As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
When no jaw will close around the bar you have four options: thread a flexible coil around it, clamp a thin jaw onto one bar of a stack and account for what the other bars are carrying, move the measurement upstream to a round cable, or read something that is already installed, such as a metering current transformer or a shunt. The flexible coil is the usual answer inside switchgear, and it works only for alternating current. None of the four is as clean as a jaw closed around a single round conductor, so the real question is which error you are willing to carry.
Why a jaw has an opening and a busbar does not care
Clamp jaws are specified by how far they open, and the figures are small compared with distribution hardware. The KAIWEETS digital clamp meter with the D shaped jaw states an opening of up to 1.1 inches, which is generous for a handheld instrument and still nowhere near a 100 mm bar. At the other extreme, the Fluke 301D with its 10 mm jaw arm is built thin on purpose, to slide between conductors rather than to swallow large ones, and it states alternating current measurement to 600 A with a CAT III 300 V safety rating.
A busbar defeats both for reasons that have nothing to do with current. It is flat rather than round, so the useful dimension is width rather than diameter. It is frequently stacked, two, three or four bars per phase with small air gaps between them, and the stack behaves as one conductor electrically while presenting a cross section no jaw can span. It is usually fixed in place on insulators with no slack to lift or bend. And the depth of the jaw throat matters as much as its opening, because a bar mounted at the back of a chamber may be unreachable even by a jaw that would otherwise fit around it.
Before any of that there is the access question. Opening a panel to reach a bar at all is work that needs planning, and knowing what is behind the cover before you remove it changes the plan. This is one of the honest uses for a borescope, which lets you look into a chamber through a small aperture and establish the geometry before committing to a method.
Four ways to get a number off a bar
| Method | What it requires | What it gives you | Where the error comes from |
|---|---|---|---|
| Flexible coil threaded around the bar | A loop long enough to encircle the stack, and an alternating current | The whole phase current in one reading | Coil position, the clasp, nearby phases, and no direct current at all |
| Thin jaw on one bar of a stack | Clearance on both faces of a single bar | That bar only, not the phase | Unequal sharing between parallel bars |
| Measure a round cable upstream or downstream | A round conductor somewhere in the same circuit | The circuit current, cleanly | The cable may carry a different total, or may not exist |
| Read an installed transformer or shunt | Access to terminals and a known ratio | Scaled current through existing hardware | Accuracy class and burden of hardware you did not choose |
The order in that table is roughly the order of convenience, not of accuracy. If a round cable exists anywhere in the circuit, measuring it with a conventional jaw is almost always the better number, and the fact that it is twenty paces away in a cable basement is a walk rather than a measurement problem.
Paralleled bars and the fraction you actually read
Here is the trap in the second method. Four bars bolted together at both ends look like a perfect four way split, and they are not. Current divides between parallel paths in inverse proportion to impedance, and at mains frequency the impedance of a flat bar in a stack is shaped by two effects that both push current outward. Skin effect concentrates current toward the surface of each conductor. Proximity effect, driven by the fields of the neighbouring bars and of the return conductors, pushes it further toward the outer faces of the outer bars. The middle bars of a stack routinely carry noticeably less than the outer ones.
So a reading from one bar multiplied by four is an estimate with an unknown error, and the error is larger the tighter the stack and the higher the frequency content of the load. If the total is what you need, measure every bar in the phase and add the readings, taking them as close together in time as you can so the load has less chance to move underneath you. Then sanity check the sum against something independent, whether that is an installed meter, the upstream cable, or the rating of the protective device. Doing the same measurement twice from the same marked position is also the only way to tell instrument scatter apart from genuine variation, which is the practical meaning of repeatability in a measuring instrument.
Where the coil sits changes what it reads
A Rogowski loop assumes a uniform winding and a conductor somewhere near the middle. Around a wide flat bar you get neither. The bar fills one axis of the loop and the winding has to take whatever shape the available space allows, so the coupling is not quite the ideal case the calibration assumed. Manufacturers quote a positional error for this, and in practice the useful discipline is simple: get the loop into one plane, keep the clasp away from the bar and away from adjacent phases, and mark where you put it so that the next reading is comparable with this one.
Drifting or jumping readings on a busbar usually have a physical explanation rather than an electronic one. The load really is changing, the loop has shifted, an adjacent phase is coupling in, or the instrument is autoranging across a boundary. The same reasoning applies to unstable figures on any handheld instrument, as set out in the explanation of why multimeter readings jump around, and the fix is nearly always to control the geometry rather than to distrust the meter.
The opposite problem is worth knowing too, because it shows how specialised these instruments become at the extremes. The Fluke 369 FC leakage current clamp carries a 61 mm jaw for large cables with a minimum resolution of 1 microamp and a maximum of 60 A, true RMS conversion, a forward facing work light and a backlit display for dark cabinets. That is a big jaw built for a tiny current, the reverse of the busbar problem, and a direct current function is not stated in the listing bullets, because leakage hunting does not need one.
Working the problem on the bar in front of you
Start by asking whether the bar can be isolated, because every method is easier and safer dead. If it must be live, settle the measurement category and voltage rating of the probe, the leads and the instrument against the position in the installation before anything goes near the chamber. Then survey what is already installed: an existing metering transformer with a known ratio beats everything you can add from outside. Look next for a round conductor in the same circuit. Only when all of that is exhausted does the flexible coil around the stack become the method of choice, and at that point confirm that the current is alternating, because a coil will report zero on a direct bus and look entirely convincing doing it.
Finally, write down how you took the reading, not just what it said. The position of the loop, which bar you clamped, whether the load was steady and what the upstream device was rated at are all part of the measurement. A busbar current figure with no record of the geometry behind it cannot be repeated, and a number nobody can repeat is a number nobody can act on.
Frequently Asked Questions
Can I just clamp one bar and multiply by the number of bars?
Only as a rough estimate, and sometimes a poor one. Current divides between parallel bars according to impedance rather than count, and at mains frequency the proximity of the other bars and the return path pushes current toward the outer faces of the outer bars. Differences of tens of percent between bars in the same phase are normal. Measure every bar and add them if you need the total, rather than measuring one and scaling it.
Does the gap at the clasp of a flexible coil matter?
It is the least uniform part of the winding, so it is the part most sensitive to what is nearby. Keep the clasp away from the conductor itself and away from neighbouring phases where you can, and keep the loop in a single plane rather than letting it twist. Doing both repeatably matters more than doing either perfectly, because a repeatable position makes successive readings comparable.
How much does conductor position inside the loop change the reading?
A well made uniform winding is only mildly sensitive to position, and manufacturers quote a positional error figure for exactly this reason. The error grows as the conductor approaches the winding itself, so a conductor pressed against the inside of the loop is the worst case and a centred one is the best. On a wide flat bar you cannot centre anything, so accept the error and keep the geometry the same every time.
Is it safe to put a flexible coil around a live busbar?
That depends on the installation and on the rating of the probe, not on the convenience of the measurement. Exposed live busbars inside switchgear carry very high fault energy, and the probe, the leads and the instrument all need a measurement category and voltage rating appropriate to the position in the installation. The fact that a flexible loop is easier to fit than a rigid jaw does not reduce the hazard it is being fitted into.
What about reading a current transformer that is already installed?
It is often the cleanest answer, because the transformer is already correctly positioned around the whole bar. You need the ratio, you need a safe way to read the secondary without open circuiting it, and you need to accept whatever accuracy class the transformer was specified to, which for a metering class device is usually good and for a protection class device is deliberately not. Never open circuit the secondary of a loaded current transformer.
Why does my reading drift while I hold the probe still?
Usually because the load itself is moving rather than the instrument. Panels feed a changing mixture of equipment and the current genuinely varies second by second. If the reading wanders when the load is known to be steady, suspect the probe position, a loose clasp, or interference from an adjacent phase, and repeat the measurement from a marked position to separate the instrument from the installation.