We've found the offending electric fence that BT weren't allowed to identify for us. There's no doubt about it, even to the fact that their new mains powered energiser was installed in November, right when the problems started. The lady was very helpful, wasn't aware that it interfered with broadband but now I mentioned it she said for the last month or so it's been really slow giving her problems with Facebook. So perfectly open to working with us to see if we can resolve things.
The fence is a single strand of plastic line with steel conductors, run at chest height on stand off insulators on fence posts. There are two gateways with expanding spring type openings. Mains powered energiser in the stable, lead out cable from there underground and coming up at a gatepost in an adaptable box from which the fence conductor is connected. Not sure what cable type is used for the underground run. There's no phone line at the stable so the energiser and it's earth are around 200m from the telephone cable at the roadside. The fence runs along the road for several hundred yard, above the 50 pair phone line.
First walk around there wasn't anything that glaring. The fence tested full power at all points, and the earth rod gave no reading on the tester at all. Nothing looked like it was shorting. So all I have is points of possible concern. (1) joints in the fence line made by knots rather than joiners. One of those was clicking when I walked round so I've fixed that one (and that seems to have had some effect). There are a few more that weren't showing any sign of problems (2) The earth rod isn't that long, in fact the guy who installed it had some concerns but hadn't fitted a longer one yet as it seems to work OK (3) the gateways, one of which is directly over the phone cable. These join with steel hooks with various degrees of rust.
Any ideas as to what sort of faults I should be looking for? I must say that if my concern was only the effectiveness of the fence, I can't see anything that would worry me.
The attached graph shows the general error rate, you can see three short periods when the fence was off, and what looks like a lower error rate after I fixed the clicking joint, not yet confirmed because I've seen short episodes like that before. My thoughts at the moment are that if this continues, it might be worth going back and fixing all the knots in the same way whether clicking or not.
Thanks in advance,