HOW-TO GUIDE
How to Check a PCB for a Short Circuit With a Multimeter
By the OhmSavvy Research Team ·
Never measure resistance on a powered board
- Resistance and continuity modes push a small current from the meter into the circuit. Any voltage already present corrupts the reading and, on mains-powered equipment, can damage the meter.
- Unplug the board, remove batteries, and confirm 0 V on the rails with the voltage range before switching to ohms. The bulk capacitor on a mains-powered board can hold a dangerous charge after unplugging.
- If the board is part of mains equipment and you are not trained to work on it, leave the primary side to a qualified repair technician.
Causes at a glance
| Symptom | Likely cause | How to check | Fix or who to call |
|---|---|---|---|
| Rail reads under 1 Ω both ways; lowest at one small cap | Cracked MLCC decoupling capacitor | Lowest reading at its pads; lift one end | Replace the capacitor |
| Low on input rail after a wrong adapter or reversed battery | Failed TVS or protection diode | Low in both polarities at the diode | Replace the diode; check the input |
| Lowest at an IC's supply pins, no cap lower | Internally damaged IC | Lift nearby caps first; compare with a good board | Replace the IC, or a repair specialist for BGAs |
| Short between two adjacent pins only | Solder bridge | Magnification; beep pin to pin | Flux and solder wick |
| Low reading after liquid spill or rework | Conductive debris or corrosion | Visual check under light and magnification | Clean with isopropyl alcohol and remeasure |
| Low in one polarity only | Diode junction conducting (not a fault) | Swap probes; compare readings | No fix needed; keep looking |
| Starts low then climbs to OL | Capacitors charging (not a fault) | Wait for the reading to settle | No fix needed |
1Power down and discharge the board
Disconnect every source: the adapter, any battery, USB cables and attached peripherals that could back-feed a rail. Large electrolytic capacitors can hold charge for a while after power is removed, so measure across them in DC volts and wait, or bleed them through a suitable resistor, until they read near zero. A 1 kΩ resistor across a 1,000 µF capacitor has a time constant of 1,000 × 0.001 = 1 second, so it discharges to under 1% in about five seconds. On boards with a mains input, the primary-side capacitor needs particular care.
2Measure each rail to ground
Find a ground reference such as a mounting hole, the negative terminal of the input connector or a shield can. Put the black probe there and the red probe on a rail, for example the output of a regulator or the positive side of a bulk capacitor. Note the value, then swap the probes and note it again. Repeat for every rail you can identify, and write the numbers down: a table of rails and readings is what you compare against a good board or the next measurement.
A reading that drifts upward is capacitance charging, not a fault. A reading that is low in one polarity but not the other usually means a diode junction is conducting. What you are looking for is a low value that holds steady with the probes either way round.
What counts as a short
Resistance from a rail to ground is the parallel combination of everything on that rail, so a busy rail reads lower than a quiet one even when every part is healthy. Hedged rules of thumb, which you should always check against a known-good board or the same rail on a working unit:
| Steady reading, both polarities | Usual meaning |
|---|---|
| Under 1 Ω | Dead short: a failed part or bridge across the rail |
| 1 Ω to about 10 Ω | Suspicious: compare with a good board before condemning a part |
| Tens to hundreds of Ω | Often normal for a core rail feeding large chips |
| kΩ and rising, or OL | Healthy rail with capacitors charging |
Low core-voltage rails on processors and graphics chips can read only a few ohms when healthy, so the comparison with a good board matters most there. For more on why the same rail can read differently depending on where and how you probe, see why resistance readings differ on a circuit board.
Why the continuity beep can mislead
Continuity mode beeps below a fixed threshold that varies by meter, typically somewhere in the tens of ohms. That is coarse for board work: a healthy rail feeding several chips can fall under it, and a large uncharged capacitor or an inductor gives a short chirp before the reading rises. Use the beeper to check a suspected solder bridge between two pins, but switch to the resistance range when you need to tell 0.2 Ω from 2 Ω. If the beeper itself behaves oddly, our note on an intermittent continuity beep covers the usual reasons.
3Locate the short by comparing readings
Copper traces have a small but real resistance, so the point physically closest to the fault reads slightly lower than points further away. Standard 1 oz copper is roughly 0.5 mΩ per square: a trace 0.25 mm wide and 10 mm long is 40 squares, or about 40 × 0.5 = 20 mΩ. Short the leads together, use the meter's relative (REL) mode to subtract the lead resistance, and pick the lowest manual range. Keep the ground probe in one place and walk the other probe across every capacitor pad, inductor and IC supply pin on the rail, pressing firmly and consistently on each pad.
Work methodically. Mark each reading on a photo of the board, and measure every pad twice, once going out and once coming back, so a single bad contact doesn't send you the wrong way. When the readings fall in one direction, keep following them until they start rising again: the minimum sits between the last falling reading and the first rising one. On a multilayer board the rail is often a copper plane rather than a trace, which spreads the current out and shrinks the differences further, and the gradient may lead to a via rather than a pad. In that case compare pads in a rough grid across the plane rather than along a line, and expect to finish the job with current injection.
Can your meter see the difference?
Resolution is the range divided by the count. A 6,000-count meter on its 600.0 Ω range steps in 600 ÷ 6,000 = 0.1 Ω, so the four readings above would all show 0.3 or 0.4 and the suspect would be invisible. A meter with 0.01 Ω resolution separates them, and a 1 mΩ range shows even small differences clearly. The last digit also wanders with probe pressure and the accuracy spec's digit term, so repeat each reading and trust a consistent pattern rather than a single number.
| Lowest ohms range | Resolution | Sees a 0.04 Ω step? |
|---|---|---|
| 600.0 Ω on a typical 6,000-count meter | 0.1 Ω | No |
| 220.00 Ω (UNI-T UT61E+) | 0.01 Ω | Yes, just |
| 600.00 Ω (EEVblog Brymen BM786) | 0.01 Ω | Yes, just |
| 50 Ω (Fluke 289) | 1 mΩ | Clearly |
4Lift the suspect and confirm
Once one part reads lowest, lift one of its ends with a soldering iron and measure the rail again. If the rail recovers, the lifted part was the short. If not, measure the lifted part on its own and keep searching. A suspect ceramic capacitor can be checked off the board with the method in how to check a ceramic capacitor with a multimeter, and soldering a surface-mount capacitor by hand covers fitting the replacement. If the cause turns out to be a bridge from rework, how to prevent solder bridges helps keep it from coming back.
01
Discharge
0 V on every rail
02
Rail to ground
Both polarities
03
Walk the rail
REL on, lowest range
04
Lift suspect
Rail recovers?
05
Replace
Then power on a limit
5When the ohms reading can't pinpoint it
On a rail with many parts in parallel, the differences between pads may be too small to resolve. The next step is current injection: connect a bench supply set to a low voltage with a modest current limit, so the short carries current but nothing else on the board turns on. Keep the voltage at or below the rail's normal level. With a 1 A limit into an illustrative 0.3 Ω short, the supply settles near 1 × 0.3 = 0.3 V, too low to power any chip, and the short dissipates I²R = 1² × 0.3 = 0.3 W. That small amount of heat, concentrated in one tiny part, is enough to find with a thermal camera, isopropyl alcohol that evaporates first on the hot spot, or a fingertip on a low-voltage board.
Lowest reading, then lift to prove it
Meters for short hunting
Any meter finds a dead short. Locating it on a crowded rail is where 0.01 Ω resolution and a relative mode earn their keep. These picks come from our verified multimeter list; see the best multimeters for electronics repair for the full comparison.
Meters that resolve small resistance differences
50 Ω range with 1 mΩ resolution using a 10 mA test current, plus relative mode to remove lead resistance.
60,000-count with 0.01 Ω resolution on the 600.00 Ω range, relative zero and BeepLit continuity under 100 µs.
22,000-count with 0.01 Ω resolution on the 220.00 Ω range and REL mode to zero the test leads.
As an Amazon Associate I earn from qualifying purchases. Prices and availability are shown on Amazon.
Frequently asked questions
What resistance counts as a short on a PCB power rail?
There is no single number, because a healthy rail feeding many chips can legitimately read tens or hundreds of ohms. Compare against a known-good board, the same rail on a sister channel, or the schematic. A steady reading well under one ohm in both probe polarities almost always means a dead short; a few ohms deserves a closer look before you condemn anything.
Why does the reading keep rising when I measure a rail?
The meter's test current is charging the decoupling and bulk capacitors on that rail. As they fill, the measured resistance climbs and may end at OL, which can take several seconds on a rail with hundreds of microfarads. That is normal behavior, not a fault. A real short settles quickly at a low value and stays there no matter how long you wait.
Should I test with the probes both ways round?
Yes. Diodes, protection devices and the substrate diodes inside ICs conduct in one direction only, so a rail can look low with the red probe on ground and normal with red on the rail. A true short reads the same low value in both polarities. Comparing the two readings is the quickest way to tell a conducting junction from a failed part.
Can I find a short without desoldering anything?
Often you can narrow it to one part by comparing readings along the copper with a meter that resolves 0.01 Ω, then confirm by lifting one end of the suspect part. If the rail reading jumps back up with that part lifted, you have found it. On rails with dozens of parallel capacitors, current injection and a heat check usually beats desoldering at random.
Is it safe to apply power to a board with a short?
Only through a bench supply with its current limit set low, at or below the rail's normal voltage. Connecting a shorted board to its normal adapter or battery can burn traces, damage the supply or, with lithium cells, start a fire. Current injection works because the limit keeps the power small, typically a watt or less, concentrated in the faulty part.
Why do I get different resistance readings at different points on the same rail?
Copper traces and planes have a small resistance, so points closer to the short read slightly lower than points further away. That is exactly what makes locating the fault possible. The differences are often only tens of milliohms, which is why you need a meter with 0.01 Ω resolution, a relative mode to zero the leads, and firm, consistent probe pressure on each pad.
