Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!sdd.hp.com!hplabs!hpda!hpcupt1!hprnd!pat
From: pat@hprnd.rose.hp.com (Pat Thaler)
Newsgroups: comp.dcom.lans
Subject: Re: Intelligent Twisted Pair?
Message-ID: <2230161@hprnd.rose.hp.com>
Date: 24 May 91 00:41:22 GMT
References: <1991May19.235055.23105@trl.oz.au>
Organization: HP Roseville Networks Division
Lines: 59

In comp.dcom.lans, hayes@blaise.trl.OZ.AU (Mark Hayes) writes:

    I have just seen an article which mentions "Intelligent" Twisted
    Pair cable manufactured by Cabling Communications Consultants Ltd.
    (Farnborough, UK). Apparently this cable is better than unshielded
    twisted pair and cheaper and less bulky than shielded twisted pair.
    CCC claim that its cable can accommodate 200 token ring stations 
    (4 or 16 Mbit/s) over 200 metres.

I haven't heard about this cable in particular.  However, it sounds
like a member of a class of cable that has been appearing over the
last two years.  These cables are unshield twisted pair, but because
of the use of different insulations (material and dimensions), and
different construction (higher twist rates), they have lower attenuation
and better crosstalk than garden variety UTP.  Some use 22 AWG, rather
than more common 24 AWG of telephony UTP.  Some are available with
an overall shield (as opposed to the individual pair shields of IBM
type 1 and 2 STP).  AT&T (which calls their line Systimax), Belden,
and Northern Telecom all make such products.

If you want to know specific specs, you should get data sheets.  The
products vary in the trade-offs made in their design, so the specs
vary too.

    I am not too familiar with physical layer restrictions on token
    rings. I assume that the attenuation per unit length is the same
    (or similiar) between the different cabling systems (assuming same
    wire gauge) and the limiting factor is crosstalk/interference. 
    
Actually, that is an invalid assumption.  The dielectric (insulation)
affects attenuation.  For instance, the attenuation of 100 m of
AT&T Systimax is several dB less than that of normal telephony UTP
even though both use 24 AWG wire.  The improvement in crosstalk
is more impressive.  I don't have the spec sheet in front of me now,
but my recollection is something between 10 and 20 db improvement
over telephony UTP.  You pay for this in a cable that is more
bulky and expensive than telephony UTP, but less bulky and expensive
than STP.
    
    As
    STP reduces crosstalk/interference, then greater distances are 
    possible (yes/no?). I would assume that the ITP cable reduces the
    crosstalk/interference below that of UTP, thus enabling their
    better performance figures. I also assume that this applies to 
    networks which are "star wired", thus having multiple stations 
    (on different pairs) in a cable.

Better distances are possible because of the attenuation improvement.
Additional improvement due to the low crosstalk is theoretically
possible, but there is a catch.  Receivers designed for UTP may
not have the sensitivity to receive a lower signal or they may have
a squelch circuit (to keep them from seeing crosstalk as incoming
signal) that won't open for the lower signal level.  

The cable I have seen is generally in 4-pair or at most 6-pair.
So far, the work that I have seen still assumes one token ring station
connect per cable.

Pat Thaler