Evaluation of the performance of several object types for electrical treeing experiments

A comparison of the efficiency between various test object types for studying the development of electrical trees in polymers is presented. Two types of newly developed wire-plane objects are investigated and compared with two versions of the traditionally used needle-needle object. The two needle-needle objects are prepared according to ASTM standard and act as references, whereas the alternative objects rely on the use of an ultra-thin tungsten wire (10 and 20 μm) for providing the highly divergent electric stress. In one of these object types the wire extends from a semiconducting tab embedded in the tested material whereas the embedded wire is extended and connected externally by means of a copper tape in the other type. The comparison is made using cross linked polyethylene (XLPE) for the testing purposes. The wire type objects provide some promising advantages, including parallel formation of several electrical trees and an exposure of a larger material volume. Further advantages include the simplification of both the manufacturing and the measuring procedures. Among the newly developed test objects, the one without the semiconducting tab is considered most advantageous as it allows for more accurate treeing initiation measurements at a lower voltage level. A disadvantage with the wire electrode is its tendency to form kinks during object preparation, which may lead to difficulties in correctly estimating the field strength locally as well put strain on the surrounding material. However the multitude of electric trees formed in each object allows for trees growing at such imperfections to be discarded in following data analyses. Although finding a statistical method that makes good use of all the data poses a challenge, a suggested approach is presented.