The 22 V Neutral: Diagnosing a High-Resistance Neutral
A two-storey residence has a bedroom receptacle that reads 121 V line-to-ground, 99 V line-to-neutral, and 22 V neutral-to-ground while a 1 500 W space heater is plugged in and running. At the service panel, neutral-to-ground reads 0.3 V. The branch circuit is 3.5 mm² copper on a 20 A breaker, roughly 25 m from the panel, and the readings fall to 0.8 V neutral-to-ground when the heater is switched off. The receptacle is a back-stabbed device. Continuity of the equipment grounding conductor back to the panel is good at 0.2 Ω.
1. Interpret the three readings. What is the load current and what does the 22 V tell you quantitatively?
Read the worked answer
The three readings are self-consistent: 99 V + 22 V = 121 V. The load is seeing 99 V instead of 121 V, and the missing 22 V appears across the neutral path.
Load current: 1 500 W ÷ 99 V ≈ 15.2 A.
Resistance of the neutral path: 22 V ÷ 15.2 A ≈ 1.45 Ω.
Expected neutral resistance for 25 m of 3.5 mm² copper: 5.3 Ω/km × 0.025 km ≈ 0.13 Ω (about 2 V at this current).
So there is roughly 1.3 Ω of unintended resistance in the neutral - about eleven times the conductor's own resistance. The drop collapsing to 0.8 V with the load removed confirms it: no current, no drop, therefore the resistance is in series with the neutral path and is being exposed only under load. The good 0.2 Ω on the grounding conductor rules out a general bonding problem; the fault is specific to the neutral conductor or its terminations.
2. Why is this a shock hazard as well as a fire hazard?
Read the worked answer
Shock. The equipment grounding conductor is bonded to the neutral at the service, so the grounding system at the panel sits at 0.3 V - essentially earth. But the 22 V neutral drop raises every bonded metal surface downstream of the bad joint (appliance chassis, the receptacle's grounding contacts, metallic conduit) to roughly 22 V above true earth. A person touching that chassis while standing on a slab, touching a water pipe, or touching a separately grounded appliance completes the circuit through their body. It will not trip a breaker, because the current is far below any trip threshold - this is exactly the class of fault that GFCIs exist to catch and that ordinary overcurrent protection cannot.
Fire. The 1.3 Ω defect dissipates P = I²R ≈ (15.2)² × 1.3 ≈ 300 W in a connection point. That is a soldering-iron's worth of heat inside a box, continuously, for as long as the heater runs. Back-stabbed devices are a common location for exactly this failure because the contact area is small and the spring relaxes with thermal cycling.
3. What is your diagnostic procedure and what do you do about the 0.8 V residual reading?
Read the worked answer
- Measure under load, at each point along the neutral: panel neutral bus → any junction → the receptacle. The voltage steps up at the defect.
- Kill the circuit and inspect the back-stabbed terminations; convert to screw terminals or pigtail to the device with a listed connector.
- Check the neutral bar termination in the panel and every splice in the run for corrosion and torque.
- Re-measure loaded; expect the neutral-to-ground at the receptacle to come back to 1–2 V at this current.
On the 0.8 V residual with the load off: do not dismiss it, but do not over-read it either. With the heater switched off the circuit is not truly unloaded - other outlets on the same circuit and any connected standby loads still draw current through the shared neutral, and meter noise contributes. The right test is to isolate the circuit completely (breaker off, nothing plugged in) and measure again: near-zero is normal. A persistent reading with the circuit truly dead points to a shared-neutral or multiwire arrangement, in which case the neutral is carrying return current from another circuit and the whole circuit topology needs to be mapped before you touch it.