Insulation valves

Keith Wilson - electrical engineer

It’s unlikely that many of today’s engineers will have heard of insulation valves, although they may have unknowingly encountered them in their work. They are, however, discussed at some length in an appendix to the seminal paper on insulation resistance that was presented to the IEE in 1913 by Sidney Evershed, the man who invented the first practical insulation tester and trade marked the word “Megger”.

But what are insulation valves? In fact, they’re not physical objects; Evershed used this slightly strange terminology to describe an effect he and many others had observed when performing insulation resistance tests with DC voltages – they found that the resistance values obtained depended on the polarity of the applied voltage.

As an example, Evershed quotes a report from an engineer in Australia who found that the insulation resistance of a new electric lighting installation was 1.5 MΩ when measured with one polarity, but 9.0 MΩ when the polarity was reversed. Evershed states that he had no trouble replicating this effect “in every particular”. Indeed, he goes on to say that “porcelain switches of inferior quality have always acted like valves.”

This last statement gives a clue to Evershed’s somewhat curious terminology. Valves – or as our transatlantic cousins would have it, vacuum tubes – had first appeared on the scene in 1904, when J A Fleming invented the diode. This was just nine years before the paper was presented, so Evershed probably considered himself to be right up to date by referring to “insulation valves”. After all diode valves, just like his “insulation valves”, conducted better in one direction than the other.

Evershed, of course, wasn’t content simply to note the insulation valve effect; he wanted to understand why it occurred. To find out, he carried out an extensive series of tests. There’s no room here for the full details, but the conclusions are interesting. 

In the main part of his paper, Evershed had proposed that conduction in insulating materials is almost entirely due to the presence of moisture, either on the surface of the insulator or within its mass where, he suggested, it accumulated in “leakage channels” that allowed the passage of current.

His explanation for the insulation valve effect was that these channels become clogged by the chemical action accompanying electrolysis when a DC voltage is applied to an insulator for any period of time, and result is that the insulator conducts preferentially in one direction.

He backs up his contention by noting that even if a new electrical installation shows no valve effect on initial testing, the valve effect is almost always apparent if testing is repeated after the installation has been in use (with a DC supply) for a while, as this will have allowed time for electrolysis to take place.

In truth, Evershed’s explanation sounds a little dubious, although there’s no doubt that it was at the frontiers of knowledge about insulation resistance at the time the paper was presented. Irrespective of the exact mechanism of the insulation valve effect, however it’s interesting to speculate whether it could be replicated today with modern instruments.

Out of curiosity, some quick-and-dirty tests were performed. Since the paper says that the effect is specifically related to porcelain insulating materials, two specimens were procured. One was a switch with a good quality glazed porcelain base with no noticeable imperfections in the glazing, the other a ceiling rose that had once been properly glazed but had acquired much damage to the glazing over its long life.

Initial tests on the specimens, in a dry room, were uninformative. Tested at 500 V with a modern Megger MIT485 instrument, both measured > 50 GΩ irrespective of polarity. After 48 hours in a humid box, however, it was a different story. After a drying out period of just a few minutes to allow surface moisture to evaporate, the switch still measured > 50 GΩ in both directions. The ceiling rose, however, measured 47 MΩ in one direction and 18 MΩ in the other.

The measurements on the ceiling rose were repeatable and, although they weren’t completely steady, they drifted only slowly when the test voltage was applied continuously for two minutes in each direction. The insulation valve effect confirmed? Possibly, although the tests were far from comprehensive.

Assuming, however, that it was the same effect seen by Evershed, a key question remains. What determines the direction in which the insulator will conduct preferentially? After all, the ceiling rose tested probably hasn’t seen a DC supply – or indeed a supply of any sort – for decade. Surely its “memory” can’t be that good!