🚀 go-pugleaf

danhicks@millcomm.com writes:

> In <35e6319c.53011606@news.lakwod1.co.home.com>, blosser@home.com (Aaron Blosser) writes:
> >On Fri, 21 Aug 1998 11:07:04 -0500, Daniel Erskine
> ><netsup75@edwardjones.com> said:
> >
> >>> Using existing 75ohm coax as if
> >>> it was 50ohm coax (with F to BNC adapters) is asking for trouble.
> >
> >>  I'll inquire with my network staff for more research material, as
> >>I'm curious as to why 75ohm cable would be that much of a problem?!?
> >
> >I wonder as well.
> >
> >For short runs of cable, I highly doubt that using 50ohm vs. 75ohm would
> >make a difference.  That measurement is a measure of the cable's
> >impedance over some distance (1000 ft?), so for a short bit of maybe
> >50-100 ft, the difference *may* be neglible.
> >
> >Of course, this all depends on how senstive the receiving equipment is,
> >and I can't make any claims to know how it would really affect the
> >performance.
> >
> >The reason you *should* use properly rated cable is because with
> >impedance matching you can achieve maximum power output.
> 
> With 10Base2 proper impedance matching is critical to prevent reflections 
> -- which cause errors and false collisions.  However, for reasonably small
> nets you should be able to use 75 ohm cable **IF** you use 75 ohm 
> terminators.  The drivers and receivers should be able to tolerate the 
> difference in current and voltage.

This is not correct. Even if you use 75 ohm coax, you MUST terminate
with 50 ohm terminators. On coax the collision detection works on DC
signaling which is sourced as a current. If the DC resistance of the
cable is wrong (37.5 ohms vs. 25 ohms), collision detection breaks.

75 ohm cable will sort of work (even in long runs), but you may see
weird behavior due to standing waves and performance will probably be
sub-optimal. Try very hard to avoid mixed runs of 50 and 75 ohm cable
as each transition will create reflections and the chances of getting
reasonable performance will approach nil.
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest Orlando Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman@es.net				Phone: +1 510 486-8634