Search results for this topic mix two different machines. An HFC coax line amplifier restores radio-frequency levels on coaxial plant. An optical amplifier restores optical power on fiber. This guide covers only the optical device.
We park candidate huts from the named project's optical budget and the selected-equipment datasheets, then we refuse to move a coordinate until that budget and those datasheets agree on the input-power window. We do not publish a mileage rule. The release question is whether eight site inputs are closed, not whether the map looks evenly spaced.
Fiber Amplifier Placement Design Guide: Optical vs HFC
A fiber amplifier placement design guide is a candidate-site release record for optical amplifiers on long-haul or middle-mile fiber, not an HFC coax line-amplifier design. G.661 (05/2025) defines the device parameters. G.663 with Amendment 1 (September 2016) sets the applications. Eight inputs must close against that project's optical budget.
Recommendation ITU-T G.661, Definitions and test methods for the relevant generic parameters of optical amplifier devices and subsystems, was approved on 14 May 2025 and is the edition in force. It applies to optical amplifiers that use optically pumped fibres, whether rare-earth doped or Raman. Semiconductor amplifiers and waveguide amplifiers sit in the same Recommendation. The 05/2025 revision added the reference diagram and definition of the remote optically pumped amplifier.
Recommendation ITU-T G.663, Application-related aspects of optical amplifier devices and subsystems, April 2011, with Amendment 1 (September 2016), is the application text in force. G.663 identifies which aspects to consider for each application and states common parameter values and ranges for each type of optical-amplifier device. Those values belong to the equipment class. They are not a hut-spacing rule for a terrestrial route.
G.661 (05/2025) is the parameter and test-methods Recommendation. Amplifier family is a datasheet fact, not a G.661 letter code with Category A as EDFA. We write the family that the selected-equipment datasheet actually is. We do not upgrade a line item to EDFA because the acronym is familiar.
G.663 amplifier applications
G.663 names three applications, and the candidate-site sheet has to say which job the hut is doing. A booster raises launch power at the transmitter. Line amplifiers restore power along the span; a pre-amplifier raises sensitivity at the receiver. Recommendation ITU-T G.662, Generic characteristics of optical amplifier devices and subsystems (07/2005), specifies those generic characteristics for the optically amplified transmitter (OAT) and the optically amplified receiver (OAR). Power draw and monitoring change with that job, and so does the optical budget.
An HFC line amplifier is a different product on a different plant. It sits on coaxial distribution and restores RF. If a cable-operator packet uses "line amp" for both machines, split the language before anyone prices a pad. This guide does not design HFC amplifiers. It does not sell optical hardware or certify a manufacturer.
Our long-haul fiber engineering explainer is the place the optical model and the physical corridor have to agree. This page is the hut record that sits inside that optical model. Draftech's in-house ISP network engineering writes both. Engineering stays in-house. When construction is in scope, Draftech-managed subcontract crews deliver full turnkey under our QA/QC and safety program.
G.661 names the device. G.663 names the job. Neither Recommendation parks a hut on a milepost. Copy gain and noise figure from the selected-equipment datasheet onto the project span sheet, then the power window. Ask whether a site can actually hold that device.
The Eight Amplifier-Site Release Inputs
We will not release an optical-amplifier candidate until every row below is closed against that project. A convenient access road is not a substitute for an optical budget. A catalog page is not a substitute for the datasheet of the unit that will be installed.
| Release input | What must be on the sheet | Hold the site if |
|---|---|---|
| Span loss | Loss for this span from the project optical budget, using the selected cable, splice, and connector assumptions | The sheet uses a generic dB/km with no event list |
| Channel plan | Channel count, band, and grid from the owner design for this system | Channel count is still TBD while gain is already written |
| Power | Hut load and backup from the selected-equipment power draw | Service is assumed available with no utility letter |
| Shelter | Space, HVAC, and cable entry from the selected-equipment outline | Pad sketch with no thermal load |
| Access | All-weather craft access from the ROW and site packet | The only approach is a locked private gate with no recorded right |
| Rights | Site control, easement, or permit path in writing | Verbal landlord interest |
| Resilience | Path and pump diversity relative to the route model | Single fibre and a single pump with no restoration note |
| Maintainability | Craft access, monitoring, and spare policy on the same sheet | No monitoring port and no spare documented |
Those eight rows are the whole release. We do not add a ninth "engineering judgment" column that lets a pretty alignment skip a blank. If a value is not on the named budget or the selected datasheet, it is not a value. The sections that follow expand the rows in the same order.
Span Loss and Channel Plan
Span loss is the first optical question because everything downstream is compensating for it. Copy the fibre type, the length of this span, the splice and connector events, and the passive devices from the project optical budget for this exact route. Then read the input-power window and the gain from the selected-equipment datasheet. If those two sheets disagree, the hut location is not ready.
Span loss from the project budget
One thing I still catch in our own reviews: a span-loss sheet that copies a catalog typical dB/km instead of the selected cable's current datasheet and the measured events on that route. That is how a hut gets parked at a convenient access road while the optical model is still fiction. G.661 defines gain and noise figure as device parameters. It does not fill in the span.
We keep the event list on the same sheet as the length. A splice moved across a jurisdiction line is not a drafting cleanup. It changes the loss the amplifier is being asked to cover. Until the event list and the datasheet input window agree, the candidate stays a candidate.
Channel plan before gain is written
Gain written against an unknown channel count is a guess dressed as a calculation. The owner design names the band and the channel count for this system, and we will not write a gain figure until that channel count is a number on the same sheet as the selected-equipment output-power limit. The selected-equipment datasheet names the gain flatness and the total output-power limit across that plan. We do not invent a channel load to make a hut look necessary.
Middle-mile packets fail this more often than long-haul packets, because the first DWDM card is treated as a future option. Our middle-mile fiber network design planning guide is the regional layer. If that layer is still dark fiber with no named channel plan, the optical-amplifier site is not a middle-mile hut yet. It is a building waiting for a system.
G.661 (05/2025) points safety parameters to IEC 60825-1 and IEC 60825-2. We record the selected product's current safety classification on the same sheet as the optical values. We do not treat an unclassified unit as released because the span loss closed.
Power, Shelter, Access, and Rights
The optical sheet can be perfect and the hut still fails. Power, shelter, access, and rights are the four civil lines that turn a coordinate into a site. Intercity fiber engineering already treats powered sites as part of corridor continuity. This section is the amplifier-specific hold that sits on those sites.
Power and shelter from the datasheet
Read the electrical load and the heat load from the selected-equipment datasheet before anyone asks a utility for a service that has not been sized to that chassis. Read the outline too. Then ask the utility and the shelter drawing whether they can take that load. "Power nearby" is not an answer. The letter that names the service and the backup, and who owns the meter, is the answer.
Shelter is the same test in cubic feet. A pad sketch with no HVAC load is a drawing, not a hut. Cable entry, grounding interfaces, and the space to open a chassis door belong on that sketch before anyone calls the site released. We hold a pad sketch drawn under a device whose current outline does not fit it. The pad was not the product.
Access and rights in writing
Craft has to reach the hut in the restoration window the owner actually wrote. A locked private gate with no recorded right is not access. A seasonal track that disappears in snow is not all-weather access. Write the approach and the key control on the sheet. Name the winter condition on that same line. If either is missing, hold the site.
Rights are written or they are not rights. Verbal landlord interest does not survive a later sale. An easement in draft is still a hold. The permit path for the pad and the generator is part of the same line, and so is the driveway, because a hut that cannot be built is not a hut. That limitation becomes a construction delay, and the next section is where we refuse to paper over it with a single-pump note.
A convenient pad with no electrical service letter is not a site. It is a wish. Copy the load from the selected-equipment datasheet, then wait for the utility letter that matches that load. Do not reverse the order.
Resilience and Maintainability
Resilience is the route-model question. Maintainability is the 2 a.m. question. They look similar on a slide. They fail differently in the field.
Resilience against the route model
A second fibre in the same sheath is not a second path. A second pump in the same hut is not geographic diversity. We test the candidate against the shared-risk register on the route, including the same bridge, the same duct bank, the same cable entrance, and the same upstream node that would take the protect path with it. If the protect path dies when this hut dies, the resilience line is still open.
G.661 (05/2025) added ROPA. A remote optically pumped amplifier puts doped fibre in the transmission path and supplies pump light from a remote site. That remote site is still a site. It still needs the power and access lines. Treating a ROPA as a way to delete huts without deleting pump locations is how the resilience sheet lies.
Maintainability the craft can use
Someone has to test this device after we leave. Replacement of failed units and live monitoring follow from that same craft visit. The sheet names the monitoring port and the spare policy. Craft access that makes a chassis swap possible belongs on the same sheet. No monitoring port means the first fault is a truck roll with no alarm. No spare policy means the restoration clock is a purchasing clock.
We do not write "standard spares" and call it done. The selected-equipment datasheet names the replaceable units. The owner names which of those units sit on the shelf. If those two lists have not met, maintainability is open. Hold the hut.
Fiber Amplifier Placement Design Guide: Release Decision by Network Type
If you are a regional ISP building middle-mile aggregation: Freeze the channel plan before you pick a hut, because a middle-mile hut released against a dark pair is a pad that will not match the first DWDM card you actually buy. If the first DWDM card is still a future option, do not release an optical-amplifier site against a dark pair and a catalog gain. Close span loss and the four civil lines on the segments you will light. Leave the unlit segments as future candidates with no pad in the ground.
If you are a carrier or long-haul operator: Run the optical model and the site packet as one revision. A route shift that adds splice events has to reopen the hut list. A datasheet change that tightens the input-power window has to reopen it too. Do not let a civil team lock pads while the optical sheet is still using last quarter's product.
If you are a dark-fiber or IRU operator handing off to a lit tenant: Keep the hut record in the landlord packet even if the tenant will own the cards. Span loss and access remain yours, and so do the rights. Channel plan and gain may be the tenant's, and monitoring often is too. Write that split on the sheet so a later dispute does not start from a blank.
When construction asks for pad coordinates, we release only the sites whose span sheet already matches the unit on order and whose utility letter names the service that will feed that load. Verbal landlord interest stays off that list. The next revision can add a held site once the named optical budget and the selected-equipment datasheet agree, without reopening pads that already match the current product.
Draftech writes that construction list in-house. Active in 24 states. Available across all 50 U.S. states. If the civil packet is already running ahead of the optical sheet, that is the mismatch in-house ISP engineering is built to close. For a candidate list already in draft, reach out at info@draftech.com.
If fiber amplifier placement design guide is part of a qualifying route, Draftech engineers the first 20,000 linear feet at no cost, from feasibility and field survey through permit approval. Request the free design package. The owner reviews each request before Draftech commits the package.
Need the hut record written against your optical budget? Talk to our ISP network engineering team about amplifier-site release. We will not sell you the amplifier. We will tell you which candidate can actually hold it.
