Clamp Meters

Intuition

A measurement that touches nothing

A clamp meter is a current meter with a hinged jaw on top. The jaw opens, goes around a conductor that is carrying current, and closes again. Nothing is cut, nothing is unplugged, nothing is unsoldered, and the circuit does not know the instrument is there.

A warehouse scanner reads the tag through the side of a sealed box. It never opens the box, and it does not need to: what it is reading was always leaking out. A clamp meter works the same way on a conductor. Current makes a magnetic field around itself whether anyone is looking or not, and the jaw is there to collect it.

The field is not large. Around a single conductor carrying 45.0 A, at the 12.0 mm radius of a typical jaw, it comes to 750 µT, which is roughly fifteen times the earth's own field and small enough that the instrument spends most of its design effort on collecting it properly.

Then there is the fact that catches everyone once. The jaw does not read a current; it reads the net current through its own opening. Close it around one conductor of a two-core flex and it reads 45.0 A. Close it around the whole flex, line and return together, and the two currents cancel inside the opening: the display settles at 0 A, and the appliance at the end of the flex goes on working exactly as before.

The instrument is honest in both pictures. Only one of them asks the question you meant.

Practitioner

Putting the jaw where it belongs

The sequence is short, and most of it is about what goes inside the jaw rather than what the display says.

1. Separate one conductor. Line and return have to be apart before there is anything to read. On a moulded flex that means a break-out lead, not a knife.
2. Choose the function before the jaw goes anywhere. On meters offering both, AC and DC current are different sensors, and a DC clamp wants its zero set with the jaw closed and empty.
3. Pick a range that suits the current you expect, or let the instrument autorange.
4. Close the jaw fully. The mating faces are a magnetic joint as much as a mechanical one, and grit, paint or a half-latched catch all read low.
5. Centre the conductor, and hold the jaw still while the reading settles.
6. Note what you clamped. A current reading with no record of which conductor it came from is an anecdote.

Range choice deserves a paragraph of its own, because a clamp built to reach hundreds of amps spends most of its life near the bottom of its span. Our 45.0 A conductor sits above a 40.0 A range, so the instrument has to be on the 400 A one, where the reading is 11.25 % of full scale. A 500 mA current on that range is 0.125 % of it, which is the territory where a specification's fixed term swamps its proportional one, as accuracy, resolution and measurement error sets out.

Both ranges start at zero. What separates them is where they stop, and how much of the span your reading is using.

There is a way out of the bottom of the range, and it is one of the more satisfying tricks on a bench. Pass the conductor through the jaw more than once. The jaw counts ampere-turns, so 10 passes of the same wire look to it like 5.00 A instead of 500 mA, and you divide the reading afterwards. The relationship is the transformer's:

With one pass the sensing winding carries 250 µA and its burden develops 2.50 mV; with ten it carries 2.50 mA and develops 25.0 mV, ten times the signal from the same circuit for the price of some slack cable.

Position inside the opening is the other thing worth a number. Where the sensor is a single element in the jaw wall rather than a closed magnetic path, what it sees depends on how far away the conductor is. Centred, it sits 12.0 mm away and sees 750 µT. Pull the conductor 4.0 mm nearer, to 8.0 mm, and the field rises to 1125 µT, 50.0 % high. Push it the same 4.0 mm the other way, to 16.0 mm, and it falls to 562.5 µT, 25.0 % low.

These are the numbers for the worst case, a single sensing point. A well-closed magnetic path collects the field all the way around instead, and cares far less. Real instruments sit somewhere between the two and publish a position error of their own.

Engineer

Two instruments wearing one shape

Two quite different sensors live behind the same hinged jaw, and knowing which one you are holding explains most of what a clamp meter can and cannot do.

Both start from the same place. A long straight conductor surrounds itself with a field whose strength falls off with distance:

with the constant 1.2566 µH/m setting the scale. At the 12.0 mm jaw radius our 45.0 A gives 750 µT; at 24.0 mm, twice as far, 375 µT, exactly half. That one-over-distance shape is where Layer 2's position arithmetic comes from.

Nothing about this curve is specific to a clamp meter. The jaw radius simply picks a point on it.

Why the jaw is made of iron

Half a millitesla is a thin signal to build an instrument on, so the jaw is not empty. It is a magnetic circuit: a split ring of high-permeability material that closes around the conductor and gives the field a path far easier than air. Ampere's law taken around that path gives a magnetising field of 596.8 A/m for our conductor, and the material's response to it is the same linear relation any core obeys:

With an illustrative relative permeability of 300, that magnetising field produces 225 mT inside the core, some three hundred times what the same current makes in open air. The arithmetic ignores the reluctance of the gap every jaw has to have, so read it as the size of the prize rather than as a measurement. It is also why a clamp cares so much about a clean, fully closed joint: the gap sits in series with the whole magnetic path and a small one dominates it.

The coil, its secondary and its burden

An AC clamp is a current transformer. The conductor is the primary, a single turn threading the core, and a winding on the core is the secondary. Instead of being left open, the secondary is closed through a small resistance called the burden, and the instrument measures the voltage across that burden.

The secondary loop is closed through the burden, and the meter reading that burden takes none of the current.

The high ratio is what makes the arrangement manageable: two thousand turns divide the conductor's current down to something a small resistor can carry without complaint. The same reasoning turns up wherever turns ratios trade current for voltage.

The element that answers to a steady field

A coil answers to a changing field. Hold the current still and it produces nothing, which is why a plain AC clamp reads a dead zero on a battery cable quietly delivering hundreds of amps. Reading that current needs a sensor that responds to the field itself, usually a Hall element in a gap cut in the core.

Take an illustrative element with a Hall coefficient of 0.00040 cubic metres per coulomb and a thickness of 200 µm, biased with 5.00 mA. In the open field beside our conductor it would give 7.50 µV, below the noise of most things you could build to read it. In the core's concentrated field it gives 2.25 mV, which is an amplifier's ordinary day's work. The core is not a refinement here; it is what makes the measurement exist.

What each one hears

The two sensors have opposite weaknesses at the bottom of the frequency range. Model the coil's roll-off as a single corner at an illustrative 20.0 Hz: at the corner itself the instrument reads 70.7 % of its mid-band value, at 50.0 Hz it has recovered to 92.8 %, and below the corner it falls away towards nothing. The Hall element is flat all the way to a steady current, and pays for it with offset and drift instead.

The corner drawn here is an illustrative one. Its position varies with the core and the burden, but the shape does not.

Professional

The readings a clamp gives away

A clamp buys its convenience with accuracy, and the trade is better stated than discovered.

A clamp's accuracy is usually written as a percentage of reading plus a fixed term, and the fixed term dominates whenever the current is a small fraction of the range, as Layer 2's 11.25 % reading already hints. Add the position error, the jaw's own gap and whatever the neighbouring cables contribute, and a clamp reading is comfortably the least precise of the ordinary current measurements. A current-sense resistor beats it by an order of magnitude and needs the circuit broken to fit one, which is the whole trade in a sentence.

A DC clamp has an extra ritual. The Hall element's offset drifts with temperature and with whatever the core last remembered, so the zero button means it. Close the jaw on nothing, zero it, take the reading, and re-zero whenever the jaw has been near something magnetic or the instrument has changed temperature. A reading taken without that step can be wrong by more than the current you were after.

Waveform matters here as much as it does on a handheld meter. The sensor reports the field faithfully; what the instrument does with it afterwards is a design choice, and a mean-responding AC clamp on a variable-speed drive, a dimmer or a rectified load is being asked a question it was never calibrated for. True RMS earns its money on a clamp, because the currents worth clamping are so often chopped.

Some jobs need a different jaw. Flexible current probes replace the rigid core with a coil on a bendable former, giving up the core's concentration and gaining the ability to go around a busbar or a whole bundle; they read AC only. Fast events need a probe designed for them and an oscilloscope behind it, because a clamp meter averages over a good fraction of a second and a motor's starting surge is over before the display has decided anything.

One property is worth ending on. A clamp takes almost nothing from the circuit it measures. An ammeter in series inserts a shunt and steals a share of the loop's voltage, which is the whole subject of meter loading effects; a clamp's only cost is a little extra reluctance around the conductor. The instrument that disturbs a circuit least is also the one that measures it worst, and choosing between them is what measuring voltage, current and resistance is really about.