Electrical troubleshooting: split the circuit before you start pulling parts

A vendor's support engineer once kept me on the phone most of an afternoon, dead certain we had lost a phase. A pick and place arm would not home and he wanted the servo drive pulled and shipped back. It was a prox sensor knocked a couple of millimeters out of alignment. Electrical troubleshooting goes wrong that way far more often than it goes wrong from not knowing enough: somebody commits to a theory in the first five minutes, then spends four hours defending it. Here is the order I actually work in, the two measurements that settle most electrical troubleshooting, and the arc flash numbers I want in my head before I open a live door.

Where electrical troubleshooting actually starts

Faults cluster in two places, and neither of them is where the alarm is pointing. Where a human touches the machine, and where something physically moves. Motors, relays, connectors, E-stops. That is not physics, it is just where the wear and the accidents land, and it is where electrical troubleshooting should open.

So E-stops first. Every time. Somebody on a scaffold leans into one, is too embarrassed to mention it, and twenty minutes later you are on your knees with a meter.

Connectors next, and those cost me more hours than I care to admit. An RJ45 that clicks in is not the same as an RJ45 making contact on all eight pins. Wiggle it and watch the state, or pull it, hit the contacts with cleaner, and reseat.

Then fuses. All of them, not the three that happen to be at eye level. Fuse blocks with a neon or LED indicator are worth the extra couple of dollars a pole for exactly this reason: the answer is visible from across the panel instead of costing you a meter check per fuse.

Ask what was worked on last, too. Half the time something got moved during an unrelated job and nobody said anything.

Split it in half, then split what is left

The most useful habit in electrical troubleshooting is refusing to check things in order.

Say you have sixteen sootblowers on a boiler, and every one of them feeds a "ready" contact wired in series into one PLC input. The input is off. One of those sixteen contacts is open and the controller cannot tell you which.

Working the string from one end is sixteen trips in the worst case. Go to the eighth one instead and check continuity there. Now you know whether the bad contact lives in 1 through 8 or in 9 through 16. Split the bad half, split it again. Four checks instead of sixteen, and a lot less catwalk.

The math holds for a long multi-conductor run, a daisy-chained network, a string of limit switches. Sixteen points take four checks. Sixty-four take six.

Four continuity checks instead of sixteen. Every test throws away half of what is left.

 

It needs two things up front: knowing what the circuit should do, and knowing where its physical middle is. The second is why the print matters more than the meter. You will find a print that is both present and current maybe a quarter of the time. The rest of the time you trace it yourself, and the decent thing is to mark up the drawing as you go so the next guy gets one.

One trick for a long cable you suspect is open: unwire both ends, twist two conductors together at the far end, and ohm that pair from where you are standing. Full out-and-back loop without walking the run twice.

Voltage present is not voltage doing work

Most electrical troubleshooting stops one measurement too early. Probes on, 24 V on the display, circuit declared healthy, go look somewhere else.

Take a solenoid valve that will not shift. Check signal and common at the cabinet terminal. Then check both at the coil. What you are after is the difference between those two places, because that difference is the resistance you are hunting.

At the cabinet terminal	At the coil	What it points to
Nominal on signal and common	Nominal on both	Coil or valve. Ohm the coil, then check whether the spool is stuck.
Nominal on signal	Nothing on signal	Open in the signal conductor between the two points.
Nominal on common	Nothing on common	Open in the return path, the leg most people never check.
Nominal at rest, sags hard when energized	Sags with it	High resistance joint. A screw backed out in a terminal block does exactly this.
Nothing at all	Nothing	Not a field problem. Output card, fuse, or the logic never called for it.

Four readings on one solenoid circuit, and what each pair rules out.

Row four is the one worth burning into memory. A connection that is loose but still touching reads perfectly with no load on it and collapses the moment current tries to pass. A static voltage check will never show it. The measurement has to happen with the circuit doing its job.

Both legs, both ends, at rest and under load. The gap between the pairs is the fault.

 

Know what the meter is telling you, as well. Zero volts can mean a dead circuit, or zero difference between two points that are both live. Two 24 V supplies in one panel are not necessarily at the same potential, and referencing one against the other gives you a number that makes no sense to chase.

Know the voltage class before the probes come out

Believing you are on 24 V DC when the circuit is actually 240 V AC is the mistake that ends careers, and it happens because nobody read the label first.

NFPA 70E handles this with a table method instead of a rule of thumb, and it is worth knowing before any electrical troubleshooting on a live panel. It starts by asking whether the equipment is in normal condition, meaning all of these are true: properly installed, maintained per the manufacturer's requirements, doors closed and secured, covers in place and secured, no evidence of impending failure. Miss any one and the equipment counts as abnormal, which raises the PPE requirement even for something as routine as operating a breaker.

For tasks that do call for arc rated gear, the categories land like this.

Equipment	Fault current	Clearing time	PPE cat.	Min arc rating	Arc flash boundary
Panelboards and other equipment, 240 V and below	25 kA	0.03 s (2 cycles)	1	4 cal/cm²	19 in.
Panelboards over 240 V through 600 V	25 kA	0.03 s (2 cycles)	2	8 cal/cm²	3 ft
600 V class motor control centers	65 kA	0.03 s (2 cycles)	2	8 cal/cm²	5 ft
Other 600 V class equipment, 277 V through 600 V	65 kA	0.03 s (2 cycles)	2	8 cal/cm²	5 ft
600 V class motor control centers	42 kA	0.33 s (20 cycles)	4	40 cal/cm²	14 ft
600 V class switchgear and switchboards	35 kA	0.5 s (30 cycles)	4	40 cal/cm²	20 ft

Arc flash PPE categories for AC equipment per NFPA 70E. Every row assumes an 18 in. minimum working distance.

Compare rows three and five. Same 600 V class MCC. The category goes from 2 to 4 and the boundary from five feet to fourteen, and the only thing that changed is how long the upstream device takes to clear the fault. Your PPE is set by the protection ahead of the panel, not by the number on the nameplate.

One provision cuts the other way. On equipment rated 600 V and below protected by upstream current limiting fuses or current limiting molded case breakers sized at 200 A or less, the arc flash PPE category drops by one, though never below category 1.

One case NFPA 70E puts in the no-likelihood bucket regardless of equipment condition: control circuits at 125 V AC or DC and below with nothing else exposed above that, including opening a hinged cover to get at it. That covers most 24 V work, which is most of what a controls guy touches.

Two tools that earn their place in the bag

A meter gets you through most electrical troubleshooting. Two other things pay for themselves fast.

A clamp on ammeter, because current tells you what voltage cannot. A pump that keeps tripping, a heater bank with one leg burned open: both read fine on a voltage check. The clamp also keeps you out of series with the circuit, which matters more than people think. A meter in current mode is close to a dead short. Put the leads in parallel across a live device with the meter still on amps and you blow the meter fuse, if you are lucky.

A 4-20 mA signal source is the other. Unwire the suspect transmitter, land the source in its place, feed 12 mA, and see what the controller reads. Mid scale means the wiring and the input card are fine and the transmitter is your problem. Nothing means they are not. Either way the loop got cut in half without a guess. And remember while you are in there: an analog input card generally does not power the loop, it only gives you a resistor to measure across, so a loop short on volts leaves the transmitter dark no matter how clean the wiring is.

An analog input card measures the loop, it does not power it. Worth knowing before you condemn a transmitter.

 

When it really is the controller

"It is never the PLC" holds right up until it does not, and that belief costs more electrical troubleshooting hours than any wiring fault I can name.

A batch plant I spent time on had a flowmeter pulsing around 30 Hz into a standard input on a remote rack. The software house was certain the meter was bad. It was not. With that I/O count on that scan, a standard input could not catch 30 Hz reliably and the job needed a high speed counter card. Hours to prove, mostly because everyone had already settled on the instrument.

Drives are their own category. Controllers mostly do not fail. Drives do. And a replacement drive almost never comes configured the way the dead one was, so before you blame field wiring, get the manual out and confirm which terminal the drive expects for run, which one it wants for the speed reference, and whether it is even set to take either of them from the terminal block. A good share of "the new drive is dead" calls come down to a start source still pointed at the keypad or a comm port.

Most electrical troubleshooting on a swapped drive ends at the control terminals, not at the motor leads.

 

Ground faults on a drive are the other regular. If it trips on a ground fault and the motor megs clean, remember the drive is measuring imbalance between output phases, not insulation. Long motor cable, a damp junction box, or nameplate data entered wrong all produce that imbalance with perfectly good insulation everywhere.

Common questions about electrical troubleshooting

What is basic electrical troubleshooting?

Working from what the circuit is supposed to do toward what it is actually doing, with measurements instead of theories. In practice: confirm the intended behavior, clear the cheap and obvious causes, then halve the suspect section over and over until the fault sits between two points you have proven good.

Can I do electrical troubleshooting myself, or does it need an electrician?

On 125 V control work with nothing higher exposed nearby, a correctly rated meter and basic PPE cover most of it. Above that the arc flash category decides what you wear, and abnormal equipment condition raises that requirement for tasks that would otherwise be routine. Work you are not trained and qualified for goes to somebody who is, and that line has nothing to do with confidence.

What does a voltage drop tell me that a voltage reading does not?

That a connection is bad without being broken. A high resistance joint reads at nominal with no load and sags under load, so the fault only appears when the measurement is taken with current flowing.

Do I need an expensive meter for electrical troubleshooting?

No. You need one with the right CAT rating and voltage rating for the panel you are standing in front of, and you need to have read its manual. A meter used above its rating is the hazard, not the price on the box.

When the answer really is a replacement part

Sometimes electrical troubleshooting ends where you were hoping it would not. The module is dead, the drive did fail, the fuse block cooked. That stings most on legacy racks, where the part that quit went obsolete years back and the migration path costs more than the line earns this quarter. We keep surplus and refurbished Allen-Bradley I/O, drives, relays, fuse blocks and terminal blocks on the shelf for exactly that, tested and ready to ship, so a diagnosis you worked hard for does not turn into a six week wait.