This is certainly not a new discovery, but it's good to re-visit certain things once in a while. One of the problems with using any foamed or micro-porous Teflon dielectric coax cable (like LMR and other modern, stiff, low-loss cable) is the hazard of having the center conductor push through the relatively soft dielectric when under temperature extremes or with rough handling. Here is an example.
I have numerous 20' lengths of LMR-400 cable, terminated with crimp-on N connectors that were commercially manufactured. You used to be able to get these things surplus on EBay, complete with test report and sealed in plastic for $10-15. They were good for jumpers and anything else that didn't move too much. If you had too much radial motion, the connectors would break off after a while. A couple weeks ago, after sorting through a jumble of RF cables used for various things (roving, portable op's, ect), I came across one 20 foot LMR-400 unit that had a loose crimp connector. I went ahead and cut off the old connector and replaced it with a nice, new-out-of-the-bag Andrew two-piece compression design and tossed it aside for test. This morning, I got around to testing it (no use having a network analyzer siting on the test bench if you don't use it to test things...) and found it had God-awful return loss (bad VSWR). Of course, I figured the connector had been installed improperly, but after disassembly, the connector looked fine. Further testing showed no change to the return loss when actively pushing and pulling on the partially disassembled end. Hmmm - maybe it's the OTHER end, so I punched up the TDR routine on the analyzer to try and find the fault. Long story short, the fault ended up being about 400mm BEHIND the connector I replaced. You can see why in this picture.
You can see that the center conductor has moved almost an entire diameter into the dielectric, but was not yet shorting. The interesting part is there was NO OBVIOUS damage to the coax outer cover - no crimp marks or crush marks - only a few scratches. I assume it was smashed under some large weight or run over by a vehicle or, maybe stepped on or subjected to less-then-minimum bend radius after laying in the hot desert sun. Sometimes, you can see a mark from a zip-tie on the cable, but not here. Whatever it was, the return loss was approaching 5 dB several places between 30 and 1500 MHz (and greater then 20 dB in the 432 MHz band).
Cutting off 18" and re-attaching the same connector gave me a cable assembly with >25 dB return loss through 1500 MHz, with insertion loss of approx 0.2 dB at 50 MHz and 0.5 dB at 432. The take-away from this lesson is to always remember how unforgiving some high-performance, low loss RF cables can be when abused - even if there is no sign. Pretty much the only way to track down this type of fault is with TDR techniques. BTW, differentiating between the reference plane and the internal conductor attachment point of an N connector using a HP network analyzer is not a problem - there's plenty of room in there.
Keith J Morehouse
Calmesa Partners G.P.