Compression testing: doing it properly, and reading what it tells you

Every other test on a car asks whether a component is working. A compression test asks something more fundamental and more useful: is the cylinder still a sealed box?

It is worth doing early rather than late, because a mechanical fault will happily absorb an unlimited budget of ignition and fuel parts before anyone thinks to check. The corpus puts it plainly on low engine compression — a misfire that no amount of ignition or fuel work resolves is where compression should have been measured first.

The original of this article is short and its diagnostic logic is sound. What it leaves out is most of the preparation, and the preparation is where the numbers come from.

Getting a number that means anything

Six things have to be true before a reading is worth writing down. The original mentions two of them.

Battery fully charged, and cranking healthy. Compression is measured by how hard the engine can be spun, so a tired battery or a dragging starter produces uniformly low figures across every cylinder — which reads exactly like worn rings, and is the commonest way this test lies. If in doubt, check the battery and the starter circuit first; the lead-acid battery covers what a good one looks like and using a multimeter covers the voltage-drop test that finds a starter circuit losing its supply.

Engine at operating temperature, if you can manage it. Cold metal has not expanded into its clearances, so a cold engine reads low. A cold test is still useful for comparing cylinders; it is not useful for comparing against the manufacturer's figure.

All the spark plugs out. Not one at a time. Leave the others in and the engine is fighting their compression, so it cranks slowly and every reading is depressed.

Throttle held wide open. The original says this and it is essential. A closed throttle starves the cylinder of air on the intake stroke, and every cylinder then reads low and, worse, evenly low — a convincing wrong answer.

Fuel disabled. Cranking with the injectors live sprays petrol into open cylinders, out of the plug holes, down the bores where it washes the oil film off, and into the sump where it dilutes the oil. Pull the fuel pump relay or the injector fuses.

Ignition disabled — and this one is not just about fuel. With the plugs out and the coils still connected, every coil fires into an open circuit on every revolution. That is the same condition described in ignition leads: the secondary has nowhere to discharge, so the voltage climbs until it breaks down the coil's own insulation or backs through the driver stage. A compression test is the third place in this archive where that hazard turns up, and it is the one nobody warns about, because the plugs are out and the danger is not obvious. Unplug the coils, or pull their fuse.

Then crank each cylinder for four to six compression strokes — until the needle stops climbing — and reset the gauge between cylinders.

Reading the result

Look at the spread before the absolute value. The manufacturer's minimum matters, but the number that identifies a fault is the difference between cylinders, and the widely used limit is 10 per cent. Four cylinders all reading a little low together is a tired engine, or a cold one, or a slow crank. One cylinder 20 per cent below its neighbours is a fault with an address.

The original's interpretation is correct and worth keeping verbatim in substance:

One addition the original does not make, and it costs nothing: watch how the needle climbs, not only where it stops. A healthy cylinder jumps most of the way on the first compression stroke and then creeps up. A cylinder that starts strong and builds slowly is leaking past the rings. A cylinder that is low on the first stroke and stays low is losing it through a valve. That distinction is free and you have already done the work to see it.

And a silent cause worth remembering here: a valve clearance adjusted too tight holds a valve fractionally off its seat once the engine is warm, which produces low compression on one cylinder with no noise whatsoever. Valve clearance covers why that happens and why the drift goes in that direction.

The wet test

This is the original's best content. Squirt a small amount of engine oil into the low cylinder through the plug hole, crank a couple of turns to spread it, and repeat the measurement. The oil temporarily seals the rings against the bore.

• The reading jumps substantially — say 40 psi / 3 bar or more — and the leak was going past the rings.
• The reading barely moves — a few psi — and the leak is somewhere the oil cannot reach: a valve, a head gasket, or a cracked head.

Use a teaspoon or so. Too much oil does not seal better; it fills the combustion chamber, reduces the volume the piston has to compress into, and gives a falsely high reading on any cylinder at all.

The leakdown test, which answers the question compression cannot

A compression test tells you a cylinder is leaking. A leakdown test tells you where the leak is going, which is the difference between knowing there is a problem and knowing what the repair is. The two tests belong together and this is why the piece on cylinder leak testing is folded in here.

The method: bring the cylinder to top dead centre on its compression stroke with both valves shut, put compressed air in through the plug hole at a regulated pressure, and read what percentage escapes. Then — and this is the whole point — listen for where it goes.

Two practical notes. The engine will try to turn as soon as air goes in — the piston is being pushed down by air on a crank that is free to rotate — so it must be genuinely at top dead centre and the car should be in gear with the handbrake on, or the crank pulley held. And the percentage figure matters less than the location: 20 per cent leaking past a valve is a cylinder head job whichever way you write the number, while the same figure heard at the dipstick on a high-mileage engine may simply be its age.

The version that needs no tools at all

The original mentions this and it deserves more space than it gets: relative compression, measured electronically without removing anything.

Every compression stroke slows the crankshaft slightly as the starter fights it. A cylinder with no compression barely slows it at all. So the pattern of crankshaft speed during cranking is a picture of each cylinder's sealing — and the module already measures crankshaft speed continuously, through the sensor described in the crankshaft position sensor. Many scan tools will display it directly as a relative compression test.

The workshop version uses a clamp meter on the battery cable instead: starter current peaks on each compression stroke, so the current trace shows the same information with nothing disconnected. A cylinder whose current peak is visibly smaller than the others is the one to test properly.

It is comparative only — it produces percentages, not pressures — and that is exactly what you want as a first pass. It takes two minutes, needs no plugs removed, and tells you whether the long version is worth doing.

What a compression test does not tell you

It does not distinguish a burnt valve from a bent one, or worn rings from a scored bore. It does not find a head gasket leaking to the outside of the engine, or one leaking coolant into a cylinder that the engine is still burning off.

And it does not find the commonest expensive fault of all on its own: combustion gas being pushed into the cooling system. That one has its own five-minute test with a chemical block tester, described in why an engine overheats, and it is the test to do before agreeing to any cooling-system work on a car losing coolant with nothing on the driveway.

A compression figure is a starting point. P0301 onward will tell you which cylinder to point the gauge at. P0300 deliberately will not — it is the code for a misfire that wanders rather than settling on one cylinder, which is itself an argument for reaching for the gauge, since compression across the board is one of the few things every cylinder shares. Either way, the gauge tells you whether to keep buying ignition parts.

Video guides

Video, for the parts of this that are easier watched than read.