In March 2023, I watched a 12-fiber outside-plant cable fail a DWDM turn-up after our EXFO OTDR gave it a clean bill of health.
The trace looked good. The event table looked good. The printed report had green checks. But the client's DWDM optics still wouldn't come up the next morning. I re-tested with a second unit, and the difference showed up fast: fiber 7 had a splice point that the first unit measured at 0.17 dB. The second unit measured the same splice at 0.39 dB.
My first move was to blame the connector. Then the launch cable. Then the technician. I cleaned the bulkhead, swapped the jumper, ran the trace again. It still passed. That's when I looked at the calibration label. It said cal due 2022-06-30. I had been using an EXFO OTDR that was nine months out of calibration.
Why EXFO OTDR Calibration Isn't a Formality
I used to think calibration was administrative. EXFO Inc. builds professional test instruments, not cheap gadgets. If the unit didn't display a warning, I expected it to stay within specification.
That model is wrong. Here's the deeper problem: an OTDR doesn't measure fiber directly. It measures light that travels down the fiber and backscatters. To turn those measurements into distance and loss, it relies on internal references—laser power, detector sensitivity, time base. All of those can change with age, temperature, and use. The changes are usually small. They can also be consistent.
A consistent offset is the worst kind of problem because nothing on the screen looks wrong. There's no warning that says the unit is off by 0.2 dB. There's just a clean trace that happens to be wrong.
The Blood Pressure Monitor Problem
My father uses a C210 blood pressure monitor at home. It's a simple device. It gives a reading in seconds. On most mornings, it says 118/78. Clean result.
Then he took it to a doctor's appointment. The doctor's device measured the same arm and got 134/86. The C210 wasn't jumping around. It wasn't broken. It was consistent. Consistency is not accuracy.
The same thing happens with an OTDR. A unit can be repeatable, clean, and wrong at the same time. If the loss reference has drifted, it shifts readings in the same direction. You won't see it in one trace. You only see it when you compare the unit against a known-good reference or when a second instrument tells a different story.
What Skipping EXFO OTDR Calibration Actually Cost
Let me put a dollar number on this one mistake.
The first report listed a 0.17 dB splice on fiber 7. That was under the client's 0.30 dB threshold. The second unit measured the same splice at 0.39 dB. The difference was small, but it was enough to push the DWDM channel outside the design margin.
That first report caused two days of argument. The client said the link wasn't working. I said the trace was clean. The debate burned time from network engineers, outside plant crews, and the project manager. Every hour made the final correction more expensive.
When we finally brought in a second EXFO unit, it found the problem in under an hour. We had to re-splice that segment. The cable was already installed, so this meant a bucket truck, a crew, and another full day. The total cost was about $3,700. The annual calibration fee for that module would have been a fraction of that amount.
That's when I understood that EXFO OTDR calibration isn't a maintenance expense. It's the cheapest insurance you can buy for a test report.
What EXFO Inc. Calibration Covers
After that job, I called EXFO Inc. in early 2024 and asked about their calibration process. The service team explained that each module is tested at specified wavelengths, checked against reference standards, and given an as-found measurement before any adjustment. If something is out of tolerance, they adjust or repair it, then verify it again. The standards are traceable, and the lab follows an ISO/IEC 17025 quality system.
That report is a point-in-time record. It proves the unit was within specification on that day, under those conditions. It doesn't make the instrument immune to drift for the next year. That's why a calibration label is only useful if someone actually checks it before the unit goes into the field.
I don't have hard data on how often an EXFO OTDR calibration catches drift across the whole industry. What I can say anecdotally from our own fleet is that we've sent in units we assumed were fine and received as-found reports showing one measurement outside tolerance. If that assumption had gone into a customer network instead of a service lab, it would have been a much more expensive conversation.
Other Things That Make a Good OTDR Look Bad
Before you blame calibration for every weird trace, check the basics. I once sent a module back for calibration and later found the real issue was a dusty launch cable.
- Launch cables and connectors. A dirty connector can mimic a dead zone, a splice, or an unusually high loss event.
- Index of refraction. If this setting is wrong, distance measurements shift. It won't change average loss much, but it makes the instrument look unreliable.
- Pulse width and averaging time. Too long a pulse on a short link can hide small events. Too short a pulse can create noisy traces that are hard to interpret.
- Warm-up. Cold equipment needs time to stabilize. Keep the unit in a cab, not in the truck bed, and let it warm up before you certify anything.
The Locked Phone Lesson Is the Same Lesson
A few weeks later, a field tech called about a locked Android tablet. The OTDR was fine, but the app wouldn't start. He asked me how to reset phone when locked. I gave the usual answer: hold the power button. That didn't work.
The reset combination was model-specific. By the time we found it, the morning's trace file had disappeared. That failure wasn't the OTDR. It was the process around it. No written procedure, no backup plan, no checklist.
The problem kept repeating because we treated the phone as an accessory. Calibration is the same trap. It's treated as a side task until the side task becomes the reason a report is wrong.
The Last Thing I'd Say to Another Fiber Tech
Check the calibration label before you need to defend a trace. If the date is overdue, the unit doesn't go in the truck. I'd rather explain why a test was delayed than explain why a test that passed was wrong.
I still have the trace file from that first OTDR. It looks like a clean pass. If I opened it today without knowing the history, I'd probably defend it. That's the real problem with measurement drift: it produces confident, believable, wrong answers. A C210 blood pressure monitor taught me the same lesson, only with lower stakes.
Searching how to reset phone when locked is easy. Explaining why a bad fiber got a clean report is not.
