Three Ways to Test a Diesel Generator’s Compression — and the One I’d Start With

The 400 kVA set looked perfect. Clean canopy, fresh paint, hour meter showing a believable 3,100 hours, and it cranked over like it meant it. The seller had it running when I arrived — smooth, no smoke, voltage steady. Any buyer in a hurry would have shaken hands right there.

I’d been hired to check it before money changed hands, so I did the thing almost nobody does: I pulled the injectors and ran a compression test. Four cylinders read strong and even. Two didn’t. They were down by nearly a third. The engine ran fine that afternoon because a diesel with two tired cylinders still runs fine — right up until it has to start on a cold morning or carry full load for twelve hours.

My client walked away. The seller called me several names, most of them about my mother. Six months later I heard the set went to another buyer, untested, for five thousand less than the asking price. Then it went to a rebuild shop. The rings were gone on those two cylinders. The “discount” cost him about four times what a compression test would have.

Here’s what I want you to take from that story: compression is the one number that tells you a diesel engine’s future, and it’s the number almost nobody measures until the future arrives. Not the controller. Not the hour meter. Not the oil analysis, useful as it is. Pressure in the cylinder, held where it belongs, is the whole game. Lose it and everything downstream — starting, power, fuel burn, smoke — degrades together.

There are three honest ways to measure it. They differ wildly in what they cost you in labor, what they tell you, and when each one is the right tool. I’ve used all three for years. This is the comparison I wish somebody had given me in year one.

The gauge test: the real number, at the cost of an afternoon

This is the absolute compression test, and it’s the gold standard for a reason: it gives you an actual pressure reading per cylinder. You remove the injectors — or the glow plugs, on engines that have them — thread an adapter into each hole, disable the fuel so the engine can’t fire, and crank it while the gauge records the peak.

What you’re looking for isn’t really the absolute number. New-engine specs vary by design, and a number from a manual written for a fresh engine won’t match a set with eight thousand hours on it. What matters is uniformity. All cylinders should read within about ten percent of each other. One cylinder reading twenty or thirty percent below its neighbors is a problem with a name and an address, even if you don’t know the name yet.

The discipline matters more than the gauge. Battery fully charged — a weak battery cranks slow and every reading comes out low, and you’ll chase an engine problem that’s actually a charger problem. Engine warm if you can manage it, cold readings run lower. Same number of compression strokes per cylinder; I use five. And disable the fuel properly, not by “just cracking the lines.” An engine that fires on one cylinder during the test gives you a reading that’s fiction.

The wet test is the gauge test’s built-in second opinion. If one cylinder reads low, squirt a spoonful of engine oil into that hole and test again. If the number jumps up, the rings weren’t sealing and the oil temporarily sealed them — ring wear. If the number barely moves, the leak is past the valves or the head gasket, where oil can’t reach. Two minutes, no extra tools, and it splits your suspects in half. I do it every time a cylinder reads low. Every time.

The honest downside: it’s invasive. Pulling injectors on a modern engine means disturbing high-pressure fuel lines, copper sealing washers that should be replaced, and on some designs, injector bores that don’t forgive ham-fisted adapters. On a V-engine in a tight canopy, plan half a day. That’s why this test lives at the top of my used-generator inspection list — when you’re about to spend real money, half a day is cheap — and further down the list for a running set with a vague complaint.

Relative compression: the shortcut that points at the guilty cylinder

Now the clever one. A starter motor works harder to push a piston up against high compression than against low compression. Clamp a current probe around the starter cable, crank the engine with fuel disabled, and watch the current trace: each cylinder’s compression stroke shows up as a hump. Even humps, even compression. One short hump, one weak cylinder.

What you get is comparison, not numbers. It won’t tell you a cylinder reads 380 psi — but it’ll tell you cylinder four is the odd one out, plain as day on the trace. For diagnostics, that’s usually enough to act on — and you got it without removing a single injector.

I reach for this test first on a running set with a symptom: hard starting that’s already survived the battery-fuel-air checklist, a slight miss at idle, a new puff of smoke. Fifteen minutes, one clamp, no disassembly. A quarry’s 250 kVA set came to me with a rough idle everyone had blamed on fuel quality. The relative trace showed one short hump. We opened that cylinder’s valve cover first — and found the exhaust lash nearly gone, the valve barely seating. Adjusted, re-tested, humps even. Total time under an hour, and we never touched an injector.

Its limits are honest ones. A weak battery flattens all the humps together, so the charger goes on first, same as the gauge test. It can’t separate rings from valves — that’s a job for the next test. And on a big slow-turning engine the humps get mushy; the method shines on the mid-size sets most of us actually service. Think of it as triage. Fast, cheap, and it tells you where to spend the expensive hours.

The leak-down test: the one that names the culprit

The third test doesn’t crank the engine at all. You bring one cylinder to top dead center on the compression stroke — both valves closed — lock the crankshaft so it can’t turn, and feed regulated shop air into the cylinder through the injector or glow-plug hole. A two-gauge tester shows what percentage of the air leaks out. Then you listen.

The listening is the whole test. Hiss at the oil fill cap: air blowing past the rings into the crankcase. Hiss at the exhaust pipe: the exhaust valve isn’t seating. Hiss at the intake: the intake valve. Bubbles in the coolant or pressure climbing in the cooling system: head gasket, or worse, a crack. Each sound is a diagnosis with the repair attached. No other test hands you that.

A hospital’s 500 kVA set taught me to respect this one. It was losing a little coolant with no external leak, and the oil looked fine — no milkshake, no drama. The gauge test showed all cylinders within spec, which nearly closed the case. But the leak-down on cylinder three hissed straight into the cooling system. The gasket was seeping combustion pressure into the coolant jacket under load, not enough to fail a cranking test, enough to push coolant out the overflow on a long run. We caught it during a planned outage. Had it let go during a real one, that hospital would have learned about head gasket failures the hard way.

The price of this precision is equipment and patience: a leak-down tester, a compressor, and the discipline to get each cylinder exactly to TDC — off by a few degrees and the valves aren’t both closed, and your “leak” is just an open valve lying to you. I use it when the gauge or relative test has already found the weak cylinder and I need to know what to order before opening the engine. Rings, valves, or gasket — three very different parts lists, three very different labor quotes.

So which one do you actually use?

An old hand I work with — thirty years on engines, and he’ll remind you — says compression testing is for car mechanics, and that on a generator, if it cranks and runs, it’s fine. He’s wrong, and the 400 kVA set at the top of this article is why. “Cranks and runs” is a statement about today. Compression is a statement about next year.

My working order, after twelve years of getting this wrong in every possible sequence: relative compression first for any running engine with a symptom — it’s fifteen minutes and it aims the rest of the work. The gauge test when money is about to change hands, or when the relative trace names a cylinder and I want numbers before I quote a repair. Leak-down when I need the culprit named before parts get ordered.

And one rule that covers all three: test before you blame the expensive parts. I’ve watched people condemn a turbocharger, then injectors, then the injection pump, chasing smoke and hard starting that a compression test would have explained in an hour. The blowby coming out of the breather was trying to tell the same story — pressure going where it shouldn’t — but blowby tells you there’s a leak, and these tests tell you where.

Put compression testing on the maintenance plan the way you’d put anything else that predicts the future: not every month, but on a schedule — at purchase, after any overheating event, whenever consumption or starting changes character. The gauge doesn’t care about your opinion of the engine. It just reads the pressure. That honesty is exactly why I trust it more than my own ears. If your set is due for a real health check — or you’re about to buy one — talk to us before the engine gives you the expensive version of this lesson.

That seller with the 400 kVA set? Last I heard, he’s still telling people compression tests are a scam buyers use to haggle. Maybe. Or maybe the scam was selling an engine with two dead cylinders to someone who didn’t check. The gauge doesn’t haggle. It just tells the truth, one cylinder at a time.