The Call That Started It All
I got the call on a Tuesday afternoon in March 2024. Our lead field tech, Mario, was on site at a major fiber deployment—think 50,000-unit annual rollout for a regional ISP. He was using our EXFO MaxTester 720C, an OTDR we'd been relying on for about 18 months. The unit had passed its last annual calibration check in September 2023. So when Mario said the trace looked 'off,' I'll admit my first reaction was skepticism.
'The trace shows a 0.35 dB loss at the first splice. Our reference was 0.15 dB,' he said. That's a 0.2 dB discrepancy. On a short-haul link running DWDM, that's enough to trigger alarms downstream. I told him to run it three more times with fresh launch cables. He did. Same result. I said, 'Extend the pulse width and try a different wavelength.' He did that too. The numbers still didn't match the reference values from the job's specification document.
That's when I had a sinking feeling. Not about the fiber. About our EXFO.
The Process Gap We Didn't Know We Had
We had a formal calibration schedule for all our network testers—every 12 months, sent back to EXFO's facility. That's standard. But what we didn't have, and I realized this mid-conversation with Mario, was a verification protocol between calibrations. We assumed that if the device passed its annual calibration, it was good for the next 365 days. That's how everyone does it (ugh).
I said, 'Swap to the backup unit and continue the job. I'll handle this.' Mario swapped to our other EXFO OTDR—same model, same firmware version. The trace suddenly looked normal: 0.16 dB on that same splice. The first unit was definitely reading high. The discrepancy wasn't in the field. It was in the instrument.
Now I had a problem. The fiber we'd already certified with the first unit—about 12 kilometers of it—might have been incorrectly flagged as marginal. Not failing, but not within spec. We were lucky in one way: the discrepancy was small enough that it wouldn't cause immediate service issues. But for a quality/compliance manager (this was back in my quality inspector role), 'close enough' is the enemy of good work. Especially when the contract specified loss thresholds with zero tolerance.
The Three-Day Investigation
I pulled the first unit from the field and ran a side-by-side comparison test in our office. Same launch cable, same fiber spool, same settings. The difference was consistent: 0.18–0.22 dB higher on every measurement compared to the backup unit. I checked the event dead zone and attenuation dead zone specs against EXFO's published data. The first unit was still within absolute tolerance per the manufacturer's spec sheet—barely. But it was outside our internal quality requirements.
To be fair, EXFO's calibration team was responsive. I emailed them the trace files and serial number. They asked for a log of the unit's last 100 tests. I sent it. They confirmed the internal laser source had drifted slightly, likely due to thermal cycling in the field. (The unit had been used in 14 different states over 18 months—temp swings from -10°C to 45°C). EXFO offered a rush calibration at no charge—warranty covered it. Turnaround was 5 business days, plus shipping.
That was the good news. The bad news: we had 8 other units in the field, all with similar usage patterns. If one had drifted, others might have too.
The Re-Calibration Escalation
We paid for rush calibration on all 9 units simultaneously. That was an expense I hadn't budgeted for—roughly $4,500 total including the expedited handling fees and return shipping. But the alternative was risking a $22,000 redo of a deployment we'd already certified for a client. Or worse, a service outage blamed on our test equipment. (Thankfully, my manager approved the expense without much pushback. We'd just had a similar incident with a different vendor's spectrum analyzer the quarter before. Lessons are learned the hard way, I guess.)
While the units were out, we implemented a monthly verification protocol: a short 3-minute test on a known reference fiber spool before any field use. If readings deviated by more than 0.05 dB from baseline, the unit was flagged for immediate recalibration. This added maybe 15 minutes per week per tech. But after the incident, nobody complained. In fact, one senior tech told me, 'Wish we'd done this years ago.'
What I Learned About EXFO Testing (and Trusting Your Tools)
So here's the thing I want to pass on. Not about EXFO specifically—their equipment is generally excellent. But about any network tester, whether it's an OTDR, a spectrum analyzer, or even how to read a multimeter reliably in a troubleshooting context. Calibration is a point-in-time event. Verification is an ongoing practice. They're not the same thing.
In that March 2024 experience, the cost of our oversight was measurable: about $1,200 in rush shipping fees we couldn't recover, plus the time I spent on the investigation. The upside: our field failure rate on that project dropped to zero for the following quarter. When I presented this at our Q2 quality review, I showed a slide comparing the $4,500 we spent on recalibrations against the estimated $45,000–60,000 in potential rework and reputational damage if we'd shipped faulty readings to the ISP client.
Now, does this mean every field tech needs to run a reference test every morning? Not necessarily. But if your work involves certifying fiber links for revenue-generating services—especially if you're using phones or tablets to log results, or managing a fleet of testers across multiple crews—a lightweight verification process is cheap insurance.
I get why teams skip it. It takes time. It feels paranoid. The equipment is expensive and you assume it's right. But uncertainty is expensive too. (I have a post-it on my monitor that says: 'An uncertain cheap fix is more expensive than a certain, pricier one.') The time certainty premium applies to test equipment decisions as much as it does to rush delivery choices. The guarantee of knowing your readings are valid—that's worth paying for.
Granted, this requires more upfront discipline. But it saves time later. When a network test shows a splice loss that's borderline, you want to be arguing about the fiber, not about whether your OTDR needs calibration.
Prices as of March 2024; verify current EXFO calibration rates.
