clive@drutx.ATT.COM (Clive Steward) (06/01/88)
From article <32990@linus.UUCP>, by bwk@mitre-bedford.ARPA (Barry W. Kort): > Roy Smith asks about "Netwatch" programs on Appletalk. Kinetics > > We suspect echos and standing waves are the culprit, caused by > impedance mismatches, but we don't have hard evidence. The > phenomenon would be similar to TV ghosts caused by poor antenna > cabling. It would be nice to have a CRT display which would reveal > echoes and standing waves so that cabling irregularities could be > more easily diagnosed and corrected. Right now it's a black art. > > --Barry Kort The answer's been around for about 30 years, actually. Call your local Tektronix sales reps and ask about Time Domain Reflectometry (TDR) analyzers. There are general purpose cable-testing tools intended just for situations like yours. In brief summary, one ships a fast transition to a (generalized) cable, then views the pulse waveshapes coming back. The analyzers include graphical measurement tools to define impedance mismatches and locations. As I remember, on a low frequency coax cable, at least, one may find things accurately to inches over hundreds of feet; range is thousands of feet. This varies with the consistency and lossiness (hence pulse spreading) of your medium. Twisted pair range is probably similar, accuracy worse. If you cared, with expensive coax and short distances, you could locate at least to millimeters. Tek makes several analyzers (though I'm ~10 years out of date with this); some of them are targeted closely to what Ether/AppleTalk nets would like. Incidentally, don't feel bad for not knowing about this. I remember (~20 years ago, groan) being a fresh young spud in a (very capable) microwave design team. There, everyone was into Frequency Domain Analysis. Generally a good idea, since many of the devices themselves were relatively narrow band. But locating mismatches (here we're talking about quite small fractions of inches) _on_the_way_to directional couplers, filters, etc. was still quite possible -- of course the reflection from nearer bumps came back before the one from the relatively indecipherable part. Made myself a hero by presenting them with Polaroid pictures of the problem they'd spent weeks trying to locate. Ah, past glory. Wish it was as easy now. Clive Steward P.S. In fairness, I think HP makes some of these too, and they're nearly as nice people as Tektronix. But I didn't work there....