kwe@bu-cs.BU.EDU (kwe@bu-it.bu.edu (Kent W. England)) (04/20/89)
In article <475@vice2utc.chalmers.se> bernerus@utc.chalmers.se (Christer Bernerus) writes: >We're having serious problems using Isolan Fan-Out's with our RT's. >Symptoms are lots of collisions and/or corrupted packets. > > A number of IBM RT 6151's, using an Ungermann-Bass baseband > adapter and running both AOS and AIX. The RT's are connected > via an Isolan 1130-1 Fan-Out box, which has been connected to > the Ether using various xcvr's. > > We also used a Cabletron ST-500 xcvr with many fancy LED's. This unit > > A test with the RT's connected directly to the cable, without the > Isolan fan-out didn't show any bad behaviour, also a test with > fan-out's from another vendor worked fine. We also tested 5 different > Isolan fan-out's and they all made things go bad again. These sorts of things are very difficult to sort out and I have not directly experienced this problem, but I think I've heard of it before. From the tests you describe, it doesn't sound like the classic heartbeat problem or loose cables. I would guess that it has to do with transceiver cable grounding. Those U-B cards don't implement IEEE grounding. Some fan-outs don't mind, maybe the Isolans do. I suggest that you try to find out from BICC about the grounding and also check with U-B. Another thing you might try is to check your xcvr cables and see what grounding they use. Maybe they can be changed out and solve your problem. You might try to use some other cables to better match one end or the other, but I'm on thin ice at this point. Maybe Jim Warner or Rob Warnock could say, they know a lot about this stuff. If anyone can tell me exactly what a card maker ought to do to be compatible with v2 and IEEE cables and xcvrs wrt grounding, I could pass that along. Some U-B people were asking me about that a while back. Anyway, check xcvr cable grounding mismatches and let me know how good this guess is. :-) Good luck. --Kent