I test a lot of multimode fiber. Most of it gets tested because someone is two days from a cutover—a data center handoff, a 5G x-haul link, or a campus backbone—and suddenly realizes the link hasn't been certified. In my role coordinating EXFO testing for enterprise and network teams, I've built a six-step checklist that works even under a deadline. Whether it's a one-rack startup or a nationwide carrier, I use the same steps. Small doesn't mean unimportant; it means potential.
When to use this checklist
Use this when you need to verify a multimode link with an EXFO Multimode OTDR before a data center move, a 5G node deployment, or a network acceptance test. It's also the checklist I follow when something doesn't look right: high attenuation, unexpected loss, or an event that disappears on the next run. This is a field sequence, not a substitute for your company's test plan.
The six-step EXFO multimode OTDR checklist
1. Define the goal before touching the OTDR
What are you proving? For acceptance testing, you need pass/fail against a spec. For troubleshooting, you need the location and severity of each event. These require different settings. I know it's tempting to switch on the OTDR and hit Auto, but that's how you end up with a trace and no answer. Take 30 seconds to confirm fiber type, link length, wavelengths, and launch conditions.
2. What is a connector? Inspect the connector and the jack
In fiber optics, a connector is the plug that terminates a fiber. The jack is the socket side—the adapter or receptacle mounted on a patch panel or wall plate. People call them both connectors, and in normal conversation that's fine. In testing, the distinction matters because a dirty jack can contaminate a clean connector in one insertion.
Before any test, inspect every end face with a scope. Per TIA-568.3-D (as of January 2025), connector end faces should be inspected before mating. That includes the connector on your EXFO launch cord, the connector that goes into the jack, and the jack itself. I'm not a standards expert, so I can't quote every clause. What I can tell you from the field is that contamination is the reason I keep getting called out for re-tests.
I don't have hard data on industry-wide contamination rates, but based on our last 200 rush fiber jobs, my sense is roughly one in four traces is affected by a dirty end face. Cleaning the connector and the jack before running the OTDR saves an hour every time.
3. Set the OTDR for multimode, not singlemode
A normal EXFO Multimode OTDR measures at 850 nm and 1300 nm. If the unit is set to 1310/1550, you're testing the wrong wavelengths. Set the index of refraction (IOR) for your fiber type—50/125 or 62.5/125—and choose a pulse width matched to the link length. For short links, keep the pulse short. Let the acquisition time compensate; 10–30 seconds is usually enough.
Auto mode gets you in the ballpark, but I have mixed feelings about it. On one hand, it's fast. On the other, auto settings are often optimized for longer singlemode routes, so they can miss short multimode events. I'd rather set the range manually and let the OTDR do the measurement.
4. Use a launch cable and receive cable
This is the step most people ignore. The OTDR has a dead zone after its test port. To measure the first connector, you need a launch cable—a known-good jumper—between the OTDR and the fiber under test. If you're certifying a wall jack, plug the launch cable into the jack so the OTDR can see the connector at that jack. Missing this step means you lose the first event, and the first event is often the bad one.
For short multimode links, also add a receive cable after the far end. It gives you the last connector's loss and a cleaner end of trace. I've seen technicians skip the receive cable and then call me about a high loss that was actually caused by a kinked launch cable. The cable matters.
5. Run the trace and read it like a technician
Now you're doing real EXFO testing. After the acquisition, zoom in on each event. Reflective spikes are connectors, mechanical splices, or bad mating. Non-reflective dips are fusion splices, macrobends, or stress points. Don't assume a long straight trace is good; zoom into the first 50 meters, where multimode problems hide.
One important reality: the OTDR gives you per-event loss and reflectance, but total insertion loss should be confirmed with a light source and power meter. This gets into OLTS territory, which is not the OTDR's strength. The OTDR is for locating and quantifying events; the power meter gives you the end-to-end loss a transceiver will actually see.
6. Name the file so your client can read it tomorrow
I can't tell you how many retests happen because someone saved the trace as TEST1. A good filename looks like this: site_rack_fiber_direction_date. For example, HQ_rack03_A_to_B_850nm_2025-01-15. Add the technician name, OTDR serial number, wavelengths, IOR, and a note if any connector was re-cleaned.
I wish I had tracked how many rush orders were delayed by bad file names. What I can say anecdotally is that it's a lot. A report that can't be matched to the physical link is useless, even if the trace is perfect.
Common mistakes to avoid
Dirty connectors are the problem
I said it in step 2, and it deserves repeating. If a trace looks weird, stop the test and clean every connector and jack involved. You'd think written procedures would prevent this, but the pressure of a deadline makes people skip the scope. It's the wrong thing to skip.
The "multimode is easier" thinking
This was true years ago, when multimode links were short and speeds were low. Today, with 100G links and 5G x-haul, the margin is much tighter. The old belief that "it's only multimode" causes people to ignore launch conditions and connector inspection. That's changed.
Forgetting the 5G path
If your team is using the EXFO G310 5G solution, test the fiber that feeds it. I've seen 5G throughput problems blamed on RF when the real issue was an LC connector at the base station cabinet. A dirty connector in the fronthaul path can look like a radio issue until someone finally runs an OTDR.
One more thing: don't panic when a deadline is tight. I've handled 47 rush requests in a single quarter, and 95% of them were completed on time. The ones that failed were the ones where someone skipped the inspection step to save five minutes. A clean connector is faster than a retest.
