Why Does My Multimeter Show a Normal Voltage but My Circuit Still Doesn't Work?

Why the meter can't see the fault

On its DC voltage function, a typical digital multimeter has an input impedance of around 10 MΩ. Across a 12 V source that lets through only about 1.2 µA (12 V ÷ 10,000,000 Ω). Voltage is lost across a resistance only in proportion to the current through it, so with 1.2 µA flowing even a badly corroded 1 Ω joint drops about 1.2 µV, far below anything the display shows. A connection that has degraded to 1,000 Ω still drops only about 1.2 mV. The meter honestly reports 12.00 V, and the reading tells you the source is there, not that it can push current through the path to the load.

That is the whole trick behind this symptom. A voltage reading taken with nothing drawing current, or taken upstream of the bad spot, proves very little about a power path. A real load can draw amps instead of microamps, which is a million times more current, and every bit of hidden resistance suddenly costs volts. The fix is to make the circuit draw its normal current while you measure. This is different from the meter loading a high-impedance circuit, which is covered in why a multimeter shows the wrong voltage, and from coupled voltage on a dead wire, covered in why a multimeter shows voltage on a disconnected circuit.

Causes at a glance

Worked example: one corroded joint

Take a 12 V source feeding a 6 Ω load, which on a healthy circuit draws 2 A (12 ÷ 6) and dissipates 24 W (12 × 2). Now add a corroded connection of 1 Ω in series, a plausible figure for a green, crusty terminal or a crimp that has loosened. The total resistance becomes 7 Ω, so the current falls to about 1.71 A (12 ÷ 7). The joint drops 1.71 V (1.71 × 1) and the load receives only about 10.29 V (1.71 × 6). The load's power falls to about 17.6 W, roughly 73% of normal, which is enough to make a motor stall, a relay chatter or an electronic device brown out and reset.

Meanwhile the joint itself dissipates about 2.9 W (1.71² × 1), all of it concentrated in a contact patch a few millimeters across, which is why bad connections run hot, oxidize faster and get worse over time. Smaller numbers still matter: just 0.5 Ω in series with a 2 A load drops 1 V (2 × 0.5) and turns 2 W into heat (2² × 0.5). The chart shows the voltage the 6 Ω load receives from a 12 V source as the joint resistance rises, alongside the reading you would get with the load disconnected.

Calculated values: load voltage = 12 × 6 ÷ (6 + joint resistance), ignoring the source's own resistance. Real circuits add wire and source resistance on top.

Why does my multimeter show voltage but the device won't work?

When the meter shows voltage and the device still won't work, the reading has proved that a potential exists between your two probe points and nothing more. The worked example above puts numbers on the gap: the meter draws about 1.2 µA to read 12.00 V, while the 6 Ω load needs about 2 A, so the path has to carry well over a million times more current than your measurement drew. Five situations produce this symptom, and each leaves a different fingerprint once you take a second reading.

• Voltage with no current capacity behind it. A wire that is broken or switched off upstream can still show tens of volts on a 10 MΩ input because it is coupled to live conductors nearby. That voltage can't supply even a milliamp, so the device sees nothing. It collapses on a LoZ setting, as explained in why a multimeter shows voltage on a disconnected circuit.
• A high-resistance connection. The 1 Ω joint in the worked example is the mild version. A back-stabbed outlet terminal, a cracked solder joint or a crimp holding by one strand can reach many ohms; at that point the loaded voltage falls so far that the device won't even start, yet the unloaded reading stays perfect.
• Voltage drop under load. A weak battery or undersized supply reads its nominal value at rest and sags as soon as current flows. Step 5 shows the arithmetic for a source with internal resistance.
• An open ground or open neutral. The return path carries the same current as the supply side. If you measure from the positive terminal to chassis rather than to the device's own ground terminal, an open ground stays invisible: the reading is full voltage while the device has none across it. On household outlets, an open neutral typically leaves hot to ground normal while hot to neutral reads low, wanders, or drops to nothing when something is plugged in.
• The device itself. If full voltage reaches the device's terminals while it is switched on, look inside it: an internal fuse, a tripped thermal cutout, an open door or lid interlock, a failed internal power supply or a dead control board.

On a 120 V receptacle, three readings sort most of these out. The table shows typical patterns, not exact values; the readings vary with the meter, the wiring and what else is on the circuit.

The fastest field version of the loaded test is to plug in a lamp or other simple load that you know works and measure again while it is on, or to switch the meter to LoZ if it has one. Either one forces the circuit to supply real current, which is the thing the bare voltage reading never asked of it. If the receptacle passes and the device still won't work, the problem has moved from the wiring to the device, and the steps below, especially 7 and 8, apply. An outlet that trips or won't reset is a separate fault, covered in how to test a GFCI outlet that won't reset.

Multimeter shows voltage but the appliance won't work

If your multimeter shows about 120 V at the outlet but the appliance won't work, either the outlet has voltage without the ability to carry current, or the appliance has an open circuit inside it. The first group covers a loose backstab or splice, an open neutral, a tripped GFCI further up the circuit and coupled "ghost" voltage on a dead wire; the second covers the cord, the power switch, an internal fuse or a thermal fuse. One swap test tells you which group you are in, and it needs no tools: plug a lamp or phone charger you know works into the same outlet, and plug the appliance into an outlet on a different circuit.

Loose backstab or splice: voltage without current

Many receptacles are wired through push-in "backstab" holes, and each outlet passes power on to the next. When one of those connections loosens or oxidizes, it still passes the microamps your meter draws, so every outlet after it reads a healthy 120 V with nothing plugged in. Plug in a 1,500 W space heater, which has a hot resistance of about 9.6 Ω (120² ÷ 1,500) and draws 12.5 A, and the joint has to carry real current. The chart works the arithmetic for a few joint resistances: at 0.5 Ω the heater gets about 114 V and the joint turns roughly 71 W into heat in a space smaller than a fingernail. Higher resistances leave the appliance starved or dead, and a joint dissipating that much power doesn't last; it chars and eventually fails open, which is when the whole chain of outlets dies.

Calculated values for a 9.6 Ω heater on a 120 V supply: current = 120 ÷ (9.6 + joint), heater voltage = current × 9.6, joint heat = current² × joint resistance. Joint heat for the 0.1, 0.5, 2 and 10 Ω cases is about 15, 71, 214 and 375 W. Wire and supply resistance are ignored, and a real heater's resistance shifts a little as it warms. Illustrative, not measured.

Open neutral: hot reads fine, nothing runs

The neutral carries the appliance's current back to the panel, so a break there stops the appliance just as surely as a broken hot. It also fools a meter. Hot to ground reads a normal 120 V. Hot to neutral can read close to 120 V as well, because the broken neutral wire is pulled up toward the hot through anything else plugged into the circuit, and a 10 MΩ meter needs almost no current to see that. Plug in a lamp and the reading collapses. If neutral to ground also shows a large voltage, the neutral is open upstream. On a multiwire branch circuit, where two hots share one neutral, losing that neutral puts the loads on the two halves in series across 240 V, so one side can run well above 120 V and the other far below it. That is an electrician job, and it can destroy electronics, so unplug sensitive equipment on the affected outlets.

Ghost voltage, switched outlets and tripped GFCIs

A dead conductor that runs alongside live ones in the same cable picks up voltage by capacitive coupling, and a high-impedance meter can show tens of volts on it, sometimes enough to look like a weak supply. That "ghost" voltage can't power anything and drops to near zero on a low-impedance setting, such as the LoZ mode on the Fluke 117, or with a lamp plugged in. The full explanation is in why a multimeter shows voltage on a disconnected circuit.

Two common reasons a wire is dead in the first place are easy to miss. One half of a receptacle is often controlled by a wall switch, so the top socket can be live while the bottom one, the one the lamp or appliance is in, is switched off. And a GFCI protects every outlet wired to its LOAD terminals, which may be in another room: a tripped GFCI in a bathroom, garage, kitchen or outdoors can kill an ordinary-looking outlet somewhere else. Press Reset on every GFCI you can find before testing further. If one won't reset, see testing a GFCI that won't reset; if it keeps tripping when the appliance starts, see why a GFCI outlet keeps tripping. When whole rooms are out with no tripped breaker, start with why half a house can lose power with the breakers looking fine.

Inside the appliance: cord, switch and fuses

If the outlet passes the lamp test and the appliance still does nothing anywhere, the fault is in the appliance. Unplug it, turn its own switch on, and measure resistance across the two flat prongs of the plug. A simple resistive appliance should read close to its working resistance, which you can estimate from the rating plate: R = V² ÷ P, so a 1,500 W heater is about 9.6 Ω and a 900 W toaster about 16 Ω (120² ÷ 900). OL means the path is open somewhere: a conductor broken inside the cord near the plug or strain relief, the power switch, an internal fuse, a thermal fuse or thermal cutout, a door or lid interlock, or the element itself. Appliances with motors or electronic controls won't give a simple reading this way, but OL across the plug with the switch on still points at the cord, switch or a fuse.

From there, test the pieces one at a time with the appliance unplugged: the thermal fuse, any cartridge fuse, the heating element and the cord, flexing it while watching continuity as described in how to find a broken wire. A blown thermal fuse is a symptom, so look for the blocked lint duct, clogged filter or failed fan that made the appliance overheat before replacing it with an identical rating. Sealed, mains-powered electronics and anything with a large capacitor should go to a qualified repair technician.

The steps below work through the circuit in the order that finds the fault fastest: first prove whether voltage is actually missing at the load, then find which half of the loop is losing it, then look at things a DC voltage reading cannot show.

1. 01

   Load on

   Measure while it runs

   →

2. 02

   At the load

   Its own terminals

   →

3. 03

   Split the loop

   Positive, then ground

   →

4. 04

   Source

   Does it sag?

   →

5. 05

   Beyond DC

   Signals, ripple, load

1Measure under load

Reconnect everything, switch the circuit on and make it try to do its job: crank the engine, turn on the pump, press the button that should light the strip. Then measure. If the device has its own switch, the reading before and after it turns on is the most useful comparison you can make. A healthy circuit loses only a small fraction of a volt when the load starts; a circuit with a hidden fault loses volts, sometimes all of them.

If the load runs only for an instant or draws current in bursts, the meter may not update fast enough to show the dip. That's where Min/Max recording, covered in step 6, earns its place. For batteries and bench supplies there are dedicated walkthroughs: how to test a 12V battery under load and how to check a 12V power supply under load.

2Measure at the load's terminals

A reading at the battery or the supply output tells you about the source, not about what reaches the device. Put the probes on the load's own power terminals, or back-probe its connector, so that every wire and joint in the path sits between the source and your meter. Wire resistance alone can matter on long runs: 18 AWG copper is roughly 6.4 Ω per 1,000 ft, so a 20 ft run out and 20 ft back is about 0.26 Ω, and at 2 A that costs about 0.5 V (2 × 0.26) before any connector is involved.

If the voltage at the load is close to the source voltage under load, the wiring is fine and you can skip to step 7. If it is noticeably lower, the difference has been lost in the wiring and connections, and steps 3 and 4 find where. The same idea applies to single components; see how to measure voltage across a resistor for the probe placement.

3Split the drop: positive side

With the load still running, put the red probe on the source's positive terminal and the black probe on the load's positive terminal. The meter now reads only the voltage lost along the positive path. Then move the probes inward one connection at a time: across a fuse, across a switch, across each connector and each length of wire. The section with the large reading is the fault. A good connector or fuse typically drops only millivolts to a few tenths of a volt depending on current; anything that stands out from its neighbors deserves a closer look.

This direct reading is more sensitive than subtracting two large numbers, because the meter can use a low range. A bad joint shows as, say, 1.7 V instead of the difference between 12.0 V and 10.3 V. On vehicles the same technique is applied to starter and battery cables in how to check voltage drop on a car, which also lists commonly used guideline limits.

4Check the ground side

Half of every circuit is the return path, and it fails just as often. Put the red probe on the load's ground terminal and the black probe on the source's negative terminal, with the circuit running. A good ground reads close to zero, typically tens of millivolts at modest currents. A corroded chassis bolt or a loose ground strap shows a reading that grows with current, and the load is short by exactly that amount.

Grounds fool people because a ground point measured to chassis with the circuit off reads 0 V, which looks perfect. Only a loaded test shows the drop. On cars, a single bad engine or body ground can make several unrelated circuits misbehave at once; see how to test a car ground wire for the specific points to check.

5Check the source under load

Sometimes the wiring is fine and the source itself can't deliver. A battery behaves like a voltage source with an internal resistance in series. Healthy batteries keep that resistance low; a sulfated, aged or deeply discharged one does not. As an illustration, a source with 0.5 Ω of effective internal resistance that reads 12.6 V at rest loses 2 V when it delivers 4 A (4 × 0.5), and sits at 10.6 V under load. Measure directly across the battery posts or the supply's output terminals while the load runs: if the voltage sags there, the problem is the source, not the wiring. The car-specific version of this is in testing a car battery that reads 12 volts but won't start.

Bench supplies and adapters fail differently. A bench supply set to a 1 A current limit feeding a load that wants 2 A at 12 V (6 Ω) switches into constant-current mode and holds 1 A, so the output falls to about 6 V (1 × 6). Many supplies show a CC light when this happens. Wall adapters and USB chargers that are overloaded may sag, fold back or shut down and restart in a cycle, which can look like a fault in the device. Check the current rating against what the load actually draws, including start-up surges.

6Catch intermittent dropouts

If the circuit works sometimes, a spot reading will usually show normal voltage because the fault isn't happening while you look. Connect the meter at the load, start Min/Max recording if the meter has it, and let the circuit run while you flex the harness, tap connectors, wiggle the plug and warm suspect areas. When the fault occurs, the minimum reading records how low the voltage fell even though the display has already recovered.

Recording speed matters. Standard Min/Max on many handheld meters captures changes lasting on the order of 100 ms or more (check your manual), while peak modes on some meters catch much shorter events. For dropouts shorter than that, an oscilloscope with a single-shot trigger or persistence display is the better tool; see using oscilloscope persistence to find intermittent glitches. The multimeter-only method is laid out in how to find an intermittent electrical fault, and readings that jump around without a fault are covered in what causes unstable multimeter readings.

7Signals, enable lines and ripple

If the supply holds its voltage at the load under load, power is not the problem. Many devices also need a control input to do anything: a motor driver needs its enable pin pulled high, an LED driver needs a PWM signal above 0% duty, a relay module needs its trigger input, and a digital board needs a valid clock or data stream. A multimeter on DC volts shows only the average of a changing signal, so a 5 V PWM signal at 50% duty reads about 2.5 V, and a data line that is toggling can read an unremarkable in-between value. A steady reading cannot tell you whether the signal is switching correctly. An oscilloscope can; see how to measure PWM duty cycle with an oscilloscope.

Ripple is the other thing a DC reading hides. A DC rail with a large AC component can average to the correct voltage while dipping low enough on every cycle to reset a microcontroller or make audio gear hum. As a quick screen, switch the meter to AC volts and measure the same DC rail: on many meters the AC function blocks the DC level and shows the AC part on its own, though some meters measure AC+DC, so check the manual. Treat the number as approximate, because handheld meters specify AC accuracy over a limited frequency band and can under-read the high-frequency ripple of a switching supply. Ripple at 120 Hz on a 60 Hz linear supply points to filter capacitors; a proper measurement needs a scope, as described in how to measure ripple voltage with an oscilloscope.

8Suspect the load itself

When the correct voltage is present at the load's terminals while it is switched on, the control signals are right and the supply is clean, the device itself has failed. Useful checks are measuring the current it draws (a reading of zero suggests an open winding, filament or internal fuse; a reading far above its rating suggests a short or a seized motor), measuring its resistance with the power disconnected, and swapping in a known-good part.

Compare against the maker's rated current where one is published. A 12 V, 24 W lamp should draw about 2 A (24 ÷ 12) when running. A motor briefly draws several times its running current at start-up, so judge it once it is turning. If the device is mains-powered or sealed, stop at this point and hand it to a qualified repair technician.

Meters that help

Any working meter can run the loaded tests above. Min/Max recording and a fast analog bar graph make sags and dropouts easier to catch. These picks come from our verified multimeter list.

Meters for catching voltage that sags under load

Fluke 88V

Min/Max/Average recording with 250 µs Peak Min/Max for dropouts, CAT III 1000V / CAT IV 600V. Average-responding, not True RMS.

Check price on AmazonRead the review →

Fluke 87V

True RMS, a 19,999-count high-resolution mode and an analog bar graph that tracks changing readings. Hands-on Tested.

Check price on AmazonRead the review →

KAIWEETS HT118A

Budget True RMS, 6,000 counts, with Min/Max and an analog bar. Maker's page and manual disagree on max voltage.

Check price on AmazonRead the review →

As an Amazon Associate I earn from qualifying purchases. Prices and availability are shown on Amazon.