# CATV to Fiber Infrastructure Survey: Coax Layers Before Overlay

**Title tag:** CATV to Fiber Infrastructure Survey 2026
**Meta description:** CATV to fiber infrastructure survey for cable operators: actives and passives, node and power sites, RF plant records, overlash, duct capacity, reuse outputs.
**Author:** Omar Molina
**Published:** September 24, 2026  
**Last updated:** September 24, 2026  
**Category:** ISP & Carrier Networks / Data Center
**URL:** https://draftech.com/blog/catv-to-fiber-infrastructure-survey
**Primary keyword:** catv to fiber infrastructure survey
**Word count:** 3187
**Read time:** 13 minutes

![Field surveyor carries a measuring wheel and equipment case along a suburban grass verge.](../../blog/img_catv_to_fiber_infrastructure_survey.webp)

---

A cable operator can hold a current node map and still lack the plant facts that decide whether fiber reuses the coax corridor or overlays it. The map shows serving areas. It does not prove amplifier identity or power-supply loading. Leakage history and remaining strand or duct space stay unproven on that map.

This guide stays on that survey. We do not run the migration engineering here. We show which coax-plant layers the field package must return so the later reuse-versus-overlay decision has evidence instead of a hopeful GIS line drawn from a serving-area polygon that never saw the housing.

## What a CATV to Fiber Infrastructure Survey Records

A CATV to fiber infrastructure survey is the field and records package that inventories an incumbent cable operator's coax plant before a reuse-versus-overlay call. Draftech's package covers 4 coax-specific layers: actives and passives; node and power-supply sites; RF plant records; overlash or duct capacity. The survey names what is present. It does not choose the overlay architecture.

The buyer for this work is the incumbent cable operator, not an overbuilder writing a greenfield HLD. The operator already has plant in the street. The question is whether that plant can host fiber. Overlay may be required. A named gap can also force a hold until evidence closes. We write that question on the kickoff sheet before a collector opens a form, because a crew that starts without a named overlay question will collect pretty photos and still miss the layer that later design will actually spend.

Coverage is not the same fact. CostQuest Associates' Broadband in America Report, Cable Market Focus, May 2026 edition, reports coaxial service at 100/20 Mbps available to approximately 95.1 million broadband serviceable locations, or 81.7 percent of locations, as of June 30, 2025. That figure describes reach. It says nothing about the amplifier housing on a given span or the spare innerduct in a given run.

Fiber pressure is real and still does not write the survey. The Fiber Broadband Association, in "U.S. Home Fiber Deployments Top 88M homes passed" dated April 17, 2025, citing the RVA LLC North American Fiber Deployment Report released in January 2025, reported 88.1 million U.S. homes passed with fiber. An operator facing that passing count still needs plant-level evidence before it spends overlay capital on a corridor that might already have usable pathway.

Safety bounds collection. The OSHA Telecommunications standard, 29 CFR 1910.268, is the named federal reference for telecommunications work. The employer's approved procedure still governs the crew. Worker qualifications and traffic control stay in force, and access authorization does too. We never ask a collector to enter a pedestal or to climb. We never ask them to approach energized plant outside that method.

The trade-off in our 4-layer package is extra office time before mobilization. A photo-only walk looks faster. It leaves power supplies and RF files as later surprises, which is when overlay bids start inventing spare capacity. We accept the setup cost because an unknown that stays named is cheaper than an unknown that ships as a design assumption and later becomes a bid quantity nobody can actually build.

> **A node map is not a survey.** Serving-area polygons can be current and still miss the housing module, the supply loading, and the dated leakage file that actually control the overlay call.

## Four Coax-Plant Data Layers That Feed Overlay

We collect the 4 layers as separate records on the same stable IDs. Merging them too early hides the conflict that the overlay decision needs to see. The table is the preview map. Later sections expand each row without turning this post into upgrade engineering.

| Layer | Field question | Required record | What overlay engineering will ask |
| --- | --- | --- | --- |
| Actives and passives | What devices sit on this leg? | Housing ID, installed module, tap values, cascade order | Can the chain stay, or must fiber replace it? |
| Node and power-supply sites | Where do optical and power plant sit? | Node port, cabinet, supply loading, access | Can a node stay, move, or be overbuilt? |
| RF plant records | What does the plant actually pass today? | Dated sweep, leakage, ingress, topology ID | Is this a health problem or a capacity problem? |
| Overlash and duct capacity | Is there physical room for fiber? | Occupancy, remaining space, method, unknown | Reuse the corridor, or cut a new overlay path? |

Each row gets one of 4 statuses we use on CATV plant: observed in the field; supported by an operator record; held as an open exception; excluded from the current overlay question. A housing that is photographed but whose module is unread is not "good." It is held. Status travels as an attribute, not as a color on a print that will be lost in a CSV.

Our [HFC network design and coax plant engineering](/services/hfc-cable-network) work consumes these layers after they close. Once those layers close, [OSP design for cable TV headend connectivity](/blog/osp-design-for-cable-tv-headend-connectivity) assigns the headend-to-hub and hub-to-node fiber. The survey's job is to make that consumption honest. If a layer is missing, the design package should show a hold, not a guessed tap value that later overlay drawings will treat as a measured fact.

We do not run the general reuse-evidence framework here. That six-group OSP method lives in the [existing infrastructure survey guide](/blog/osp-existing-infrastructure-survey-guide). This post is narrower: coax plant, incumbent cable operator, overlay question. Poles, guys, and generic handhole continuity stay in that other package unless they are the node site or the duct the overlay will use.

## Node Sites and Power Supplies on the Active-Passive Chain

Start with the node. If the node port and the serving-map leg disagree with the field label, stop. Do not credit a cascade that belongs to a different branch. Give the node a stable identifier that appears on the site sketch; the device chain; the RF file names; the exception log. One wrong port number makes a precise-looking chain describe plant that is not there.

### Start with the node and the supply

Node sites and power-supply sites are locations with access, space, and loading, not just dots. We record the cabinet or pedestal identity, the optical interface when it is visible under the approved method, the supply or inserter, protective devices, and whether a technician can actually reach the plant. A supply that exists on the schematic and cannot be opened is a hold, not a reuse candidate, because overlay engineering cannot spend capacity it has never verified at the cabinet.

Powering belongs in the survey because overlay fiber does not erase the coax chain that still has to run until cutover. Current loading and voltage drop at actives are operator-limit questions. We capture the installed state and the evidence source. We do not declare that a supply can support a later spectrum plan. That declaration is upgrade engineering, and it waits for the [HFC coax upgrade engineering](/blog/hfc-coax-upgrade-engineering) package.

Access failures stay granular. A locked cabinet is not a missing node. A flooded vault is not an absent supply. Record what could not be observed and why. Name the next authorized action. Guessing a module to close the form is how a later designer inherits a tap value that never existed.

### Record the cascade without crediting bandwidth

Actives and passives are a chain. We want housing identity, installed module or faceplate, tap values, splitters, directional couplers, and order from the node port to a representative endpoint. A family name on the housing is not the module. If the faceplate cannot be read, the calculation that would have used it stays blank.

We refuse to assign a convenient nominal bandwidth to an unread passive. One thing I still refuse is treating a housing photograph as proof of the installed response. The photo proves the box was there that day. Response data comes from the module identity and the manufacturer's document revision, or it does not come at all.

Mechanical facts stay on the same record. Connector type, spacing, enclosure constraints, and grounding can make a device electrically interesting and still unusable in place. Record those as mechanical exceptions. Do not bury them in a constructability note that overlay engineering will never see.

The cascade record is not a spectrum model. CableLabs' DOCSIS 4.0 Physical Layer Specification, CM-SP-PHYv4.0, describes Extended Spectrum DOCSIS and Full Duplex DOCSIS interfaces. It does not survey the installed chain. We keep that specification out of the survey form so a collector cannot "pass" a housing against a document it never measured.

## RF Plant Records Versus Pathway Occupancy

RF health and physical room are different overlay inputs. A clean sweep does not prove spare overlash space. An empty innerduct does not prove the return path is quiet. We keep the two layers on the same IDs and we keep their statuses independent, because overlay engineering will ask both questions and will get a false yes if we fold them together.

### Dated RF files travel with topology

RF plant records are files with dates, not a verbal "the plant is fine." We attach leakage captures; ingress history; sweep or telemetry traces; the topology revision they belong to. A trace taken after a temporary plant adjustment cannot silently become the ordinary baseline. If the filename cannot be reopened against the same branch ID, return it.

ANSI/SCTE 257 2024, Techniques for Accurate Measurement of Cable TV Network Downstream RF Signal Levels, addresses accurate downstream RF level measurement. It does not establish overlay acceptance. The operator's approved procedure still names instruments and calibration. It also names test points and limits. We record which procedure was used so a later reviewer is not left with a number and no method.

Use 47 CFR 76.605, Technical Standards, only for the cable-system technical requirements within its scope. 47 CFR 76.611 addresses signal leakage and CLI. Aeronautical-band rules sit in 47 CFR 76.610, with frequency-separation in 76.612 and emergency-frequency limits in 76.616. Those rules are not an overlay recipe. The operator identifies applicability. The survey preserves the leakage practice the operator already runs, dated and tied to the branch, instead of inventing a pass rate.

Intermittent return-path noise is the usual trap. A quiet morning capture can hide an evening ingress condition. We record when the issue appears. Duration and threshold policy stay with the operator. The evidence record should not invent persistence it did not observe.

> **Do not let a clean trace hide a crowded span.** RF health and remaining pathway space answer different overlay questions. They may sit on the same pole and still require different crews.

### Pathway occupancy is not a strand map

Overlash and duct capacity are occupancy attributes, not a full aerial plant assessment. We record how many sheaths are visible on the support, whether a duct is occupied at an authorized structure, the method used, and whether remaining space is measured or estimated, or else unknown. Unknown is a valid status. Converting unknown to spare before verification is how overlay drawings grow a second cable that will not fit.

We defer the strand-walk method and span-by-span aerial assessment to the strand mapping and aerial plant assessment process. This survey asks a narrower question: is there physical room for a fiber sheath on this CATV corridor, and what evidence supports that claim. A crowded photograph is occupancy evidence. It is not a pole-loading conclusion and it is not a make-ready ticket.

Underground occupancy follows the same honesty rule. An authorized handhole visit can show visible ducts and apparent plugs. It does not prove destination or internal condition. Pull feasibility stays unproven until a named method runs. Until that method runs, the duct stays held for overlay use even if the structure itself is fully documented.

The OSHA bound still applies here. Rodding and mandrel work stay outside this survey unless the operator authorizes that method. A collector who cannot enter a structure records the access failure and leaves. Overlay engineering can request a specialized verification later. It cannot spend a guessed spare duct today.

## Survey Outputs the Overlay Decision Will Consume

The package has to be usable by someone who was not in the field. We hand overlay engineering a layer register, not a folder of unmarked photos. Each accepted segment names the 4 layers and their statuses. It also names the files that support them. Each held segment names the gap and the design consequence. It also names the closure action.

- **Layer register:** one row per node-to-end chain with the 4 statuses plus source revisions and exception IDs.
- **Device chain:** housing and module, then tap and splitter, with coupler order tied to the same chain ID.
- **RF file index:** dated captures linked to topology, not to a generic node name.
- **Pathway occupancy sheet:** overlash or duct evidence plus method; remaining-space status; unknowns left visible.

Those 4 outputs are what the reuse-versus-overlay call will actually read. The call itself lives in cable plant upgrade engineering. We do not make the architecture choice in this post. We make it possible to make that choice without pretending spare fiber already exists in the coax plant because a corridor photograph was taken from the road.

Export is part of the survey, not a later IT task. We test at least 1 representative chain through the receiving GIS or CAD workflow before we call the package closed. If export strips IDs, methods, or evidence links, the native dataset is not finished. Correct field work that dies in a shapefile is still a failed handoff.

Revision history stays with the record. When a module ID changes after a second visit, keep the first observation and the reason it changed. Overlay engineering needs to see that the value moved, because a silent overwrite is how two teams argue about which tap was "always" there.

Draftech performs this survey engineering in-house. When construction is later included on an overlay, delivery is full turnkey through Draftech-managed subcontract crews under our QA/QC and safety program. Field deviations still return to engineering. A crew photograph does not rewrite a held layer.

## CATV to Fiber Infrastructure Survey Decision by Role

**Plant records manager at a cable operator:** freeze source revisions before the first field day and keep GIS and RF files on one ID scheme with the node map. Do not send crews until conflicts between those sources are listed as investigation tasks. A clean-looking extract that mixes three dates is worse than a dated conflict log.

**OSP engineering lead preparing an overlay HLD:** accept survey segments only where all 4 layers are closed or explicitly held. Do not credit pathway occupancy from a node map. If the overlay must move while RF files are still missing, isolate those chains from the bid quantities rather than averaging them into a spare-capacity percentage.

**Overlay program owner:** fund the 4-layer survey as a gate, not as a week of photography. The Fiber Broadband Association's April 17, 2025 passing count explains why the program exists. It does not tell you which of your own corridors can take a sheath. Spend the survey money before the overlay capital, not after the first jammed duct.

**Field operations lead:** schedule against access and approved method, not against a mile target. A locked supply site is a held day. Counting it as production hides the layer that overlay engineering will ask for on Monday.

The recurring CATV-to-fiber failures we still see are unread modules credited as passives, node ports that do not match field labels, RF files with no topology, and pathway space marked spare because the photo was taken from the road. We remove those by keeping the 4 layers separate until each one has evidence. If a node area needs that register built before overlay design, [email our HFC survey team](mailto:info@draftech.com).

A cable operator that wants the overlay question tested on a qualifying corridor can start with a route. We [engineer the first 20,000 linear feet of a qualifying build at no cost](/free-design), from feasibility and field survey through permit approval, and the 4-layer CATV record described above is how that coax plant enters the design. The offer is open to cable operators and ISPs. Utilities and cooperatives qualify too.

> **[Talk to our HFC survey team about your node area.](/#dt-contact)** Bring the node map, one RF file set, and the current pathway record so the review starts on the layers that actually change the overlay call.

## Frequently Asked Questions

### What does a CATV to fiber infrastructure survey include?

A CATV to fiber infrastructure survey records 4 coax-specific layers before a reuse-versus-overlay call: actives and passives, node and power-supply sites, RF plant records, and overlash or duct capacity. Each layer carries identity, evidence source, method, and remaining gaps. The package names what is present. It does not choose the overlay architecture. The cable operator retains spectrum policy and network release.

### Does a current node map replace the survey?

No. A node map shows serving areas. It does not prove amplifier identity, tap values, power-supply loading, or remaining strand and duct space. CostQuest's Broadband in America Report, Cable Market Focus, May 2026 edition, reports coaxial 100/20 Mbps service at 95.1 million broadband serviceable locations as of June 30, 2025. Coverage is not plant evidence. Hold overlay credit until the 4 layers are recorded.

### Which RF records belong in the survey package?

Keep dated leakage, ingress, and sweep or telemetry captures with the topology they belong to. ANSI/SCTE 257 2024, Techniques for Accurate Measurement of Cable TV Network Downstream RF Signal Levels, addresses downstream RF level measurement and does not set overlay acceptance. 47 CFR 76.605 states cable-system technical standards. 47 CFR 76.611 addresses leakage and CLI. The operator names applicability. The survey preserves files, dates, and branch IDs.

### How is overlash or duct capacity recorded without a strand walk?

Record occupancy and remaining space as a pathway attribute with method and confidence, not as a full aerial plant assessment. Visible overlash count, duct occupancy at an authorized structure, and a named unknown are valid states. The strand-walk method lives in the aerial plant assessment guide. Do not convert a photo of a crowded span into a loading conclusion. One authorized structure visit proves only what that method can see.

### Who decides reuse versus overlay after the survey?

The cable operator decides. The survey returns per-layer dispositions: reuse candidate, overlay required, hold for evidence, or exclude. Cable plant upgrade engineering and HFC coax upgrade engineering own the architecture and spectrum decisions that consume those outputs. Draftech performs engineering in-house. When construction is included, delivery is full turnkey through Draftech-managed subcontract crews under our QA/QC and safety program.

### What closes the survey for engineering handoff?

A closed package lists accepted segments and open exceptions in the receiving GIS or CAD path, with the 4 layers attached to stable IDs. At least 1 representative node-to-end chain should survive export with identity, method, confidence, and evidence links intact. Design may start on accepted segments. Held layers stay out of overlay credit. Completeness of the last photo dump is not the test.