Alan B said:

+/- 2% plus 4 digits, so effectively +/-5V,

If it's a 3-digit display, 500 V range, the accuracy in volts for FSD is given by [(%) + (digit value)], so full-scale deflection accuracy in volts is actually +/- [(2*500/100)+(4)] = +/- 14 V. For a reading of 250 V, the accuracy in volts is +/- 9 V.

It is similar in accuracy overall to the Megger MFT1700 series on voltage measuring range ... ±3% ± 1 V ±2 digits would give +/- 10.5 V.

For analysis of the supply, other instruments, such as a Power Quality Analyzer, are far more appropriate.

It's also worth remembering that the laboratory (intrinsic) accuracy of MFTs are much better than the quoted service accuracy in the field ... the more you pay for instruments, the more accurate they will be in different conditions in the field.

Alan B said:

+/- 2% plus 4 digits, so effectively +/-5V,

If it's a 3-digit display, 500 V range, the accuracy in volts for FSD is given by [(%) + (digit value)], so full-scale deflection accuracy in volts is actually +/- [(2*500/100)+(4)] = +/- 14 V. For a reading of 250 V, the accuracy in volts is +/- 9 V.

It is similar in accuracy overall to the Megger MFT1700 series on voltage measuring range ... ±3% ± 1 V ±2 digits would give +/- 10.5 V.

For analysis of the supply, other instruments, such as a Power Quality Analyzer, are far more appropriate.

It's also worth remembering that the laboratory (intrinsic) accuracy of MFTs are much better than the quoted service accuracy in the field ... the more you pay for instruments, the more accurate they will be in different conditions in the field.

If you really are seeing swing of 40V, and there is not 100A of load being keyed on and off at the house between the two states, then there are a number of possible nasty things happening - but some we can eliminate with a bit of local knowledge.
If you need more evidence to get the DNO out of bed, then perhaps look if there is any evidence for  some of these issues.

(and while one may argue the toss about the absolute meter accuracy,  a swing of that magnitude is really notable, even if both voltages would read higher or lower on a better instrument.) An L-N PSSC test may be revealing - have you done that ? Is this a 3 phase supply - I presume not but if it is, then looking at the other phases may  be instructive.

Voltage drop at the transformer - this should be no more than about 5% if the substation or transformer is not horribly overloaded, but it could be,  or there may be a burnt  contact on one of the LV side fuses.

Pole transformers where the paint has burnt off, or the oil has dripped are getting rarer, but are still seen occasionally and are a clear overload marker.

Voltage drop on the line from transformer to the house - is the house at the end of a long thin street with the tranformer at the other end, and what loads (houses or businesses) does it share with.  (sharing such a supply with a welding shop can be bouncy, or somewhere with big loads like kilns or industrial ovens)

Normally the DNO network designs for no more than about 10% voltage drop. But then the loads sort of grow...

poor network neutral - prohibited by law, but not so uncommon in older aluminium clad buried mains. Varying N-E voltages and large currents in bonded services like water mains and things are a symptom.

Faulty cut out or cut out fuse - smell of burning, dripping tar or plastic.

then the only one that is not a DNO issue,

Fault on load side - loose contacts scratchy main switch etc, all likely to be obvious by the heat produced, and the fact the house lights dim when the kettle goes on...
test the volts  as near origin as is sensible.

Mike

mapj1 said:

the house lights dim when the kettle goes on..

maybe a bit less diagnostic these days than it used to be - as filament lamps are replaced by electronically driven LEDs or CFLs.

A high current loop test can sometimes be revealing too.

   - Andy.