Indeed - and if it came to a choice, I'd rather forego the insulation test in favour of a thorough inspection. Eyes, ears fingertips and nose can be very good at detecting things meters can miss.

I think a few years ago there was a discussion about which test was the most important, in a sort of Desert Island Discs way ' if you could only take one meter with you ?' sort of way. We decide that an R2 test was the most useful, as any other fault would become  obvious in a safe way on switch on.

I think for PAT or EICR then similar considerations often apply.

We measure conductor to conductor IR and not surface to conductor. (Granted metal containment systems will be earthed.)


In a new installation, we expect hundreds of MΩ. For periodic testing, it is not so much the absolute value, but any change which is of interest. (Assuming that previous reports are available.) If we get a result around 1 MΩ now, it has probably been falling and may well continue to fall. In any test regime, an acceptable value must take account of any expected change before the next test. It follows that the pass value will be above the minimum safe value. IMHO, a margin of a factor of 10 would not be unreasonable.


Or Zoomup may be correct - it was simply a reasonable engineering judgment.

We measure live to live with the circuit disconnected at the distribution board/ consumer unit.


But when we measure lives to earth the circuit protective conductor should be connected to the main earth terminal with all the installation earthing in place to hopefully pick up any earth fault between the lives and anything they may come into contact with that is or maybe close to earth potential. 


If you see a YouTube video of a circuit being insulation tested with the lives and CPC all disconnected it is wrong, the CPC should be terminated into the earth bar with the tester clip attached to the earth bar or similar.


Andy Betteridge

I have done the inspection and testing in a three storey house today, on the job sheet it said to reinstall the hall light as well.


The tenant said the hall light had been taken down because the shower was leaking and water was running down through the light fitting, the tenant said he had an electric shock whilst he was in the shower, so an electrician had taken the light fitting down.


Then he went on to say he had also had an electric shock when touched the dishwasher and another electrician had replaced the plug socket in the back of the cupboard under the sink, though it looks original to me.


There is a 30 mA DP RCD upfront of the old Square D consumer unit with BS3871 MCBs, so I turned the RCD off along with the CU main switch and did a 250 volt insulation test of the whole installation lives to earth  with everything connected, 0.00 Mohm absolutely no resistance at all.


So I tested the RCD with the CU main switch off, tested perfectly both on time tests and the ramp test, tripping at 25 mA.


I now had a C1 to start the list of observations and I knew the tenant had been shocked whilst using the shower and dishwasher. So it was a case of either find send fix it or advise that the installation needed to be isolated. A couple of hours later after breaking the installation down  I was replacing the socket circuit cable drops behind the central heating boiler that the plumber had drilled whilst installing the boiler some years ago creating an 8 ohm resistance fault between the neutral and CPC on one leg of the ring.


It is now all sorted out, but I am pondering two things, first that the healthy RCD didn’t trip despite a zero insulation test result, secondly that appears that two electricians went to investigate two different electric shock events at different times and presumably neither of them did a global insulation test.


Andy Betteridge 

"...two electricians went to investigate two different electric shock events at different times and presumably neither of them did a global insulation test."


They were unprofessional in their attitude and work, and just wanted to get out quickly. You are professional, dedicated, conscientious and competent Andy, and a credit to our industry.


Z.

The NE fault may not trip an RCD if the fault path impedance is high enough. Here you are putting the 8 or so ohms of the fault across one of the two coils in the RCD ( and that reading is such a funny no.  it may well be quite variable from day to day if it is a rusty screw not really touching copper cores very well that creaks and groans as the boiler heats up).


If the RCD coil is say 0.1 ohms and the fault path shunting it is indeed 8 ohms, then the current will split in the ratio of 80:1 - the imbalance seen by the RCD  may not have reached 25mA if the flat was never heavily loaded, for example  20A was running down the live sensing coil of the RCD and 79/80 of 20A down the neutral sensing coil would give an imbalance of 20A/80 - which is about 25mA.


I have seen RCDs that seem to fire off in a load dependent way because of this sort of situation - and maybe if it only happened once or twice the occupant reset it and kept quiet having assumed they had overloaded something a bit.


As you say, it seems that both sparks before you were keen to get in and out a bit quick and do the minimum of investigation. Or maybe to be charitable, perhaps on that day the rusty screw was not in such good contact. However, after a reported shock, I'd hope you might expect a bit more rigour than the minimum.


(and that is why we check at 500V if we can, not with cheap general purpose meter with a 9V battery - which will find dead shorts and opens, but maybe not a scratchy contact where a few extra volts break the oxide barriers and reveal things like this.)