How to Test Ground Resistance

1. Know what the code actually requires — and what the spec adds

NEC 250.53(A)(2) says a single rod, pipe, or plate electrode must be supplemented by an additional electrode unless the single electrode shows 25 ohms or less to earth. That's the entire code obligation — and the practical read is important: driving a second rod (6 ft minimum apart, per the same section) satisfies the code with no test at all, which is why most contractors just drive two rods and move on. There is no NEC requirement to re-test after supplementing, and no NEC requirement that the final system hit 25 ohms.

Testing enters the picture when the design demands it: a spec that requires a maximum system resistance (5 ohms for substations and telecom sites is typical), a commissioning requirement, or an owner's insurance or lightning-protection standard. In those cases the engineer's number governs, the test method is usually specified, and your documentation is a deliverable. Read the spec section before mobilizing a tester — the target, the method, and the required weather conditions are often all spelled out.

One more distinction worth keeping sharp: ground resistance is about earth coupling for lightning and surge. It is not the fault-clearing path — NEC 250.4(A)(5) is explicit that earth shall not be the sole equipment grounding conductor. A beautiful 2-ohm electrode does not excuse a missing EGC.

2. Isolate the electrode before you measure

A fall-of-potential test measures one electrode system. If the electrode under test is still tied to the utility neutral, building steel, metal water piping, or a bonded fence, you're measuring the parallel combination of everything — and the number will look great while telling you nothing about your electrode. Disconnect the electrode from the system at the test point, which requires the service to be de-energized or the electrode to be on a test link designed for this.

This is the step that separates a real test from a wishful one, and it's also where the safety stakes live: never lift a GEC on a live service. On new construction, test before the service is energized and before the electrode ties into the building network — it's the one moment in the building's life when isolation is free.

Clean the connection point to bright metal before clamping test leads. A corroded clamp adds series resistance that reads as bad earth.

3. Set up the fall-of-potential test geometry

The three-terminal test drives a known current from the electrode under test (E) to a current stake (C) placed far away, and measures voltage at a potential stake (P) placed between them. Resistance = measured voltage ÷ injected current, computed by the tester. The whole art is in the distances: the current stake must be far enough away that the electrode's and stake's ‘effective resistance areas’ don't overlap — for a single rod, a C-stake at roughly 10 times the rod depth is the working rule, so 80 ft for an 8 ft rod, and 100+ ft doesn't hurt.

Run all three points in a straight line: electrode, potential stake, current stake. Keep the line away from and roughly perpendicular to buried metal — pipes, tracer wires, other grounding systems — which distort the voltage gradient. On a congested site, this is often the hardest part; walk the utility locates before choosing your line.

Drive the stakes firm and wet them if the surface is powder-dry — the stakes themselves need tolerable contact resistance for the tester to drive its current. Most testers will flag high stake resistance; fix it with depth and water, not by ignoring the warning.

4. Run the 62% method — and verify with the curve

With the current stake fixed, theory puts the true resistance reading at the point where the potential stake sits at 62% of the distance from the electrode to the current stake. So for a 100 ft C-stake: read at 62 ft. That's the shortcut — one reading, done, and it's what most specs mean by ‘the 62% method.’

Trust it only after you verify the plateau. Take readings with the P-stake at roughly 52%, 62%, and 72% of the distance. If the three readings agree within a few percent, the 62% value is your resistance — you're on the flat part of the fall-of-potential curve. If they climb steadily, your current stake is too close and the resistance areas overlap: move C farther out and repeat. A full curve — P-stake walked out in 10% increments and plotted — is the gold standard for a spec deliverable and takes twenty minutes.

Record actual distances, readings, tester model, and the date's soil conditions. A resistance number without its geometry and weather is a number nobody can reproduce.

5. Know when the clamp-on tester is legitimate

The clamp-on ground tester induces a voltage in a loop and measures the resulting current — no stakes, no disconnection, one-handed. The catch is in the word loop: it only reads correctly when the electrode under test is part of a multi-grounded system that provides a low-resistance return path — like a pole ground on a multi-grounded utility neutral, or one rod among many on a large bonded site. The instrument reads the series combination of your electrode and everything else in parallel; when ‘everything else’ is many electrodes, their parallel contribution is small and the reading approximates your electrode.

Where the clamp-on lies to you: a single isolated electrode (no loop — no valid reading), a rod whose only return path is the thing you're trying to measure, or any reading taken through a bonding jumper that bypasses the electrode. It also can't produce the engineering-grade curve a spec deliverable wants. Use it for maintenance surveys and spot checks on distributed systems; use fall-of-potential for acceptance testing and anything contractual.

If a clamp-on reading and a stake test disagree, believe the stake test.

6. Interpret desert numbers and improve a poor electrode

Soil resistivity rules everything, and ours is poor: dry caliche and decomposed granite run orders of magnitude higher resistivity than moist loam. Expect single rods in Tucson to blow past 25 ohms routinely, expect readings to improve dramatically for a few weeks after monsoon rain, and record the season with every test — a 15-ohm August reading and a 60-ohm May reading can be the same electrode telling the truth twice.

To genuinely lower resistance: length beats diameter (deeper rods reach more stable, moister soil — coupled sectional rods driven 16 or 24 ft work where 8 ft won't), multiple rods spaced at least their own length apart, conductive backfill or ground-enhancement material in drilled holes, or a concrete-encased electrode — the Ufer ground in every new footing is the desert's best electrode and the reason 250.50 makes you use it when it's there. Chemical rods are the specialty option where a spec demands single-digit ohms in hostile soil.

What doesn't work: driving a second rod 2 ft from the first (overlapping resistance areas waste most of the second rod — 6 ft is the code minimum and a rod-length apart is better), pouring salt water on the ground before the acceptance test (it reads well for a week and then lies for a decade), and hoping.

Does the NEC require 25 ohms for a grounding system?

No. The 25-ohm figure applies only to a single rod, pipe, or plate electrode — and the alternative to meeting it is simply adding a second electrode. Once supplemented, the NEC imposes no resistance target at all. Lower targets (like 5 ohms) come from specs and industry standards, not the code.

Why do most contractors drive two rods instead of testing?

Economics. A second rod costs a few minutes and satisfies 250.53(A)(2) with no test, no tester, and no documentation. A proper fall-of-potential test costs an isolated electrode, stakes, and time. Testing happens when a spec demands a number.

How far apart do the test stakes go?

Rule of thumb: current stake at 10× the electrode depth (80–100 ft for an 8 ft rod), potential stake at 62% of that distance, all in a straight line away from buried metal. Larger electrode systems (rings, grids) need proportionally more distance — IEEE 81 covers the geometry.

Can I test without disconnecting the electrode?

Not meaningfully with stakes — you'd measure every bonded path in parallel. A clamp-on tester works without disconnection but only on multi-grounded systems where a valid return loop exists. For acceptance testing of a specific electrode, isolate it (de-energized) and use fall-of-potential.

What resistance should I expect from one 8 ft rod in Tucson soil?

Often well above 25 ohms — dry high-resistivity soil is the norm, and triple-digit readings from a single rod in caliche are not surprising. Deeper coupled rods, multiple spaced rods, enhancement backfill, or leaning on the Ufer electrode are the realistic paths to low numbers here.

Does a better ground electrode make breakers trip faster?

No — that's the most common grounding myth. Fault clearing depends on the low-impedance metallic path back to the source (the EGC and bonding), per 250.4(A)(5). The earth electrode's job is lightning, surge, and voltage stabilization. Both matter; they are different jobs.