I am a bit confused with this one. In the lab, I have two Netgear 10/100 hubs (used to be one, but same thing happened). The feed from the router interface is about 10 feet to the hubs. So, in the lab, all the machines on 50 foot or less of CAT-5 and CAT-5e cables (just the ones from Computer Support, ready made, that I can sneak out the door). They could all be 10 foot in the lab... but I use what I've got. So all the machines here work fine (most using Intel 100B 10/100s, one 3com and a couple of Linksys 10/100 cards) in half-duplex 100BaseT (what they are supposed to do).

But, the problem is with the two boxes that are down the hall in my office. The longest "ready made" cables here are 50 foot, so I made the cables from the spools of CAT-5, then the CAT-5e, we have. It isn't that far from the hubs, but they WILL NOT work on 100BaseT! They will negotiate to 100, but no packets will pass. If I force them down to 10BaseT... they work fine. I've replaced the cables a few times (thinking I messed up the RJ-45), but still the same. There is no way I could be hitting the 200 meter limit for 100 or the 100 meter limit for 10.

So what gives?? Could I just be messing up the wire layout in the RJ-45s *every* time?

I don't mention this at all to the "network" guys (the guys that lay out all the cabling for the building) for fear of ridicule. So, maybe you can all save me from being laughed at. Or maybe I really have a strange situation here.

Thanks. (Oh, BTW, to stay on topic... the machines in my office are Linux boxes... and the RJ-45 crimper sits on the bookshelf right on top of some Linux books too!!)

-Brian

is the cable in a false ceilling? florescent lights cause lots o' problems with throughput

try this pinout on the RJ's :

1 brown 2 brown/white 3 green/white 4 white/blue 5 blue/white 6 white/green 7 orange 8 orange/white

No, not near any field generators (obvious ones at least). Anything special about this color order? Please don't tell me it is to please the Ethernet Gods. ;-)

the colors are unimportant as long as receive data + (pin3) and receive data - (pin6) are on the same color pair. normal short jumpers run rd+ on white/blue and rd- on orange. thats fine for very short lengths, but not long runs. to get even further into the theory, the reason each color pair is twisted around itself(and there actually are spec's on how many twists per meter)is that a tiny magnetic field generated when a signal is passing helps lower the crosstalk between the transmit and receive pairs. running the signal on half of two different color pairs doesn't allow the mag field to be generated because (from above)white/blue and orange are not twisted around each other, no mag field=more crosstalk=less throughput. but yeah, chalk it up to the ethernet gods

rshaw speaks the truth. I once, in my early career (about 6 months ago), cut a 100ft cable lining up the colors just so they would match, not caring about the color code. The cable never worked for who I made it for... worked fine for me; and then one day I plugged it into my Sparc, which hated it, regardless of mode.

I cut the heads, re-crimped them by the color code, the ethernet gods are appeased, and even the Sparc is happy.

Cheers,

Finegan

Ok, I will give that a shot and report back. (So little time, that I'll stand for the boxes to be at 10MB!!). It's a toss up between the fields-turns/meter theory and the Ethernet Gods story. If it works... I'll stick with the Gods on this one. Hang on for that report... posted here.... in the not too distant future! It could proove that we all need to make regular Macintosh sacrifices at the altar of Ethernet!

Brian