those eight distribution fibers per OLT port are the expensive part and each enclosure can be large — 24 boxes, sixteen drops each, one door to open for sixteen customers.

The optical cost is real but modest. A PLC 1:4 adds roughly 7.3 dB and a 1:8 roughly 10.7 dB, so a two-level 1:4 + 1:8 cascade lands near 18 dB of splitter loss against about 17 dB for a single 1:32 — plus the extra connector pairs at the second splitter. On a GPON class B+ link, which allows 28 dB of total loss, that remains workable; on a long-reach or high-split design it is the first thing to model, which is why splitter ratio selection and the loss a passive splitter adds are worth settling before you order enclosures. XG-PON and XGS-PON raise the split ceiling to 1:128 — set out in ITU-T G.987.1 for XG-PON and G.9807.1 for XGS-PON — but the cascade arithmetic does not change, only the power budget you have to close.

Sizing math: how many subscribers per cabinet

Port counts on a datasheet tell you what a box can hold. Sizing tells you what to buy. The inputs are homes passed, take rate at the end of the planning horizon, total split ratio, and a spare-port policy.

Worked example — one FDT service area:

1. Homes passed: 640.
2. Take rate at year 3: 55% → 352 subscribers.
3. Total split: 1:32 as 1:4 at the FDT and 1:8 at the FAT.
4. OLT ports: 352 ÷ 32 = 11 → provision 12 (one spare port).
5. Feeder fibers: one per OLT port → a 12-fiber feeder cable.
6. FATs: 352 subscribers ÷ 8 drops per FAT = 44 FATs → provision 48 FATs, which is 384 drop ports (about 9% spare).
7. Distribution fibers: one per FAT position → 48 FAT positions need a 48-fiber distribution cable, i.e. four 12-fiber elements.
8. Level-1 splitters: 12 × 1:4, one per OLT port, in 12 cassettes.
9. Coverage per FAT: 640 ÷ 48 ≈ 13.3 homes passed, which at typical suburban frontage is a service radius of a few hundred meters.

Three habits make that arithmetic survive contact with reality.

• Size fiber in 12s. Cables come in 12-fiber elements, so the 44 FATs you need become a 48-fiber cable, not a 44-fiber one. The same rounding applies to FDT ports: a 48-port cabinet is a catalog item, a 44-port cabinet is not.
• State the spare policy explicitly. Ten percent spare ports on the FAT side and one spare OLT port is a defensible default; zero spare is a decision that shows up as a mid-build redesign. Spare fiber matters more than spare ports — you can add a splitter later, you cannot add a duct cheaply.
• Watch the two ratios against each other. Moving from 1:4/1:8 to 1:2/1:16 halves the FAT count (48 → 24) and doubles the drops per box (8 → 16): you buy and install half as many enclosures, but each is larger and puts twice as many homes behind one door. Moving the other way, to 1:8/1:4, doubles the FAT count to 96 for four drops per box. Neither is wrong; pick deliberately, write down why, and re-derive the box count from the level-1 ratio before you commit.

Enclosure ratings, port counts and the cost trade-off

Both boxes are passive, so the specification argument is about mechanical protection, not electronics. The rating that actually matters is the one the local environment justifies — IP54 (dust protected, splash resistant) is common on sheltered wall boxes, IP65/IP66 (dust tight, water jets) on exposed pole and cabinet positions, and IP68 only where standing water or frequent flooding is real. Over-specifying to IP68 on a dry wall buys nothing and costs door-seal service life; under-specifying on a coastal pole buys water ingress and a return visit.

The trade-off that decides most projects is not the rating but the node count:

Two mechanical details deserve a line in the spec sheet because they cause most of the real-world rework. First, bend radius: a FAT with tight internal routing needs bend-insensitive fiber — ITU-T G.657.A2 fiber is rated to a 7.5 mm minimum design radius (A1 to 10 mm), against roughly 30 mm for standard G.652.D — and a cabinet that stores large slack coils needs the space to keep them above that radius. Second, adapter orientation: angled adapter plates shed laser backscatter and keep patch cords from kinking, which matters more in a box that a technician opens at night than in a catalog photo.

Installation and serviceability

A FAT that cannot be serviced without a ladder in a live roadway is a design fault, not an installation problem. The practical rules that separate a good build from a cheap one:

• Working height. Mount FATs at roughly 1.2–1.8 m above finished level. Below that, kneeling in wet ground; above that, a ladder for every connection.
• Clearance. Allow about a meter of working space in front of an open FAT door and 1.2 m at an FDT, and check it against the sidewalk width, not the drawing.
• Slack storage. Drops need a service loop; a box with no slack chamber forces splices into the cable tail and shortens the next repair.
• Access route. Pole-mounted boxes need a safe climbing position; wall-mounted boxes need a wall that will still be there in ten years.
• Records. Label the FDT port, the splitter output and the FAT port in one scheme. Fault-finding starts at the splice closure and ends at the right port only if the records agree with the plant.

Grounding is the detail people forget. A metallic FDT in aerial plant should be bonded to the messenger strand, and in high-lightning or coastal areas a non-metallic enclosure is often the better choice precisely because it removes that requirement. Condensation is the other quiet failure: a sealed box that heats and cools daily will pull moisture through any gland that is not properly sized, which is why vented or breathable designs exist for humid climates.

A decision checklist

Before the order goes out, each line should have an answer on paper:

1. Which boundary does this node own? Feeder/distribution → FDT. Distribution/drop → FAT. If a single node is being asked to do both, split it.
2. What total split, and how is it divided? Fix 1:32 (GPON) or 1:128 (XGS-PON) first, then choose the level-1/level-2 pair, then check the optical budget with real splice and connector counts.
3. How many homes behind one door? Derive FAT count from take rate, not from homes passed.
4. What comes in and what goes out? Feeder count and distribution count set the port count; round to 12s.
5. What is the spare policy? Ports, fibers and OLT ports each need a stated number.
6. What does the environment demand? IP rating, UV stability, metallic or non-metallic, pole or wall or plinth.
7. Can it be serviced safely? Working height, clearance, slack, access, labeling.
8. Does the record scheme match the plant? One naming convention from OLT port to subscriber.

FAQ

Is an FDT the same as an FDH? Largely yes. "Fiber distribution hub" is the term common in North America, "fiber distribution terminal" in ITU-influenced markets and Europe; both describe the cabinet where feeder cable breaks out into distribution cable. Where a vendor distinguishes them, the FDH usually implies cross-connect patching rather than splicing, and often a larger port count.

Can a FAT contain a splitter? Yes, and in distributed-split designs it usually does. A FAT with an integrated 1:8 splitter and eight drop ports is the standard second level of a 1:4 + 1:8 build. A FAT can even carry the whole split in a single-level design — but then it is doing the FDT's job, the feeder/distribution boundary has moved out into the street, and the distribution cable is being used as feeder.

How many subscribers can one FAT serve? Four, eight or sixteen, with eight the most common compromise between enclosure size and node count. The number that matters is homes passed, not subscribers: at a 55% take rate an 8-port FAT covers about 14–15 homes passed.

Does either cabinet need power? No. Both are passive optical enclosures; the only powered equipment in a PON is the OLT at the head end and the ONT at the subscriber. That is precisely why splitter placement, not electronics, is the design lever.

Can you skip the FAT and run drops straight from the FDT? In low-density rural builds where drops are short and few, sometimes. In any suburban grid it moves the distribution/drop boundary into the cabinet, which means every new connection is a cabinet visit and every drop cable is a long home run. The cost shows up later, in operating expense rather than capital.

The short version

The FDT is the last point where you think in cables; the FAT is the first point where you think in customers. Choose the split ratio in pairs, size the boxes from take rate and spare policy rather than from the datasheet maximum, and spec the enclosure against the environment it will actually live in. If you are still choosing between access technologies rather than cabinets, the OLT, ONU/ONT and ODN overview is the right place to start, and the FTTH/FTTx solution set covers the hardware that fills both boxes.