Ohm's Law as a Field Tool Rather Than a Formula

Why this matters

Most techs can recite Ohm's law and almost none of them use it. It gets treated as something you did in class, not something you do standing in front of a panel. That is backwards. The law is only worth anything in the field when you run it before you put the meter on something, so that you walk up with a number you expect. Then the measurement is not data, it is a verdict: the reading either agrees with the nameplate or it does not, and the size and direction of the disagreement usually names the fault outright.

The shop cost of skipping the prediction is the "no fault found" ticket on equipment that is genuinely running at half output. Nothing trips, nothing smells, every reading looks plausible in isolation, and the customer calls back three weeks later about the same complaint.

The gate, stated once

Per load, at steady state, compute the expected value from the nameplate, then measure. Treat a disagreement as an equipment fault only when both of these are true:

• the load is purely resistive - a heating element, an incandescent lamp, a fixed resistor, and specifically not a coil, a motor winding, a transformer, an electronic power supply, or anything with a capacitor in it, and
• measured differs from predicted by more than 10 percent of predicted.

Both conditions, not either. Under 10 percent on a resistive load, suspect your measurement before you suspect the equipment: lead resistance, probe pressure and winding temperature all live in that band. And when the threshold does trip, work in whole-branch steps rather than percentage steps, because the disagreements that mean something arrive as ratios - half, double, one third - not as a few percent.

That gate is going to pass one case below and fail the other, and the two cases look identical on the meter until you ask which kind of load you are holding.

Case one: the heater that is exactly half there

A resistance water heater is rated 4,500 W at 240 V and the customer says it takes forever to recover. Nothing has tripped.

Predict first. Resistance follows from the nameplate: R = V squared / P = 57,600 / 4,500 = 12.8 ohms. Expected current = 4,500 / 240 = 18.75 A. That is what a healthy element string should read.

Before touching it: this is a de-energized measurement, so open the branch disconnect, lock and tag it under 29 CFR 1910.333(b)(2) (29 CFR 1926.417 if the job falls under construction), and prove dead live-dead-live per NFPA 70E-2021, 120.5 before a probe goes anywhere. If you are also pulling the element rather than just metering it, the tank is hot and under supply pressure: close the cold inlet, relieve pressure through the drain or the temperature-and-pressure valve until flow stops, and let it cool, because a scald from a tank at service temperature is a burn injury inside a second.

Now measure. The element string reads 12.6 ohms. Against a predicted 12.8 that is 1.6 percent low, well inside the gate, and cold metallic elements read slightly below their hot value anyway because their resistance rises with temperature. Verdict: the string is intact. That reading is not the fault.

Now the same job on a different unit, same nameplate, same complaint. This one reads 25.6 ohms. That is not "a bit high", it is exactly double, and a clean ratio of 2.0 is the tell. A 4,500 W, 240 V heater is normally two 2,250 W elements in parallel, each 240 squared / 2,250 = 25.6 ohms, and two 25.6-ohm elements in parallel give 12.8. Reading exactly one element's worth means one branch is open.

Carry it through. Supply at the terminals measures 232 V rather than 240. With both elements alive, output is 232 squared / 12.8 = 4,205 W, which is 93.4 percent of the 4,500 W rating; the missing 6.6 percent is the voltage, not a fault. With one element open, output is 232 squared / 25.6 = 2,102 W, exactly half of the 4,205 W the unit should be making at that supply. Current falls to 232 / 25.6 = 9.1 A against a design 18.75 A.

That last number is why nobody caught it. A 30 A breaker sees 9.1 A and holds forever. There is no nuisance trip, no burning smell, no error code. The only symptom the equipment can produce is time: a tank that recovered in 55 minutes now needs about 110. The customer described the fault accurately and the previous tech had no number to compare their reading against, so 25.6 ohms looked like a reading rather than an answer.

Case two: the same gate, run on a coil, invents a fault

Same method, a 24 V contactor coil that a tech suspects of dragging the control transformer down. Measured winding resistance: 18 ohms. Ohm's law says 24 / 18 = 1.33 A, which is 32 VA, and on a 40 VA control transformer that looks like the whole problem.

Clamp it instead and the coil pulls 0.5 A holding, which is 12 VA. The prediction was high by a factor of 1.33 / 0.5 = 2.7, and the equipment is fine.

Nothing is broken except the gate's first condition. On alternating current an inductive load opposes current with impedance, not resistance: Z = square root of (R squared + X squared), where X is the reactance the winding's magnetic field produces. Since X is never negative, Z is always at least R, so a resistance-only prediction on any coil, winding or transformer is always high, never low. The direction is fixed, which makes it easy to catch and impossible to correct for by eye, because X depends on the core, the air gap and whether the armature has sealed in.

The winding resistance is still worth measuring on a coil. It is just answering a different question: compared against an identical coil, or against the same coil's own history, a resistance that has fallen sharply says shorted turns. What it will not do is predict current.

Reading the direction of the disagreement

The middle row is the one that costs the most time. An ohmmeter's own leads commonly contribute a few tenths of an ohm, which is invisible against 12.8 ohms and completely swamps a 0.4-ohm reading. A sibling article covers what a resistance reading cannot see at all; this table is only about interpreting a disagreement you already trust.

Where the squared term changes the stakes

Power on a fixed resistance follows voltage squared, and that asymmetry decides how seriously you take a sag.

At 4 percent low supply, a resistive load delivers 0.96 squared = 92.2 percent of rated output, so you have lost 7.8 percent of the heat for 4 percent of the volts. At 10 percent low, 0.90 squared = 81 percent, a 19 percent loss. That is the whole explanation for the seasonal "it does not heat like it used to" complaint on a shared service that sags under neighbourhood load, and it is why measuring supply voltage under load is not optional on a heating complaint.

A motor does the opposite on the same sag: it holds its shaft load, so current rises as voltage falls. Same law, opposite sign, because the load is not resistive. Do not carry the heater intuition onto a motor.

The case the resistance check cannot catch: a right element on a wrong supply

There is one common fault this whole method passes, and it is worth knowing because it produces a complaint identical to the open branch above.

Fit a 240 V, 4,500 W element to a 208 V nominal circuit. Its resistance is still 12.8 ohms, so it measures perfectly against a prediction derived from its own nameplate, and it will keep measuring perfectly for its entire life. Output, though, is 208 squared / 12.8 = 3,380 W, which is (208 / 240) squared = 75.1 percent of the rated figure. The customer loses about a quarter of the heat, the breaker sees 208 / 12.8 = 16.25 A instead of 18.75 A and is happier than ever, and every de-energized test you run reports a healthy element.

The prediction is only as good as the pair of numbers you fed it, and this is the case where the nameplate is right and the supply is the mismatched half. So take the supply voltage on any output complaint before you take the resistance, and check the nameplate voltage on any replacement part against the circuit it is going onto rather than against the part it is replacing. Elements, heaters and coils are commonly stocked in more than one voltage with near-identical appearance, and the wrong one installs perfectly.

If the equipment is genuinely on a 208 V system, the fix is a part rated for 208 V, not a bigger breaker and not a second element.

Verifying your own prediction before you act on it

Three checks, in this order, and they take under a minute:

1. Re-derive the prediction from a second nameplate quantity. If the plate gives watts and volts, you predicted resistance; now predict current the other way (P / V) and confirm the two agree. A transposed digit shows up immediately.
2. Confirm the ratio names a real branch count. A measured-to-predicted ratio of 2.0 on a two-element load means one element. A ratio of 1.5 on a three-element load means one of three is open, because two 3R branches in parallel give 1.5R against three giving R. If the ratio maps to no plausible branch count, you are measuring through something else.
3. Close the arithmetic against the supply you actually measured, not the nameplate supply. Predicted output at 232 V is not the plate's 4,500 W, and quoting the plate figure to a customer after measuring 232 V is how a correct diagnosis turns into an argument at the invoice.

The failure mode worth naming: writing the prediction down after the measurement. Once you have seen 25.6 ohms it is remarkably easy to convince yourself that 25.6 is roughly what you expected. Predict on the way to the truck, out loud or on the ticket, and the disagreement stays honest.

References

• Trade-standard practice for resistance-element heating equipment and nameplate interpretation
• 29 CFR 1910.333(b)(2) (general industry) and 29 CFR 1926.417 (construction) for de-energizing, lockout and tagging before a de-energized measurement
• NFPA 70E-2021, 120.5, for the live-dead-live verification sequence
• See related: What a Resistance Reading Tells You and What It Hides; How to Use Voltage Drop to Find a Bad Connection; How a Motor Draws What It Draws