So if I have R1+R2 and R1 + RN original readings I have a point of reference for when it's 2nd fixed. I admit that it would have to be a bit worse than 0.1 ohms out for me to spend long on it though.
That is exactly the point I was trying to get across. The usual test for a radial is to link L & CPC at the CU and then measure R1+R2 at the end socket (which you know, if just installed) and ideally at all sockets to proves polarity at each one (or if you don't know the end, or it might be a "tree" instead of "linear" radial so has more than one end so proving CPC continuity at one end is not telling you about the other branches).
What you get is proof of polarity and CPC continuity, and if the R1+R2 is sane you know it is probably OK and it would disconnect on a fault within time (even before a live Zs test).
But it is
very unlikely you would spot a small resistive connection as you would not know the expected R1+R2 to that accuracy as you probably don't know the installed length to within a couple of meters, nor any tolerance on the cable (5%?).
You also do the same for L & N so get R1+RN which is excellent practice! But more than most do as it is not usually required
Whereas the end-end RFC continuity measurement
specifies testing all three and to look for discrepancies between the measured resistances and the expected cable ratios.
On EICR's it's interesting that just because an RFC is easy to identify faults with we routinely do the tests even though it's quite a bit of work, but we don't tend to do an equivalent bit of work on radial circuits with a nulled wander lead checking R1 = RN and R2 is proportional. Maybe the wander lead should come out more....
Using a nulled wander lead is the simplest way to do the same, but a much greater pain in the posterior to actually do compared to the RFC end-end at the CU. More so if the site and any on-going use makes using a wander lead risky or impractical.
You could link L, N, and CPC at the CU and then measure all 3 pairs at the radial end to get R1+R2, R1+RN, and R2+RN to find out the same, but computing the individual conductor values is more work and most folk don't relish solving 3 simultaneous equations before lunch time.
Of course to spot an error you don't need to solve them, as you could check the expected ratios of the pairs of conductor CSA (e.g. if all 3 are the same, say singles in trunking, then all 3 measured pairs should be identical).
I've not yet seen anyone else test both sides of a double docket when doing a fig8 test, so maybe there is a 50% weighing on this one?!
In the above I meant checking sockets on both the RFC and the radial to find which was bad.
Though I do check both halves of double sockets as I have seen problems before, and I have had the luxury of not being under the same job time-pressure that professional sparks are to get it in and completed at the going rates.
Really the question was to look at the issue of testing using the "figure of 8" for the RFC. As every socket then should measure the
same R1+R2 value it becomes easy to spot one that is high, either a spur (yeuch!) or a bad socket.
Doing the same check on each socket on the radial is harder to identify as, again, you probably don't know the cable length socket-socket to get an accurate idea of what each one's R1+R2 progression should be. (Same for doing R1+RN to check neutral for poor connections).
TL;DR Illustrating that the RFC test may be more work, but it gives much better fault coverage.
