Both tests put a voltage across insulation and both are described in the same paragraph of most test reports, which is why they are so often treated as interchangeable. They are not. Hipot is a stress test: it looks for a breakdown that either happens or does not. Leakage current is a measurement: it quantifies what flows through and around the insulation at normal operating voltage.
What each test can and cannot see
A hipot test at 3,000 V AC for 60 s is looking for a defect that cannot survive the stress: a creepage path that is too short, a barrier with a pinhole, a wire pinched against a chassis. It is a good production screen because it is fast, binary, and it fails loudly.
It is blind to anything that is supposed to conduct. A Y-capacitor across the barrier passes the hipot without complaint, because a capacitor is not a defect. That same capacitor is what makes the leakage current measurement interesting: at 253 V and 50 Hz, a 4.7 nF capacitor passes about 0.37 mA, which is a meaningful fraction of a 0.75 mA limit. Add a second capacitor, or a longer cable with more capacitance to earth, and a product that is perfectly safe exceeds the limit.
| Test | Voltage | What it finds | What it misses |
| Dielectric strength (hipot) | 1,000 to 3,000 V AC or DC | Breakdown, pinched wire, short creepage | Excessive capacitance, resistive leakage below trip |
| Leakage current | Operating voltage, 1.06 × rated | Capacitive and resistive current to earth | A barrier that only fails above operating voltage |
| Insulation resistance | 500 V DC typically | Moisture, contamination, degraded material | Capacitive current entirely |
| Touch current | Operating voltage, weighted network | What a person would feel | Internal leakage that does not reach accessible parts |
Why the hipot current never settles to zero
On a product with any capacitance across the barrier, the hipot current starts high and decays. The initial spike is the charging current, and the steady value is the resistive leakage plus the capacitive current at the test frequency. On a 10 nF barrier at 3,000 V and 50 Hz, the capacitive current alone is about 9.4 mA, which is above the trip setting of many bench testers. The tester trips, the operator reports a failure, and the product is fine.
Two legitimate ways out: raise the trip limit with a documented rationale, or test with DC. A DC hipot removes the capacitive component entirely after charging, and the steady current is then purely resistive, which is what the test was trying to measure in the first place.
The measurement that needs a network
Leakage current is not measured with a multimeter. The value that matters is the current a person would experience, and a person is not a 10 MΩ resistor. IEC 60990 defines the weighting networks: a resistive network for perception and let-go, a network with a parallel capacitor for reaction, and a network that approximates the impedance of the human body at the relevant frequencies.
Using the wrong network changes the answer by a factor of two or more at 50 Hz, and by much more when harmonics are present. A switch-mode supply with a 20 kHz component can read 0.3 mA on a resistive network and 1.1 mA on the weighted one, because the capacitor in the network presents a low impedance at high frequency.
Practical sequence
- Insulation resistance first, at 500 V DC, with the product unpowered and discharged.
- Leakage current at operating voltage, at the worst-case supply, with the product running at rated load.
- Hipot last, because it is the only test that can damage the sample, and a damaged sample must not be used for anything afterwards.
- Record the actual leakage value, not just pass or fail; a product at 0.72 mA against a 0.75 mA limit is a design warning even when it passes.
Standards referenced: IEC 60990 (measurement of touch current and protective conductor current), IEC 61010-1 clause 6.3 (dielectric strength) and clause 6.4 (limits for accessible parts), IEC 60335-1 clause 13 and 16.
