A dielectric test on a dry sample and a dielectric test on the same sample after a damp heat soak are not two versions of one measurement. They are two measurements of different things. The dry test measures the insulation. The soaked test measures the insulation plus whatever path the moisture has found across its surface, and the second number is the one that predicts a failure in a bathroom rather than in a laboratory.
That is why IEC 60884-1 clause 17 does not simply say "apply 2000 V". It says the sample spends 48 hours at 40 degrees C and 93 percent relative humidity first, and it gives one minute to get from the chamber to the test. Everything difficult about the clause is in those two sentences.
The condition and where it is defined
The soak itself comes from IEC 60068-2-78, damp heat, steady state: 40 degrees C plus or minus 2 K and 93 percent relative humidity plus 2 or minus 3 percentage points, held for 48 hours unless the product standard says otherwise. The 48 hours is not a warm-up. It is the time a moulded thermoplastic needs to take up the moisture it is going to take up at that humidity, and a 24 hour soak reaches a different answer on a polyamide part and nearly the same answer on a filled PBT part.
Note the asymmetry in the humidity tolerance. Plus 2 and minus 3 is not a rounding; it reflects that the water content of the air changes faster above saturation than below it, and a chamber that overshoots to 96 percent is further from the condition than one that undershoots to 90.
The one-minute window is the whole test
Clause 17 requires the insulation resistance and the dielectric strength to be measured while the sample is still in the chamber, or within one minute of taking it out. The reason is that the conductive path the soak created is a film on the surface, and a film in a dry room evaporates from its edges inward. A sample that sits on the bench for ten minutes has a dry rim and a smaller effective surface path, and the reading improves for a reason that has nothing to do with the product.
In practice this drives the layout of the test cell. The high voltage lead has to reach into the chamber, or the chamber has to be next to the test bench with the sample still dripping. A lab that runs the soak in one building and the dielectric test in another is not running clause 17, however good its records look.
What the numbers should be
| Check | Condition | Limit |
| Insulation resistance | 500 V DC, read after the value is stable | At least 5 MOhm |
| Dielectric strength, live parts to body | 2000 V, 1 min, rated voltage above 130 V | No breakdown, no flashover |
| Dielectric strength, live parts to accessible metal | 3000 V, 1 min, rated voltage above 130 V | No breakdown, no flashover |
| Chamber condition | 40 ± 2 °C, 93 +2 / -3 % RH | 48 h |
Two of those entries punish impatience more than bad product. The insulation resistance reading rises as the sample charges: at 500 V a clean dry part may read above 100 MOhm in a second, while a soaked part climbs from a few megaohms at 15 seconds to a stable value at 60 seconds and beyond. Reading at 15 seconds and calling it the result is how a marginal product gets a good number, and reading at 60 seconds on a product that was still climbing is how a good product gets a bad one. The honest method is to watch the display until it stops moving, and to record both the time and the value.
The dielectric test has the same shape in the other direction. Ramping to 2000 V in under a second produces a capacitive inrush current that trips the trip point on a perfectly sound sample; a controlled ramp over 5 to 10 seconds does not. Most modern hipot testers offer both, and the fast one is usually the default.
The chamber is part of the measurement
Four chamber faults account for most of the disagreement between laboratories running this clause.
- Water quality. A humidifier fed with tap water deposits scale on the heating element and, worse, aerosols dissolved solids onto the samples. Deionised water below about 5 microsiemens per centimetre removes both problems.
- Air movement. A fan aimed at the samples dries the surface faster than the humidity control can replace it. The samples should sit in still, saturated air.
- Drip from the ceiling. Any condensate that falls on a sample adds a water path that the condition does not intend. A chamber with a flat roof and no drip rail will do this on the first cold sample of the day.
- Loading. Samples stacked against each other share a moisture film and produce a result that belongs to the stack rather than to any one sample. Keep at least 100 mm between samples and between a sample and the wall.
Recovery comes after, not before
IEC 60068-1 defines a recovery period of 1 to 2 hours at 15 to 35 degrees C and 45 to 75 percent relative humidity, and it is easy to apply it in the wrong place. Recovery is for tests that measure a property of the material after the conditioning has finished. Clause 17 measures a property that exists only while the moisture is present, so the electrical tests come first and recovery comes after, if it comes at all.
A report that lists a 2 hour recovery before the dielectric test has documented a sequence that no longer tests what the clause describes. The number in it may still be passing, and it is passing for the wrong reason.
Why the soak finds the products that pass dry
The failures this clause catches are almost never bulk insulation failures. They are surface paths: a moulding flash between two pins, a seam where two halves of a housing meet without a gasket, a coating that pulls back from an edge during cure, a terminal recess that holds a bead of water against a live part. None of these show up on a dry sample, because dry air at 2000 V per millimetre is an excellent insulator and the path is only a millimetre long.
Add a water film and the same millimetre carries a leakage current that the trip point sees. That is why a product can pass a 3000 V dry test and fail a 2000 V soaked one, and why the second result is the one that matters.
The chambers and hipot testers used for this sequence are listed under electrical safety test equipment; the full range starts at the product index.
