Why is my splice loss suddenly so high
Number one question, every single time. Yesterday the machine threw 0.02 dB splices and today everything reads 0.1 dB or worse. Your fiber didn't change, so the machine did.
Most of the time it's one of three things, and they're all cheap to fix: dirty V-grooves, worn electrodes, or a bad cleave. That's the order I check them in.
Start with the cleaver, not the splicer. Sounds backwards, I know. But a worn cleaver blade or a chipped end face gives you a high cleave angle, and the splicer will happily fuse a bad cut into a bad splice and report loss for it. So pull the fiber after cleaving and look at the end face under the splicer's own camera. Flat and clean, you're good. Angled or ragged, rotate the cleaver blade to a fresh spot or replace it. A decent blade gives you around 48,000 cleaves across all its positions, though field abuse eats that fast.
Then clean the V-grooves and clamps. Dust here causes more high-loss splices than anything else I see. One speck holds the fiber slightly off center and your core alignment is gone. Cotton swab, 99% isopropyl, gentle hand. For stubborn dust, drag a cleaved scrap of fiber through the groove to lift it. If you splice every day, clean these every day. I mean it.
Still high after all that? Your electrodes are probably finished.
When do I replace the electrodes
Depends on the machine, but the numbers are worth knowing. A Fujikura 90R wants fresh electrodes around 1,500 discharges. A lot of older single fiber units run to about 6,000. Ribbon splicers chew through them faster, sometimes 1,500. Sumitomo single fiber units sit near 6,000. INNO and the budget field units, check the manual, but plan on somewhere between 1,000 and 3,000.
The machine warns you, just not in words. The arc starts wandering side to side. The discharge looks too bright or too dim. You see "Too Left" or "Too Right" on a Fujikura, or the splice position drifts between identical splices. Worn electrodes, every time.
What an arc calibration actually does
People run it because the manual says to, with no idea why. So here's why.
The fusion arc is sensitive to altitude, temperature, humidity, and how worn your electrodes are. The arc power that gave you a perfect splice in an air-conditioned office at sea level will under-fuse or over-fuse on a mountain in July. Arc calibration reads the current conditions and trims the arc power so your fusion energy stays where it should be.
Run it when you reach a new site, when the weather shifts, right after you change electrodes, and any morning the first few splices look off. On a Fujikura or AFL unit you want "Good" on both arc power and position. If it says "Not Adequate," run it again. With fresh electrodes it sometimes takes two or three passes to settle.
Error messages, and what they really mean
This is where people get stuck, because the message on screen almost never names the real cause. Here's the bench translation.
"Fiber not found" / left or right fiber won't detect
The fiber's too far back in the holder, or the V-groove and lenses are dirty, or in the bad case the camera or optical block has failed. Reseat the fiber. Clean the groove and the objective lenses. If it still can't see fiber that's plainly sitting there, now it's an optical problem and it needs the bench.
"X/Y Motor Overrun" / Z motor overrun
The fiber's outside the camera's range, nearly always because it isn't seated at the bottom of the V-groove. Clean, reseat. If it keeps happening on a clean groove, the motor's worn or it lost its home position, and that's a repair.
"Large cleave angle" / "Cleave shape NG"
That's your cleaver, not your splicer. Clean the cleaver, re-prep the fiber, rotate to a fresh blade position. Dirty objective lenses can fake this one too, so wipe them if a clean cleave doesn't clear it.
"Strong arc" / "Weak arc power"
The auto mode can't pull the arc into range. Run an arc calibration on standard G.652 single mode. If it won't reach "Good," your electrodes are worn or there's a fault on the high voltage side.
Bubbles in the splice / splice too thin or too fat
Arc power or contamination. Clean everything, recalibrate, recleave. Bubbles usually mean dirt or a bad cleave. Too thin means weak arc or not enough fiber overlap. Too fat means too much.
Won't boot / freezes on a screen / comes up dead
If you see the boot or motherboard info appear and then it locks up, that's often firmware or a board fault. If nothing comes up at all, suspect the power board, the battery, or the main board. Those are bench repairs, not field fixes, but they're very fixable. Don't scrap a splicer over a no-boot.
Which fiber goes with which splicer
This trips up more people than the error messages do. Buy the wrong type for your work and you either overpay or you can't hit your loss numbers. Quick rundown.
Core alignment splicers
Image and line up the actual light-carrying cores. Lowest, most consistent loss, often under 0.03 dB, and they handle single mode, multimode, dispersion shifted, NZ-DSF, bend insensitive G.657, and splices between different fiber types. Telecom backbone, long haul, FTTx headend, anything where loss is non-negotiable, this is the class you want. Fujikura 70S, 90S, 90R, Sumitomo Q101, INNO View — that tier.
Cladding alignment splicers
Line up the outer glass and trust that the core follows. Cheaper, faster, smaller. Fine for standard single mode in less critical work, and genuinely good on multimode, where the big core forgives a lot. Last mile FTTH, LANs, premise cabling, high volume short runs. The catch is specialty fiber or older fiber with concentricity error, where the core doesn't follow the cladding and your loss climbs.
The rule I give people: multimode and routine drop work, cladding alignment saves you money. Single mode backbone, specialty fiber, or any job where someone's going to OTDR your splices and complain — core alignment, don't argue with it.
Ribbon / mass fusion splicers
Fuse up to 12 fibers at once, which is the only sane way to handle data center and high density backbone work. They cost more and run cladding alignment, but terminating thousands of fibers, the time saved pays for the machine.
PM fiber
Polarization maintaining fiber — the Panda and bow-tie types — needs a splicer that rotationally aligns the stress rods, usually to 45 or 90 degrees. A Fujikura FSM-40PM, FSM-45PM, or FSM-100P, or a dedicated PM unit. A standard splicer will fuse PM fiber, but it won't hold the axis and your extinction ratio falls apart. If PM is your work you already know this. If someone hands you PM fiber and a standard machine, stop and get the right one.
Repair it or replace it
Depends entirely on what's wrong. A no-boot, a dead screen, a worn motor, a failed camera, board level faults — those are repairs, and they usually run a fraction of a new machine. A four or five figure splicer with a dead battery board is not scrap. We fix those all the time.
What tips a machine toward replacement is when the repair cost and the age and a dropped OEM support line all land at once. An older unit the manufacturer won't service anymore, with several failing subsystems, sometimes that money is better spent on a current model. But get it looked at first. I've watched people throw out a splicer that needed a $200 part.
Not sure? Send it for an evaluation before you decide. A good shop will tell you honestly what it needs and what it costs before you commit, and you make the call from there.
Buying in volume
Outfitting a crew, or putting through a procurement order? A few things save real money and grief.
Buy the splicer, cleaver, and stripper as a matched set. Mismatched tools are where loss problems and slow techs come from. The cleaver especially has to be good, because a cheap cleaver wrecks a good splicer's results. I'll keep saying it until it sticks.
Standardize the fleet on one or two models so electrodes, blades, holders, and spares interchange and your techs aren't relearning a menu on every job. A truck carrying five different splicers is a maintenance headache nobody needs.
Stock consumables ahead of the work. Electrodes, cleaver blades, sleeves. Running out of electrodes mid-job because nobody ordered spares is the most avoidable downtime there is.
Put calibration in the budget from day one. Every splicer drifts, and a fleet that never gets recalibrated will quietly start failing splices across the board, usually right when you're busiest.
For bulk buys, used and refurbished units deserve a serious look. A properly serviced, recalibrated previous generation core alignment splicer will outperform a brand new bargain cladding unit, for less money. That's a lot of what we do at Aumictech Labs, so take my bias into account, but the math holds, especially across ten machines instead of one.
If you only remember a few things
Most splicer trouble is dirt, electrodes, or a bad cleave — roughly in that order. Match the machine to your fiber: core alignment for single mode and specialty, cladding for multimode and volume, ribbon for density, PM for PM. Calibrate your arc. And don't bin a splicer that only needs a board or a battery.
Got a splicer doing something I described here and want a second opinion? That's what we're here for. Send us the model and what it's doing, and we'll tell you straight whether it's a five minute fix or a bench job.