FOA CFOS/FC common mistakes
FOA CFOS/FC · Understand the traps and how to avoid them.
When measuring a link, the candidate takes the reference (one-jumper or two-jumper) with the launch cords connected to the power meter and source, then disconnects at the meter and plugs the meter into the far end of the link without re-zeroing or accounting for the connector removed from the reference. The reported insertion loss is therefore low by roughly one connector pair (about 0.3–1.0 dB), and the link appears to pass when it may not.
The candidate memorized 'set the reference first' as a ritual but never learned which connectors are included in each reference method. Because the one-jumper reference deliberately excludes one mated pair, the meter must be connected to the link with that same excluded connector still in place; swapping cords silently changes the math.
What to do instead
Before every test, write down the reference method and list exactly which connectors are inside the reference. For one-jumper reference, keep the launch cord on the meter and mate it to the link's far-end connector, so the excluded pair is the one you add back. For two-jumper reference, both pairs are included and the link loss should be read directly. Re-zero only when the source, meter, or wavelength changes, and record the reference value on the test record.
The candidate cleans a connector by wiping the ferrule endface on a dry shirt or by blowing on it, then inspects and sees a pass. Later the same connector shows contamination or a scratch, and the candidate blames the test equipment rather than the cleaning process.
Dry wiping and blowing are common habits from consumer electronics and seem harmless. In reality, dry wiping drags grit across the endface, and breath carries moisture and particles. The candidate has no written cleaning procedure to compare against, so the habit survives.
What to do instead
Use the wet-then-dry method: lint-free wipe with approved fluid for the first pass, then a dry lint-free wipe, one direction only, no back-and-forth. Use one-click cleaners for adapters and never reuse a wipe. Inspect with a scope before and after, and log the result. Treat any visible scratch or pit as a fail and re-terminate or replace, not as 'probably fine'.
During endface inspection, the candidate looks only at the center of the ferrule, declares it clean, and moves on. Contamination or damage near the cladding edge or under the ferrule shoulder is missed, and the connector later fails intermittently under temperature change or vibration.
The core is where the light travels, so attention naturally goes there. But the acceptance criteria define zones out to the cladding and contact areas, and edge chips or epoxy beads can migrate onto the core or prevent physical contact. Candidates also inspect at too low a magnification or with poor illumination.
What to do instead
Use a scope with adequate magnification and coaxial illumination, and systematically scan the whole ferrule face in zones: core, cladding, contact, and ferrule edge. Apply the standard zone criteria for scratches and defects rather than a gut call. Capture an image for the record when a defect is borderline, and compare against the acceptance table, not memory.
The candidate performs an OTDR trace from one end, sees a large reflective spike near the far end, and reports it as a break or a bad connector without checking the launch and receive conditions. The event is actually a ghost or a gainer caused by a reflective far-end connector, and the real fault is elsewhere.
OTDR traces are visually dramatic, and a big spike looks like a fault. Ghosts and gainers require understanding of reflection and of the difference between splice loss and reflective event loss, which many candidates skip because they only practice on clean links.
What to do instead
Always use a launch cord (and receive cord where possible) long enough to move the first and last events out of the dead zone. Check the trace from both ends when a fault is suspected. Recognize ghosts as events that appear at distances beyond the fiber length or that repeat at regular intervals, and recognize gainers as negative loss at a splice caused by a reflective event before it. Re-test with a shorter pulse or with the far end terminated in index-matching gel to confirm.
The candidate tests a PON from the OLT with a live 1490 nm downstream signal present and either damages the power meter or reads a value that includes the live traffic, then reports the splitter loss as wrong.
PON testing looks like ordinary link testing, so candidates reuse the same setup. They do not account for live wavelengths, the need to test at 1310, 1490, and 1550 nm separately, or the fact that a live PON carries traffic that corrupts loss and ORL measurements.
What to do instead
Confirm whether the PON is live before connecting anything, and use a test port or maintenance window as required. Measure at each wavelength separately with a wavelength-selective meter, and keep the upstream and downstream directions distinct. For loss testing through splitters, use a source and meter sized for the split ratio and verify the expected splitter loss against the design budget before trusting the result.
When calculating a loss budget, the candidate adds fiber attenuation and connector loss but forgets splices, splitter loss, or the margin, or uses the wrong attenuation coefficient for the wavelength in use. The computed budget passes a link that the field test then fails, or rejects a link that is actually within specification.
Budget formulas look simple, so candidates do them from memory and skip the line items that do not appear on every link. They also mix 1310 nm and 1550 nm coefficients, or forget that a PON splitter contributes a large fixed loss that dominates the budget.
What to do instead
Build the budget from an explicit table: fiber length times the coefficient for each wavelength, number of mated connector pairs times the per-pair allowance, number of splices times the splice allowance, splitter loss from the split ratio, and a stated margin. Show the units and the wavelength on every line. Compare the field measurement to the budget only after confirming the reference method matches the connector count.
The candidate reports a measured loss to two decimal places and treats a 0.05 dB difference between two links as meaningful, without stating the uncertainty of the measurement or the conditions under which it was taken.
Meters display fine resolution, so candidates equate resolution with accuracy. They have not been taught that source stability, meter linearity, connector repeatability, and reference method each contribute uncertainty, and that a single measurement cannot resolve differences below that combined uncertainty.
What to do instead
State the measurement uncertainty alongside every result, and treat differences smaller than the combined uncertainty as inconclusive. Control the contributors you can: stabilize the source, use the same reference cords and method for compared links, clean every connector, and repeat measurements to check repeatability. Report the wavelength, reference method, and uncertainty on the test record.
The candidate uses a visual fault locator to declare a link good because no red light escapes, without ever measuring loss or checking the trace. A high-loss but continuous link passes the 'VFL test' and is shipped.
The VFL is cheap, fast, and satisfying, so it becomes the default check. Candidates forget that it only detects breaks and severe bends near the surface, and says nothing about attenuation, splice quality, or connector loss.
What to do instead
Use the VFL only for what it does: locating breaks and tight bends, and confirming fiber identity at a patch panel. Always follow with an insertion loss test against the budget, and use an OTDR when the loss fails or when splice and connector events must be located. Record which instrument produced which conclusion.
The candidate tests multimode fiber with a source and meter set to the wrong launch condition, or tests a link at 850 nm with a patch cord designed for 1310/1550 nm single-mode, and reports loss that is several dB off.
The instruments look interchangeable and the connectors mate, so the mismatch is invisible. Candidates do not check core size, launch condition (mandrel wrap or mode filter for multimode), or whether the source is the correct type for the fiber under test.
What to do instead
Before testing, confirm fiber type, core size, wavelength, and required launch condition from the design. For multimode, apply the specified mandrel wrap or mode filter and use multimode-rated cords. Verify the source and meter are set to the same wavelength, and record the launch condition on the test record so the next technician can reproduce it.
The candidate works through a troubleshooting checklist in a fixed order regardless of the symptom, spending time on OTDR traces when the failure is an unseated connector, or replacing connectors when the real problem is a dirty adapter or a bend radius violation.
Checklists are learned as sequences to complete, not as decision tools. Under time pressure the candidate follows the list top to bottom instead of starting from the symptom and the simplest, most likely cause.
What to do instead
Start from the symptom and the topology: confirm power at the source, then at each patch point, then at the receiver, cleaning and reseating before replacing anything. Use the VFL for continuity and the OTDR for event location, and check bend radius and cable management before condemning hardware. Keep a written record of what was changed so the fix can be verified and repeated.