During an over-voltage withstand test, equipment in which the conductors are misplaced or the insulation is contaminated or undercured may fail, or it may suffer only undetected deterioration by discharges, e.g. a subsequent increase in dielectric losses in oil-impregnated paper or the propagation of discharge channels partly through or over the surface of the insulation. It is shown that such damage is quite probable during a.c. tests, whereas d.c. and impulse tests are more likely to cause either negligible damage or complete breakdown. If impulse tests can be applied for routine quality control, they are preferable to d.c. tests as they more closely simulate surge conditions which may occur in service. If neither impulse nor d.c. tests are appropriate to the particular equipment, a test with a few cycles of low-frequency (e.g. 0.1 c/s) voltage would be preferable to the usual power-frequency test. To detect very local defects which might cause thermal breakdown in service it is proposed that the impulse (or equivalent d.c. or 0.1 c/s) test should be made with insulation at the normal operating temperature.Over-voltage withstand tests of short duration give no assurance against deterioration and eventual breakdown in service by electrical discharges or electrochemical processes. It is therefore recommended that design tests should include time-to-breakdown tests and appropriate non-destructive tests in order to determine the magnitude of discharges and degree of ionic contamination which can be tolerated for a required service performance. Tests to assess resistance to breakdown by discharges should be made at the maximum operating temperature; often such tests can be accelerated by applying a higher frequency than that used in service. Life tests should be made at at least three stress levels to permit extrapolation to the service stress. Life tests to assess resistance to electrochemical deterioration may be similarly accelerated by raising the temperature.
Read full abstract
Jan 1, 1967
Setsuo Fukuda 
