It Starts With a Late-Night Call
I'm an emergency field engineer for a regional fiber contractor. Over the last four years I've handled more than 40 after-hours outage calls for carrier transport networks. I can only speak to my corner of the industry, but in that corner one habit keeps costing us time.
In March 2024, the NOC called me at 2:15 AM. An edge router in a metro hut was dark. The on-site technician had already checked the PDU with a voltage tester and reported 121.4 VAC, so he assumed the power supply was healthy. Then he went quiet. When I asked what the transceiver showed, he read a receive power of -21.7 dBm and a CRC error counter climbing in the log. That isn't a power-supply failure. That's a contaminated connector or a bad splice somewhere in the optical span. We lost 80 minutes that night because we anchored on the first test.
I should add that I'm not anti-multimeter. I carry one. It has a place. But the order matters.
The Power-Supply Reflex
When a router or an optical transport shelf goes dark, the first object in a technician's hand is often a multimeter or a voltage tester. Why? Because electrical checks feel concrete. A power-supply LED is binary, a VAC reading is a number, and the problem seems containable. But a network can be perfectly powered and still dead.
The more I see this, the more I think the reflex is a hangover from the copper era. On an older Ethernet run or a T1 circuit, most field faults really did end in power, cable, or interface. The failure domain was small. Optical transport changed the domain without changing our training habits.
How to Read a Multimeter Correctly
Since the term gets searched, let's cover it. If you genuinely need to check a power supply: set the meter to AC V for AC supplies and DC V for DC supplies. Plug the black lead into COM and the red lead into V. Test a known-good outlet first, because your meter is only as reliable as its leads and fuse. Then test the socket or the PSU output. A non-contact voltage tester tells you if a circuit is live; it does not tell you if the supply is stable or adequately loaded.
Notice what that sequence does: it proves electrons are moving. It says nothing about light.
The Invisible Failure: Why Fiber Outsmarts the Multimeter
Fiber faults don't look like copper faults. A broken electrical cable gives a hard open. A contaminated fiber connector can lose 0.5 dB one moment and 8 dB the next as temperature shifts. A micro-bend in a patch panel may not appear until the cabinet door is closed. No voltage tester on earth can see that.
The deeper reason we misdiagnose, in my opinion, is that we trust what we can see. Electricity gives visible cues: sparks, brownouts, a blown fuse. Light in a single-mode fiber is invisible. A -20 dBm signal doesn't make a dark patch cable glow. You need an instrument to translate it into numbers.
Why does that matter? Because your first tool should be the one that can falsify the most likely failure. On a fiber-fed site, the most likely failure is optical.
What a Misdiagnosis Really Costs
In my own call logs from 34 after-hours tickets between 2022 and 2024, 22 were eventually traced to optical-layer faults. Not power, not hardware. Of those 22, 14 had spent at least an hour on electrical troubleshooting first. Those hours don't show up on the invoice line item labeled power check because many emergency vendors bury them in a low-priced first dispatch.
The most frustrating part of those late-night calls is that everyone on the bridge has a theory. You'd think an alarm log would point clearly to layer one, but the log says link down for both an electrical failure and an optical failure. The interpretation falls on the human who arrived with a screwdriver.
There is also a pricing transparency issue. I've learned to ask what's not included before I ask the price. Some emergency quotes look cheap because they only include electrical verification and basic fault finding. The OTDR trace, the second truck roll, and the truck with the right test gear are separate fees. If the up-front quote had included optical testing, we would have started there.
The vendor who lists optical testing on the first line usually ends up cheaper than the one who saves it for the footer. If you ask me, that's a better way to buy.
A Better Sequence (Keep It Short)
Now the fix, because the problem is clear by this point.
- If the device is completely dark, do a quick electrical sanity check. Use a voltage tester to confirm the outlet or PSU. Then stop.
- Verify the optics. Look at the transceiver receive power and error counters. If the receive light is in the expected range but FEC or CRC errors are climbing, suspect contamination or a damaged jumper first.
- For total loss or intermittent degradation, grab an EXFO MaxTester OTDR. A single trace shows where the loss is: connector, splice, bend, break. It beats a night of guessing.
- If you don't need distance-to-event, use an EXFO laser source and an optical power meter to measure end-to-end loss. This is the optical equivalent of a continuity check, but with actual numbers.
I've watched a tech with an EXFO MaxTester get an alert that said event at 4.82 kilometers, reflective loss 2.1 dB. That put him at the right cabinet in about ten minutes. On the same route, a multimeter could have burned the entire maintenance window.
Let me rephrase that: the goal isn't to become an optical expert overnight. It's to let the right instrument ask the right question early.
The Takeaway
The next time a node dies at midnight, feel free to remember how to read a multimeter. Use it as a first filter, not as the whole investigation. A power supply can be healthy and the network can still be down, because the fiber is the long, fragile, invisible part of the path.
Transparency works the same way. Show the full diagnostic path in the quote, not just the cheap first hour. In my experience, honest pricing plus the right test sequence ends a maintenance window before sunrise rather than after it.
