My phone went off at 2:14 on a Tuesday morning. Logistics warehouse outside town, utility had blipped for maybe ten seconds, and their 500 kVA standby set was doing the saddest thing a generator can do: cranking strong, cranking healthy, cranking like it meant it — and never firing. Not a cough. Not a puff of smoke. Just that endless, confident cranking sound that tells you the starter motor is doing its job and something else has completely checked out.
The maintenance guy met me at the gate looking wrecked. “Batteries are six months old,” he said. “Tank’s three-quarters full. The other company came yesterday afternoon, changed both fuel filters, bled the system, charged me for a service call, and told me to call them back if it happens again.” It happened again. That’s why I was there in a hi-vis vest at 2 a.m., drinking gas-station coffee.
Here’s the thing I’ve learned in twelve years of chasing generators that won’t start: when an engine cranks at full speed and never even tries to fire, stop touching the fuel system. The fuel system is probably innocent. The question you should be asking is much dumber and much cheaper: does the controller even know the engine is turning?
The magnetic pickup — MPU, speed sensor, whatever your controller manual calls it — is a small threaded cylinder screwed into the flywheel housing, its tip hovering a hair’s breadth from the ring-gear teeth. Every time a tooth passes, it disturbs the pickup’s magnetic field and the sensor spits out a little pulse of AC voltage. Faster teeth, higher frequency. The controller counts those pulses and that’s how it knows the RPM.
That’s it. That’s the whole job. And yet this ten-dollar-looking part has more authority over your generator than almost anything else on the skid. No speed signal, no fuel — the controller is designed that way on purpose, as a fail-safe. It will not open the fuel solenoid for an engine it believes is standing still, because injecting fuel into a stationary engine is how you get a crankcase full of diesel and a very bad day. So when the pickup goes quiet, the engine cranks forever on a controller that calmly insists nothing is rotating. Which, from the controller’s point of view, is true.
I used to walk right past this sensor on PM visits. It’s small, it has no moving parts, it doesn’t leak or drip or make noise. Twelve years later it’s one of the first things I check, because I’ve watched it cause more expensive misdiagnoses than any other part I can name. There’s something almost funny about a forty-thousand-dollar machine being held hostage by a part that costs less than the service call to look at it. Almost.
A pickup doesn’t usually die cleanly. It lies, and it lies in three distinct ways, and each lie sends you chasing a different expensive ghost.
Lie number one is silence. The tip is magnetic — it has to be, that’s the physics — which means it collects every fine iron filing floating around the flywheel housing. After a starter replacement or ring-gear work, there’s always debris. Give it a few months and the tip looks like it’s wearing a fur coat of metal dust. The magnetic field gets smothered, the pulses get weak, and one cold morning the signal drops below what the controller can read. Crank, crank, crank, nothing. That warehouse at 2 a.m.? I pulled the pickup and the tip was caked solid. The “other company” had changed fuel filters on an engine that was never getting fuel in the first place, because the controller never knew it was turning. Wiped the tip clean with a rag, reinstalled it, and the set fired on the second crank. The maintenance guy just stared at me. I didn’t have the heart to tell him what the filter change cost.
Lie number two is stuttering. The wiring from the pickup to the controller is usually a thin shielded pair, and it runs through exactly the kind of places wires go to die: zip-tied to vibrating brackets, draped over sharp sheet-metal edges, baked near the exhaust. When the insulation chafes through and the signal wire kisses ground intermittently, the controller sees a speed signal that jumps and doubles. And what does a controller do when it thinks the engine just doubled its RPM? It trips on overspeed and shuts the whole thing down. I watched a data center chase this ghost for three weeks. First company replaced the AVR. Second company quoted a new controller — four figures, two-week lead time. The actual problem was a harness rubbing on a bracket edge, worn through to copper. Fifteen minutes with a multimeter, a piece of split loom, and a reroute. The controller was never the problem. The controller is almost never the problem, and yet it’s always the first thing people want to replace, because it’s the expensive box with the screen and replacing it feels like doing something.
Lie number three is the one the old-timers don’t believe. A pickup that works fine cold and dies hot. Heat changes the coil’s resistance; a marginal pickup that reads fine on a cool morning drops out after twenty minutes of running, then “heals” when everything cools down. My old mentor — thirty years in, hands like sandpaper — insists sensors either work or they don’t. “It’s a coil of wire, kid. It doesn’t have moods.” Respectfully, it does. I had a telecom site that only failed on summer afternoons, always the same fault, always fine by the time anyone arrived with tools. Three visits, no fault found, everyone embarrassed. Fourth time I drove out at 1 p.m. in August, left the set running, and watched the speed signal on a meter decay as the housing soaked heat. New pickup, problem gone, and I still bring it up every time he says sensors don’t have moods. He hates that.
There’s a pattern to these misdiagnoses and it’s worth naming, because it costs real money. The controller gets blamed because it’s the brain and brains are suspicious. The AVR gets blamed for hunting and surging that’s actually a noisy speed signal feeding garbage to the governor — if your engine’s frequency won’t sit still, read what the governor is actually reacting to before you touch the voltage regulator. Batteries get replaced because “it won’t start” must be the batteries, even though the batteries are audibly cranking the engine like a champ — strong cranking with zero firing is the single best clue that your problem is signal, not power. And fuel filters get changed, bless them, because changing filters is what you do when a diesel won’t start, except nobody asked whether the controller had even authorized fuel yet.
My rule is simple now, and I wish someone had tattooed it on my forearm in year two: cranks strong but never fires means the controller doesn’t see rotation. Check the eyes before you operate on the stomach. For the full crank-no-start decision tree, start here — but if the cranking sounds healthy, skip straight to the pickup.
This is the whole diagnostic, and it fits in one short list because there genuinely isn’t more to it:
One more trick, free of charge: a lot of engines carry two pickups — one for the governor, one for overspeed protection. If you suspect one, swap them. If the fault moves, you’ve found it. If it doesn’t, you’ve ruled out the cheapest suspect in sixty seconds, and that’s worth knowing too.
The gap deserves its own sermon, because it’s where good mechanics create bad problems. Too close and the gear teeth will machine the tip down over time — I’ve pulled pickups with the face ground to a shiny nub, the coil exposed, after someone “snugged it up” during an overhaul and never backed it off. Too far and you get the cruelest failure of all: the signal is strong enough at rated speed but drops out at slow cranking speed, so the set runs fine for months and then won’t start on the one morning it matters. The engine hasn’t changed. The physics of the gap just finally caught up with it.
After any engine work near the flywheel housing — starter replacement, rear seal, overhaul — the pickup gap is guilty until proven innocent. It takes ninety seconds with a wrench. I check it the way pilots check fuel: every time, no matter who touched it last, no matter how sure they are. The mechanic who set one “by feel” on a 1 MW set I commissioned was a good mechanic. He was also wrong by two full turns, and we found out at 6 a.m. during the witness test, in front of the customer. Nobody enjoyed that morning.
Here’s my observation after a dozen years, for whatever it’s worth: the parts that fail dramatically get all the respect — turbos, injectors, controllers with their glowing screens. The magnetic pickup fails quietly, or rather it fails quietly adjacent: everything around it looks fine, the engine is mechanically perfect, the fuel is clean, the batteries are strong, and the machine just sits there cranking into the void because one dumb sensor can’t see the teeth going by.
Phantom overspeed trips that clear themselves? That’s usually the harness, and false trips have a whole family of causes worth knowing. A set that cranks like a champion and never fires? Before you spend a dollar, ask the cheapest question first: does the controller know the engine is turning? And if you ever do get a genuine overspeed event — real, mechanical, terrifying — know what the protection system is supposed to do, because that’s the day you want every sensor telling the truth.
That warehouse set has run fine for two years now. Every PM, somebody pulls the pickup, wipes the tip, checks the gap, and writes it in the log. Ninety seconds. The maintenance guy told me last month he shows the new hires the rag — the gray, metallic-smelling rag — and says, “this is what a three-thousand-dollar service call looks like before it happens.” He’s not wrong. If your standby set is due for a look, talk to us — we’d rather find it with a rag than with a 2 a.m. phone call.
Email：[email protected]
Work：Monday to Friday,9:30-18:30
