You isolate a circuit, lock it off, and put a meter across two conductors you know should be dead \u2014 and it reads 80 V. Most technicians read that as “still live somewhere” and go looking for a second feed that doesn’t exist, or worse, read it as confirmation that a component downstream is faulty when it never was. By the end of this article you’ll know exactly why a high-impedance meter does this, how to tell a ghost from a genuine fault in under thirty seconds with low-impedance mode, and how to prove a circuit is actually dead rather than assume it from one reading.<\/p>\n\n\n\n
Run two conductors together in the same conduit or trunking and you’ve built a capacitor, whether you meant to or not. Over a 20 m run, the capacitance between two adjacent conductors is typically a few tens to a few hundred picofarads. If one conductor is live at 230 V and the other is disconnected at both ends, that capacitance couples a small current into the floating conductor.<\/p>\n\n\n\n
The numbers explain why the effect is so large. At 50 Hz, 100 pF has a reactance of roughly 32 M\u03a9. That reactance forms a voltage divider with whatever is measuring the floating conductor. A standard digital multimeter on AC volts has an input impedance of around 10 M\u03a9 \u2014 high on purpose, so it draws almost no current and doesn’t disturb the circuit it’s testing. Put that 10 M\u03a9 meter across 32 M\u03a9 of coupling reactance fed from 230 V, and the divider lets a large fraction of the supply voltage appear on the meter. It is entirely normal to read anywhere from 30 V to 150 V on a conductor that is genuinely, safely dead.<\/p>\n\n\n\n
The current behind a ghost voltage reading is in the microamp range. That is nowhere near enough to do work: it cannot light a 2 W neon indicator, it cannot hold a contactor coil in, and it cannot deliver a shock. A relay coil, a solenoid, a motor winding \u2014 anything with a real, low-impedance path to earth \u2014 collapses that voltage to almost nothing the instant it’s connected, because a few hundred ohms or less swamps the microamp source completely.<\/p>\n\n\n\n
This is the single fact worth holding onto: a voltage with no capacity to do work is not a live circuit.<\/strong> But that is a reason to prove it, never a reason to assume it. You confirm a ghost reading with a second measurement, every time \u2014 you don’t wave a meter near a wire, get a number under 100 V, and decide from memory that it must be capacitive coupling.<\/p>\n\n\n\n Most service-grade meters carry a low-impedance mode, marked LoZ<\/strong> or Low-Z<\/strong>. Switching to it drops the meter’s own input impedance to roughly 3 k\u03a9. That loads the coupling capacitance heavily \u2014 the microamp source cannot sustain any meaningful voltage across 3 k\u03a9, and a ghost reading collapses.<\/p>\n\n\n\n The rule that makes the two modes useful together: a real voltage barely moves between high-Z and LoZ; a ghost voltage collapses to under about 5 V.<\/strong> An 80 V reading in high-Z that drops to 1\u20132 V in LoZ is a dead conductor. A reading of 230 V in high-Z and 228 V in LoZ is a live conductor, and it will kill you exactly as readily as any other live conductor.<\/p>\n\n\n\n Two things will damage equipment if you get them wrong. First, never use LoZ on electronic control circuits, thermistor inputs, or 0\u201310 V signal lines \u2014 the 3 k\u03a9 load pulls the signal down hard enough to damage the driving device, not just misread it. Second, LoZ does not replace a proving unit. It tells you whether a reading is real or induced; it does not confirm your meter is working correctly in the first place.<\/p>\n\n\n\n Safe isolation under the OHS Act 85 of 1993 is a sequence, not a single reading, and skipping steps is how technicians get hurt on circuits they believed were already proven dead.<\/p>\n\n\n\n A live-work safety note belongs here: any time you are taking a reading on an energised panel rather than a proven-dead one, wear the correct PPE for the incident energy present, use a meter rated CAT III 600 V or better for plant panels and switchgear, and never treat a plausible-looking reading as a substitute for isolation when isolation is the safer option. Ghost voltage is a reason to look closer at a dead circuit, never a reason to skip proving one that should be live.<\/p>\n\n\n\n A walk-in freezer compressor would not start. A previous technician measured 228 V at one contactor output terminal (T1) and 86 V at the other (T2), both against earth, both in the meter’s default high-impedance mode, and concluded the compressor itself must be open-circuit \u2014 a full supply on one leg and a partial reading on the other looked, at a glance, like voltage was reaching the load.<\/p>\n\n\n\n Switching both readings to LoZ told a different story: T1 held at 227 V, barely moving \u2014 a genuinely live terminal. T2 collapsed to 2.4 V \u2014 a ghost, coupled entirely from the adjacent live conductor in the same gland plate. T2 was dead, not partially live.<\/p>\n\n\n\n With the contactor energised, a further reading across the line and load terminals of the T2 pole itself \u2014 not to earth, but across the closed contact \u2014 read 226 V. Voltage across a closed switch means it isn’t actually closed electrically, whatever the armature is doing mechanically. Isolated, locked off, and proven dead by the same three-step sequence above, resistance across that pole read over-range; the other pole read 0.2 \u03a9. One contact face had burnt away.<\/p>\n\n\n\n Insulation resistance on the compressor windings came back over 500 M\u03a9, and the three winding resistances summed correctly with no shorted turns \u2014 the compressor was, and always had been, sound. The fault was a burnt contactor pole, a fraction of the cost of the replacement compressor that had already been quoted. The wrong diagnosis traced back to exactly one high-impedance reading taken on what turned out to be a dead conductor. If you want the full pole-by-pole method for confirming a contactor before you condemn it, see how to test a contactor properly<\/a> \u2014 the same LoZ discipline applies before you touch a single pole. The same proving-dead sequence matters just as much before you handle a capacitor, which can hold a dangerous stored charge long after the supply is removed; see how to test a capacitor on an air conditioner<\/a> for the discharge procedure.<\/p>\n\n\n\nThe Cure: Low-Impedance Mode<\/h2>\n\n\n\n
Reading in high-Z<\/th> Reading in LoZ<\/th> Verdict<\/th><\/tr><\/thead> 80 V<\/td> 1.2 V<\/td> Dead \u2014 ghost voltage from capacitive coupling<\/td><\/tr> 45 V<\/td> 3.8 V<\/td> Dead \u2014 ghost voltage<\/td><\/tr> 230 V<\/td> 228 V<\/td> Live \u2014 proceed as live<\/td><\/tr> 210 V<\/td> 195 V<\/td> Live \u2014 some loading, still real supply<\/td><\/tr> 86 V<\/td> 2.4 V<\/td> Dead \u2014 ghost voltage on a disconnected pole<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n How to Prove a Circuit Genuinely Dead<\/h2>\n\n\n\n
Worked Example: The Contactor Pole That Looked Like a Dead Compressor<\/h2>\n\n\n\n
Final Thoughts<\/h2>\n\n\n\n