The carburettor: how it meters fuel, and why it needs adjusting when injection does not

A carburettor is a remarkable piece of engineering that solves a hard problem with no electricity, no sensors and no moving decisions: it measures airflow and adds the right amount of fuel to it, continuously, using nothing but the shape of a tube.

And the answer to the question in the title is contained in that sentence. A carburettor is a fixed mechanical calibration with no feedback. It is exactly right at one set of conditions, approximately right near them, and progressively wrong as anything moves — temperature, altitude, wear, the fuel itself. Nothing puts it back except a person with a screwdriver.

An injection system does the same job by measuring the result and correcting it, several times a second, forever. That is the whole difference, and it is why one needs adjusting and the other does not.

How it does it without electricity

The source explains this well and it is worth keeping in compressed form.

Air being drawn into the engine passes through a venturi — a deliberate narrowing of the tube. A fluid speeding up through a restriction must lose pressure to do it, so the narrow section is at a lower pressure than everywhere around it. Drill a hole there, connect it to a bowl of petrol held at atmospheric pressure, and fuel is pushed out of the hole by the pressure difference.

The elegance is that the effect scales with airflow. More air means more speed through the venturi, means lower pressure, means more fuel. The mixture stays roughly constant across the range without anything deciding it should.

Around that core sit the parts that patch the places where the simple idea fails:

• A float and needle valve keep the bowl at a constant level, because the fuel level sets how hard the venturi has to pull.
• An idle circuit, because at idle the throttle is nearly shut and there is not enough airflow through the venturi to lift any fuel at all. It feeds from a separate hole below the throttle plate, where the vacuum is high.
• A choke, which restricts the air rather than adding fuel, to enrich the mixture for a cold engine — cold fuel does not vaporise well, so some of what is delivered never burns and the rest has to be enough on its own.
• An accelerator pump, a small squirt of neat fuel when the throttle is opened suddenly, because air accelerates through the venturi faster than fuel can follow and the mixture would otherwise go momentarily lean.

Each of those is a workaround, and each is a separate thing that can block, stick, wear or perish — which is the practical half the source never reaches.

Which circuit has failed, read from the symptom

This is the useful table, and it works because the circuits operate in different parts of the throttle range.

That last one deserves emphasis because it is the one people chase with the adjusting screws for hours. A worn throttle spindle lets air in below the venturi, unmetered, so the engine gets air the carburettor did not account for. No screw can compensate, and the tell is that the idle speed changes when you squirt a little carburettor cleaner around the spindle ends with the engine running.

Tune the engine, then the carburettor

The single most useful rule about carburettor adjustment is that most of it is not carburettor adjustment.

The idle mixture and idle speed screws set two things, and they can only be set correctly on an engine that is otherwise right. Before touching them: ignition timing, plug condition and gap, valve clearances, and no vacuum leaks. An engine with a leaking manifold gasket or a tight exhaust valve will accept a carburettor setting that makes it run, and that setting will be wrong for the engine once it is fixed.

That ordering matters more here than on an injected car, precisely because there is no feedback: on an injected engine a vacuum leak shows up as a positive fuel trim that the module reports and partly corrects, and the oxygen sensor and reading live data explain how to see it. On a carburettor the same leak is silent, and the only evidence is that nothing you do to the mixture screw produces a good result.

What has changed since the source was written

Modern petrol is harder on carburettors than the fuel they were designed for. Ethanol content attacks the rubber and fibre components — float needle tips, accelerator pump diaphragms, gaskets — and it holds water, which corrodes the zinc alloy the body is cast from. A car that sits for months is the worst case, because the fuel left in the bowl evaporates and leaves behind the varnish that blocks the idle jet.

The practical consequences are simple and worth stating:

• A carburetted car that stands should be stored with a full tank and, ideally, a drained bowl.
• Rebuild kits are cheap and the perishable parts are all in them. A carburettor that has been standing usually needs the kit, not a replacement.
• The blocked idle jet is nearly always the whole fault on a car that ran when it was parked and will not idle now.

Why injection won, honestly

One merged piece argues the carburettor's corner against injection, and the fair version of that argument is short.

A carburettor is cheaper, simpler, repairable with hand tools, and needs no electrical system to run. For an engine in a remote place, that is a real advantage and it is why they persisted in agricultural and small industrial engines long after cars moved on.

What it cannot do is adapt. It cannot lean the mixture out when the air is thin at altitude — the source's own material notes the effect — and it cannot richen it when the air is cold and dense, and it cannot notice that its own jets have worn. Above all it cannot hold the mixture at the narrow ratio a catalytic converter needs to work, which is what actually ended it: emissions law required closed-loop control, and closed-loop control requires something that measures. See the catalytic converter for why that window is so narrow.

So the carburettor did not lose on principle. It lost because the thing that replaced it can measure what it did, and a carburettor is a very good guess made once at a factory.

And it is still worth knowing

A large part of the fleet this archive was written for is carburetted, and those cars are now old enough that the failures above are not hypothetical — they are what is wrong with them. The parts are available, the faults are diagnosable by symptom, and almost all of the repairs are cheap.

The one thing to carry away: when a carburetted engine runs badly, identify which throttle range it runs badly in before you touch anything. That single observation names the circuit, and the circuit names the part.

Video guides

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