Thanks for all the input.

On meter accuracy the KT65DL measures voltage to  4 digits, example 230.5, so my max error is 4.6V + 0.4V, total 5V.

Interesting to see that the meter linked above is much less accurate for AC voltages than DC, I guess it's difficult to compensate for voltage drops across diodes, assuming meters use a bridge rectifier.

Meter appears to be more accurate than that in real life, the error becomes a concern when we are on edge of acceptable, if I am seeing 216V, is it actually 216 or 214 for example.
So the voltage variations I was seeing were occurring around about 5.30 in the evening. It's in a village that is fairly long and strung out, with a mixture of overhead and underground lines. Looking at the SSEN network map, the substation is about 4km away, marked as 33/11KV, which I assume is input and output. Also states fully constrained by thermal on the 33KV input line, no constraint down stream. There is also a solar farm covering a few hectares presumably on that substation. Looks like the substation can take a 50% reverse feed.
I think the transformer for the village is about 1km away and the house in question is close to the end of line; with a couple of houses another 0.5 km outside the village, may be on a different feed. Main part of the village is spread out over about 1Km. Haven't managed to find a diagram of the local network

Customer thinks he has actually solved his charging issues with a new 12V battery in the car, but he was reporting issues with both cars, so not sure. I am also seeing quite big swings in my house, 5 or 6 houses down from them. Now looking at this out of interest as much as anything else and wondering if there is actually a bit of an issue with the neutral. I am going to take some measurements of neutral voltage over the weekend to see what I get, will use a different meter as the KT65DL doesn't measure voltages below 25V.

I shall put my money on a network problem rather than a metering one.

That said, I fully appreciate GK's comment about the type of meter which is required. The DNOs must have them.

The real question must be where the burden of proof lies. Does a customer have to show that a supply is problematic, or does the DNO have to show that it is sound? At first glance, ESQCR 2002 is unhelpful.

when you measure NE, do so between supply N relative to a test electrode in the garden - which could be as simple as a garden fork and a  croc clip or some random old (non insulated of course) screwdriver  or similar for voltage readings - its not an electrode to carry massive fault currents after all. Measuring NE between network N and E tells you very little if there is any PME type wiring on the network it tends to hide such problems, as CPC and N then move together. True terra-firma earth does not move and then the full Neutral voltage lift is evident..
I'd be slightly surprised if you are more than 500m from your 11kV to LV transformer, but occasionally it happens.  The location of the 33kV 11KV step-down means your 11kV line is about 5km long, which is long but not very long for an 11kV line.

The one metre per volt rule for economic line length is a rule of thumb, not hard and fast but it is a good start as to what sort of distances to expect or consider 'unusually long'.

Mike.

Just managed to get hold of the local supply network diagram. As you described it looks like there is actually a lot of 11KV cables going around the village, with what looks like small pole mounted transformers dotted all over the place. Probably no more than 150m from a transformer to any house.
But I assume that the pole mounted or small ground mounted transformers are fixed ratio without any other technology. Therefore voltage drop in the 11KV network could still be an issue.

I will be measuring earth/neutral voltage reference to something conductive stuck in the ground.

I learnt quite a bit this evening digging around in diagrams etc, not sure how much use it will be day to day but interesting anyway.

well poor joints and voltage drop on the 11kV side has an effect of course, but because of the lovely way transformers change voltage and current in inverse ratio, you find you need a lot of ohms on the 11kV side to introduce the same effect as a resistance on the 400V side - well (11000/400) squared as many ohms to be precise (*)- so half an ohm on the LV side is more like a few hundred  ohms on the 11kV side, which is why one can afford HV lines of many km of course, and why there have to be quite so many LV transformers.

So I'd expect an LV  fault first.

Mike

* and I was deliberately vague about are we comparing at voltage drops phase to phase or phase to neutral, as I know that the transformers are delta primary on the 11kV  and star wired 240V phase to neutral so perhaps  you may want to consider (11000/230) squared instead. In any case, it  is common to ignore the drops on the HV side, as a first approximation, and the smoothing effect of many loads pm the same 11kV branch mean that steps due to any given load are less serious.

Alan B said:

Interesting to see that the meter linked above is much less accurate for AC voltages than DC, I guess it's difficult to compensate for voltage drops across diodes, assuming meters use a bridge rectifier.

I suspect it's more about, unless it filters for 50Hz, the impact of the actual waveshape. It doesn't look to me as if MFTs filter 50Hz, but I may be wrong, but if they don't then the actual waveshape - which by the time you get to the property is vanishingly unlikely to be a pure sine wave - will have a big effect on the accuracy. Even with the regs on switch mode convertors (including car chargers) there is a huge amount of rubbish on domestic 50Hz circuits these days. (Even more in industrial circuits.)

I wouldn't expect a modern (in the last 30 years!) professional meter to use diodes, I'd expect it to calculate the rms voltage from the waveshape - although many will use the average of each half cycle.

However re the comments of "it actually seems to be more accurate than that", yes very probably - in fact I'd be worried if it wasn't. 4% is the point at which you could complain to the manufacturers that it hasn't met their spec, they'd have designed it to stay well within that so they don't get loads of returns. And as Graham suggests, they have to allow for, particularly, temperature extremes which throw both the voltage reference and the components in the analogue front end. It would be nice if they quoted accuracy between, say, 10C and 30C as well as over the full temp spec of the meter, but that won't sell many more meters to do so!

Thanks,

Andy

Alan B said:

Interesting to see that the meter linked above is much less accurate for AC voltages than DC, I guess it's difficult to compensate for voltage drops across diodes, assuming meters use a bridge rectifier.

I suspect it's more about, unless it filters for 50Hz, the impact of the actual waveshape. It doesn't look to me as if MFTs filter 50Hz, but I may be wrong, but if they don't then the actual waveshape - which by the time you get to the property is vanishingly unlikely to be a pure sine wave - will have a big effect on the accuracy. Even with the regs on switch mode convertors (including car chargers) there is a huge amount of rubbish on domestic 50Hz circuits these days. (Even more in industrial circuits.)

I wouldn't expect a modern (in the last 30 years!) professional meter to use diodes, I'd expect it to calculate the rms voltage from the waveshape - although many will use the average of each half cycle.

However re the comments of "it actually seems to be more accurate than that", yes very probably - in fact I'd be worried if it wasn't. 4% is the point at which you could complain to the manufacturers that it hasn't met their spec, they'd have designed it to stay well within that so they don't get loads of returns. And as Graham suggests, they have to allow for, particularly, temperature extremes which throw both the voltage reference and the components in the analogue front end. It would be nice if they quoted accuracy between, say, 10C and 30C as well as over the full temp spec of the meter, but that won't sell many more meters to do so!

Thanks,

Andy