Using a multimeter on a car: the four measurements that find most faults

A multimeter measures what is there. It does not measure what a circuit can deliver, and almost every wasted afternoon in automotive electrics lives in the gap between those two things.

The reason is a number on the back of the meter. A digital multimeter has an input impedance around 10 megohms, so on the volts range it draws a few millionths of an amp. That is deliberate — it means the meter barely disturbs what it is measuring. It also means the meter will happily report 12.4 volts at the end of a wire whose connection has corroded so badly that it could not light a bulb. There is enough of a circuit left to satisfy a few microamps. There is not enough left to run anything.

So the meter says the supply is fine, and the supply is not fine, and nothing about the reading looks suspicious. Corroded or loose connector is the most-referenced cause in this site's fault code corpus by an enormous margin — it appears on over three thousand codes — and this is precisely why: it is the commonest fault and the hardest one to see.

This article is a translation of a Persian original with 297,000 views. It is a competent general electronics tutorial and it does not contain the measurement above. Nor does it contain the one that fixes the problem, which is where this starts.

1. Voltage drop — the one nobody does

Every other measurement here asks a question about a component. This one asks a question about the wire, and it is the only measurement that answers it honestly.

Instead of putting the black lead on ground and the red lead on a point in the circuit, you put both leads on the same conductor, one either side of the section you are suspicious of — across a connector, across a length of cable, across an earth strap and the body it bolts to. The meter then reads the voltage the circuit is losing in that section. On a perfect conductor it reads zero, because a perfect conductor cannot have a voltage across it.

And it only works with the circuit switched on and working. Voltage drop is current multiplied by resistance, so with no current flowing there is no drop, and a badly corroded joint reads a flawless 0.00 volts. Testing an unloaded circuit is how you produce a confident clean bill of health on a wire that is about to strand somebody. Crank the engine, switch the headlamps on, run the fan — then read.

The figures are widely agreed and they are tighter than most people expect:

On a starter circuit those same figures apply to the whole of each side rather than to each joint, because the cable runs are long and the current is enormous: about 0.5 V across the supply side and 0.2 V across the ground side, measured while cranking. Anything beyond that and the starter is not receiving the battery voltage the battery is producing. The starter motor has where to put the probes, and the arithmetic that makes those limits look so small — at 300 amps, 0.5 V is under two milliohms.

Start on the ground side. Most of these faults are earths, because an earth is a bolt through painted metal into a body that spends its life in water and salt, while the supply side is at least a purpose-made terminal. Put the black lead directly on the battery negative post — the post, not the clamp — and the red lead on the metal body of the thing that is misbehaving, then load the circuit.

Then halve the distance. If a section reads badly, move one lead to a point in the middle of it and read again. Two or three moves puts you on the joint. This is the whole technique, and it beats replacing parts in order of price.

2. Why a resistance test does not find the same fault

This is worth doing with numbers, because it is the thing that makes the rest make sense.

A connection that drops half a volt while carrying ten amps has a resistance of 0.05 ohms. That is a serious fault — it is enough to stop a fuel pump priming properly or to make a heated oxygen sensor take twice as long to come up to temperature, which is P0135 territory.

Now try to find it with an ohmmeter. The meter's own test leads have more resistance than the fault does — typically 0.1 to 0.3 ohms, and more if they are cheap or old. The reading will be dominated by the leads, the fault will be inside the meter's own tolerance on that range, and the display will show something that looks entirely normal. Touch the two probes together before you believe any low-resistance reading; whatever number that shows is being added to everything you measure.

And the continuity beeper is worse, because it beeps on anything below a couple of hundred ohms. A joint four thousand times worse than it should be beeps exactly like a perfect one.

That same 0.05 ohm joint, with a starter pulling two hundred amps through it, drops ten volts and turns two kilowatts into heat inside a plastic connector. It is not a subtle fault. It is simply invisible to the instrument most people reach for.

3. Voltage — what it does prove

Straight voltage measurement is still the first thing you do, as long as you know what it is worth. It proves a supply is present, it proves a switch has changed state, and it is the right tool for checking battery and charging figures — see the lead-acid battery and the alternator and charging system for what those numbers should be, including the range the archive got wrong.

Two habits worth building.

Back-probe, do not pierce. Slide the probe in alongside the wire at the back of the connector, or use a proper back-probe pin. Pushing a sharp probe through the insulation opens a path for water into the copper, and the corrosion that follows is the exact fault described at the top of this article — arriving months later, several inches from anything anyone would suspect.

Use min/max on an intermittent. Most meters will record the lowest and highest reading they saw while you were not watching. On a fault that appears once every twenty minutes, that turns a measurement into a trap you can leave set — which is how you catch the kind of failure described in the crankshaft position sensor without having to be looking at the screen at the moment it happens.

4. Current — and the mistake that costs a meter

Current is the one measurement that requires you to break the circuit and put the meter in the middle of it, and the original is right to lead with that. It is also where meters die.

The amps range is a near-short circuit inside the instrument, protected by a fuse. Leave the red lead in the 10 A socket, forget, and go to measure a battery voltage, and you have connected a dead short across the battery through a fuse — best case, a blown fuse and a meter that now reads nothing on that range; worse case, an arc at the probe tip. Move the lead back to the volts socket the moment you have finished.

Most meters split it into a milliamp socket and a 10 A socket, and the 10 A one is usually unfused or fused far higher. Anything on a car above a few amps wants a clamp meter instead: the jaws read the field around the conductor, so nothing has to be disconnected and nothing is in series. The original mentions clamp meters and they are the sensible answer for anything at starter or alternator scale.

The genuinely useful low-current job is finding what is flattening a battery overnight — that is a whole method in itself, and it is in finding a parasitic drain.

5. Resistance — the one the original teaches best

The original's advice here is sound and worth keeping.

• Disconnect the power first. An ohmmeter works by pushing its own small current through the thing being measured; applying an external supply to it gives a meaningless reading at best and destroys the meter at worst.
• On a manual-ranging meter, start on the highest range and work down. Autoranging has made this less important and it has not made it wrong.
• Do not measure a component while it is still connected to its circuit, or you are also measuring everything in parallel with it.

What resistance is genuinely good for is components rather than wiring: an injector winding, an ignition coil primary, a heater element, a temperature sensor against its resistance-versus-temperature table. Those have resistances measured in ohms or kilohms, comfortably above the noise floor of the instrument.

Add a wiggle test to it. With the meter on continuity across a suspect length of loom, move the harness with your hand. A break that only appears when the engine rocks on its mounts will announce itself as a chirp from the beeper, and it will never show up on a static reading.

The diode test, which the original gets wrong

It says: put the probes across the diode, and if the meter makes no sound, the diode is good.

That catches a shorted diode and nothing else. A diode that has failed open — which is at least as common, and is what happens inside an alternator rectifier — is silent too, and by this test it passes.

A diode test is a test in both directions, and it needs the meter's dedicated diode range rather than the beeper:

• Forward — red lead to the anode — should read roughly 0.4 to 0.7 volts on a silicon diode. That is the junction's forward voltage, not a resistance, which is why the diode range exists at all.
• Reverse — leads swapped — should read open, with no number and no beep.

A good diode gives you one number and one open. Two opens is a dead diode; two numbers, or two beeps, is a shorted one. Anything other than exactly one of each is a failed part.

A note on test lights

One of the pieces folded into this article was about building an LED test lamp, and it is worth ending on, because it is the same argument in a different tool.

An old-fashioned test light with a filament bulb in it draws current — a quarter of an amp or so. That makes it a load, and it is why it finds faults a meter does not: on a supply crippled by a bad connection, the bulb glows dim or not at all, while the meter beside it reads twelve and a half volts.

An LED test light draws almost nothing. It is bright, it is cheap to build, it will not blow, and on that same crippled supply it lights up perfectly. It behaves like the voltmeter, not like the bulb.

Neither is better. They answer different questions, and knowing which one you are asking is most of the skill. If you take one thing from this article, take that: presence is not capacity, and the only measurement that tells you the difference is the one taken while the circuit is working.

Video guides

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