[comp.dcom.lans] Summary of responses to my question about punchdown blocks.

roy@phri.UUCP (01/28/87)

	A while ago (in article <2551@phri.UUCP>), I asked for advice on
re-doing the terminal wiring in my machine room.  The basic situation is
that we've invested a lot of time and money in stranded wire.  I want to
get rid of the rats-nest wiring and find punchdown blocks a very attractive
way to go, but I've heard that you can't put stranded wire in them.

	True to Usenet form, if you ask a simple question, you get at least
six different, mutually exclusive, answers.  I havn't done a careful count,
but it looks like we're about evenly split between "sure, go ahead and do
it; it works fine, I've wired up half of Texas like that." and "anybody who
tells you it's OK to put stranded wire in punchdown blocks deserves to be
shot!".  Oh well.  I tend to believe the latter, partly because that seems
to be the word from people that I have reason to trust.  But then again,
suposedly reputable people have told me just the opposite.  Draw your own
conclusions.  For what it's worth, the folks at Glasgal Communications also
don't like putting stranded wire in punchdown blocks.

	There were a smattering of replies posted to the net over the past
couple of weeks; they are not included in the summary.  Thanks to everyone
who responded; my apologies if I missed anyone.  Comments [in brackets with
my initials, like this -- RHS] are my additions.  Elipsis [...]  indicate
sections I've cut out to save space.  And, no, I havn't yet figured out how
I'm going to solve this mess.  Next building I wire up, I'm going to use
solid wire, but it's too late now.

/roy

----------------
From: Bob Paver <cmcl2!ames!im4u!milano!paver>
Organization: MCC, Austin, Texas

Having just finished cabling a 200,000 square foot building with lots of
twisted pair, I'll offer my "professional" opinion.

Telephone punch blocks WILL cut the stranded wire.  The edges of the
"clip" that holds the wire are sharp (enough to cut the insulation on
the wire).  We used punch blocks (mfg. by Ortronics; bought from Anixter
I believe) that are designed (supposedly) for stranded wire.  The "clips"
have rounded edges.  So far so good.  What we found, however is that
the punch blocks are constructed such that one has to punch both wires
into the same clip.  This often results in a loose connection.  If we had
to do it again, we'd use something else.

The most recent catalog from Black Box shows circuit boards with modular
(RJ-11) jacks that appear to be suitable for stranded wire.  I've never 
seen one so am not sure

----------------
From: David G. Cantor <hplabs!dgc@CS.UCLA.EDU>

We (UCLA Mathematics Department) have been using punch blocks and such
for several years.  We have runs of over 500 feet at 9600 baud using
standard Telco twisted pair (NOT SHIELDED) and have very few problems. 
The reason is, of course, that the power wiring in commercial buildings
is all in steel conduit and so is shielded from our wiring.  We use
slightly different wiring from you with 5 wires (in effect, DTR in both
directions) and use patch panels using 6 pin modular jacks.  We have
punch blocks in many locations.

Impedance mismatches are not a problem.

I wouldn't want to connect stranded wire to a punch block.  Punch blocks
are designed to make a good connection by spring pressure against solid
wire.  And the probability of a short (from a mislaid strand) is quite
high.

Let me make a simple-minded suggestion.  Buy some 1/8 inch thick by 3
inch wide plastic strips, 2 1/2 feet long. and mount machine screws on
them, from the back, using one 1 inch 6-32 screw and one nut.  Then
mount the strips to a wall.  Wires can then be fastened to these screws
using two washers and one more nut.  The screws can be mounted in groups
of three, horizontally, say every inch.  Thus five strips would take
all of your cables with some to spare and two strips would take all of
your ports with plenty to spare.  Then, after all terminals and ports
are connected, you can mount the jumpers on top using a nut and two more
washers (alternatively you can connect the jumpers between the same nuts
as the other wires).

The telcos used terminal strips like this until about 15 or 20 years ago.

----------------
From: sun!amdcad.AMD.COM!phil (Phil Ngai)
Organization: Advanced Micro Devices, Sunnyvale, California

>	What I want to do is set up a panel with a termination for each
>port on one side and a termination for each external line on the other and
>patch between them as necessary.

Good plan. That is what we do. Works great. Cheap, too. For us,
quality and reliability is most important but when you can have a low
price too, it's wonderful. We have tried RJ-11 patch panels and had
some problems.  Not a lot, but we have none with the punch down
blocks. Wire is something that The Phone Company really knew how to
handle.

> I understand that punchdown blocks are made to only work with solid wire.
> [...] The last thing I want is to have hundreds of connections start to
> work themselves loose in a year or two.

You are correct to be concerned. One of our buildings in Texas was
wired for voice with stranded by mistake (not by our group) and they
have intermittent problems. The problem is that the clip which pierces
the insulation and contacts the wire doesn't have enough spring action
to follow the compression of the stranded wire. Solid wire, as you can
imagine, resists compression much better and provides a much more
reliable connection. This is for the 66 block.

There are blocks designed to work with both solid and stranded. Have
you talked to your local telco supply house? Here in California we
like Anixter Brothers. [I havn't yet called them, but they are on my
list -- RHS]  They are a national company, I think. Also there is
Graybar. If you get really stuck, maybe ATT can help.

>	The frame ground is connected to the shield only at the computer
> end and xmit and rcv are each twisted with a signal ground

This also sounds like a good technique.

> If the 20-30 feet between the patch
> panel and the computer is 8723, as is the external line, will I loose much
> if I make the patch between them plain unshielded solid phone wire?

No problem. Your cross connect (the telephone term) won't be more than
five or six feet, right? [right -- RHS]

>  Do impedance mismatches make any difference at these rates?

At these data rates, absolutely not. Impedance mismatches are
important when the period of a bit is comparable to the time it takes
a signal to propogate down the cable. If the reflection settling time
is a small fraction of the bit period, it will never be seen.

> should I just be worried about how many feet of wire are exposed to EMI?

I wouldn't even worry about that in an office environment, except for
florescent lamp ballasts and they can usually be handled by rerouting
the cable.

----------------
From: cmcl2!seismo!rochester!kodak!ornitz (barry ornitz)
Organization: Eastman Kodak Co, Rochester, NY

I do not believe that the stranded wire will work well in your application.
A simple approach might be to use punch blocks with paralleled connections:
solder the stranded wire to one terminal and use solid wire from a paralleled
terminal to patch elsewhere.  If you do not have blocks with paralleled
terminals, use short wire lengths of solid wire to jumper to an adjacent
terminal pin.  Only the terminal pins with stranded wire will be soldered so
the "bite" of the punched-in connections will not be degraded.

As a rule of thumb, line length must usually be over one twentieth of a 
wavelength before impedance mismatch creates any problems.  This works out
to about 2500 feet for 19.2 KB so I am sure this will cause no problems with
differing types of wire you are considering.

One way to handle the shielding is to use a zipper shielding braid over the
entire bundle of wires.  Alternately you could run shielded wires in and out
of a metal box with the punch blocks located inside.  Ground the box and 
connect the cable shields to the box.  If you are not concerned about looks,
aluminum foil can be used to shield the bundle.  A spiral wrap ground wire
over the foil is a good way to make connections.

----------------
From: cmcl2!jhc@mtune.ATT.COM
Organization: AT&T ISL Middletown NJ USA

At risk of pushing my own company's products, we have a whole building in
Middletown NJ wired up with 110 hardware, not the 66E blocks to which I
think you are referring. The 110 hardware is the building block of AT&T's
Premises Distribution System. It does work with stranded wire, it is easy
to make patches, you can do pretty much anything you want with it. There
are enough different configurations that a lot of things are possible -
factory-or-field terminated, 3/4/5 pair blocks, logical closet layout, and
so forth. The down side is that it costs a lot (I think - I never bought
any 66E blocks! Any patch cables other than 3-pair cost an arm and a leg
though).

----------------
From: cmcl2!harvard!axiom!lpi!jdc
Organization: Language Processors Inc., Waltham MA

Roy:  I have done what you are want to do and punching stranded wire into
telco blocks works just fine even though it is more difficult to handle
than solid copper wire.  Here are a few suggestions:

1) Invest in a good impacting termination tool.  In case you don't know
   what I mean it is a tool wherein the blade is spring loaded so that
   at the last moment it "snaps" forward cutting off the excess wire
   cleanly without effort.  It works beautifully on 24 ga. solid copper
   wire but may take a couple "snaps" to cut off the stranded wire.

2) Buy a spare blade at the same time you get the tool.  If you nick the
   blade you can forget about terminating stranded wires.

3) Use 2-pair, 24 ga. solid, unjacketed wire for your patches.  It comes on
   1000 ft. spools.

4) Lay out your patch panel so there is room for wire guides and then use
   them.  I saw a settup where a guy tried to do all the connections
   point-to-point and it was a disaster!

5) If you have any thought of ever installing a MICOM port-selector (which
   is very easy if you install your patch panel correctly) you might
   consider punching your terminal lines in the following order,

	3 - Received Data
	7 - Ground (or DSR, on a micom it can be either)
	2 - Transmitted Data
	20- DTR

   Your should also do you computer lines the same way.  This way your NULL
   cross-over is made via your patching wires.

   This is the order that MICOM uses for its interface so if you used telco
   blocks that have been prewired with group terminators you could just
   plug the MICOM between the two sides of you patch panel, strip out the
   patching wires and have yourself a network.


I've rewired two different companies using telcos and for RS232 I wouldn't
have it any other way.  I even have some custom parts that I order that 
give me extra flexibility on the patch panel and remote connections.

----------------
From: David Marston <cmcl2!harvard!dartvax!marston>

I have punched stranded wire onto Telco-type punchblocks; it works but it
isn't ideal. It seems to help if you give the termination an extra squeeze
with pliers after punching. The (chemical) type of insulation may have an
effect on the results, also--ours are PVC.

----------------
From: sun!hplabs!rutgers!ems!barker
Organization: EMS/McGraw-Hill, Eden Pairie, MN

	I would recommend against using stranded wire in telco punch
blocks.  As you said, they are intended for solid wire.  I'm not sure what
to reommend that you use for stranded wire.

	As far as shielding and EMI go... I would advise you to use
shielded wire where ever possible and to ground it at no less than 1 end,
preferably both.  [I was under the impression that grounding both ends
doesn't give any better shielding, and in fact is bad because it causes
ground loops; can anybody comment on this? --RHS] Two things govern a
cables suseptability (sp?) to EMI, shielding and proximity to EMI sources.
EMI sources are not only computers, but power supplies, florescent lights
and also other cables.  I've heard of people running 2 long unshielded runs
of RS-232 cable right next to each other (ie. 1 for a remote terminal and
one for a remote printer) only to find that when one of the devices is on
and idle, it picks up the data intended for the other.  This is due to the
fields generated in one of the cables being induced into the other.

----------------
From: sun!hplabs!rutgers!meccts!mecc!sewilco
Organization: Minn Ed Comp Corp, St. Paul

Find a nearby cabling company and ask them what they've got for stranded
cable.  Dave Olson at Computer Systems Products, here in Minneapolis, says
that insulation-displacement connectors (ie, punchdown blocks) don't work
well with stranded cable.  The wires deform as the insulation is deformed,
and the contact points slice strands and otherwise make for loose
connections.

For a year I've been using a cabling system based on RJ-45 connectors.  These
are the 8-wire version of the RJ-11 telephone modular connectors.  RJ-11
cabling systems are available, and with their four wires should work well.
Components for these are available from CSP, Black Box, and many other
sources.

If your existing 100 wires are mostly separate cables and you can't justify
a patch panel, maybe you can just get a dozen RJ-11 wall plates or surface-
mount jacks (3" square boxes).  When you need to change an old cable,
connect it to a jack.  You'll eventually end up with all the jacks neatly
mounted on a plywood panel.

If some of the 100 wires are in telephone-style 25-pair cables you can
terminate it with what is called an "Amphenol" connector (you've seen
them on multiline phones).  It might be best to have a cabler do this.
These can be connected to converters to RJ-11.

On the terminal and/or computer you'll need a data jack, which converts
from RJ-11 to DB25 (and a cheap cable from the data jack to the "patch
panel").  Black Box is expensive; CSP list is $6.95.

With a little study, I found I could easily cost-justify this cabling
system.  If changing one cable uses $8 of your time, these cables pay for
themselves after two or three cable changes.  If putting in new cables,
the 25-pair cables are cheaper than single cables over a distance of
less than four feet (yes, 4 feet!) if more than a few terminals are grouped
together.

----------------
From: Jim Wright <allegra!ulysses!ecsvax!jcw>

We do it a lot.  We use Belden 8457 (12-cond) and 8444 (4-cond) to run to
offices, and we use 8444 to patch between blocks.  We have run up to 1000
feet (rs232) at 9600 bps.  We don't connect anything to pin 1... we use 2,
3, 7, and 20, because DTR from the terminal is used to wake up our Develcon
Dataswitch (we also have terminals wired similarly connected to Timeplex
switching muxes and DEC terminal servers).  In six years of doing this in a
large network with at least 300 terminals, I don't think there have been
more than a half-dozen times when we had a bad connection where the
insulation wasn't sufficiently pierced by the connector on the block.  As
for impedance... remember, this is all unbalanced, so you could just as
well use barbed wire. [I don't understand this comment.  Why does a
balanced line have to be impedence matched, but not an unbalanced line?
I'm used to thinking of balanced meaning like TV twinlead, and unbalanced
meaning like coax -- RHS] We patch with 8444 so that it is easier to follow
which wires go where in this maze of spaghetti.  But using solid wire
without an external sheath (a la Telco) means that a given bundle of
conductors will have less cross-sectional area... this can become important
with a LOT of patching.

----------------
From: allegra!@EDDIE.MIT.EDU:phr@hermes.ai.mit.edu (Paul Rubin)
[Note: this gets my award for "unparsable return address of the month" -- RHS]

1. I expect stranded wire will lose pretty bad with punchdown blocks,
because of the way the punchdown tool works.

2. Punchdown is a gigantic pain in the neck; why don't you use
modular telephone-style cables (available in 4, 6, or 8 conductor)?
I don't expect it would be much more expensive than punchdown.

----------------
From: hplabs!hpscda!hpscdw!garyg

Stranded wire seems to work OK if your wire size is 1-2 gauges larger than
the smallest rated for the punchblock. It just needs to be big that the
strands don't get pushed aside while the insulation is being displaced.

We have runs of over 400' running in 25 pair phone cables in an R&D lab,
totally unshielded, 8 circuits per 25 pair, without any crosstalk or noise
problems. Your mileage may vary.  If anything, the shielded cable with it's
higher capacitance may cause problems. Good luck!

----------------
From: sun!cognito!randolph (Randolph Fritz)
Organization: Sun Microsystems, Mountain View

> Has anybody tried using stranded wire in punch-down blocks?  Will it work?

It won't -- sorry.  Try screw-terminal barrier strips instead.  You could
bring your existing "cable plant" -- your terminal cables -- to barrier
strips and re-wire your computer room with solid wire.


> If the 20-30 feet between the patch panel and the computer is 8723,
> as is the external line, will I loose much if I make the patch
> between them plain unshielded solid phone wire?

Generally, you should see no problems as long as you do not leave
unterminated quads running into your computers -- capacitive coupling
between send and receive lines produces an on-again off-again loopback.  A
more serious problem with RS-232 is ground shifts and ground loops.  If the
money is available, I would strongly recommend the use of line-powered
line-drivers.  These things prevent vast amounts of grief, but were costly
last time I looked (but look again -- that was almost a year ago).

Two further suggestions:

  Glasgal Communications (Northvale, NJ) offers a very good line of
  cable management products.  If you have a need for serious telco-
  style equipment, connection frames and so forth, try Greybar Electric.
  
  Nevada-Western (San Jose, CA?) makes a very useful line of
  connectors and patch panels.

----------------
From: cmcl2!ihnp4!ihdev!dlr
Organization: AT&T Bell Laboratories

Having just gone through something like you describe at a field test site,
I would suggest that you go ahead and try the setup you describe, but with
a couple of minor changes.

1.  When you terminate the stranded wire in the pushdown block, solder the
incoming side to the pushdown block for stability.  This will prevent
unraveling of the strands.  Only do this if there are only minor changes
over time.  Unsoldering the blocks gets messy and tedious.

2.  On the cables running from the termination blocks to the computer, I
suggest some sort of shielding, even if it is only taking the wires and
twisting them into cables (labeling the pairs, of course) and then wrapping
them with lead or copper foil (if available) or getting shielded 25-pair
cables from your local wiring source (should be some of that type of wire
around for the old 6-button key system phones).

----------------
From: cmcl2!ihnp4!ihlpf!spear

You might try the little spring loaded clips that you can buy at radio
shack - these work well for twisted wire and are fairly cheap.  I am
talking about the ones with a button and a hole beneath it which closes on
the wire when the button is not depressed.
-- 
Roy Smith, {allegra,cmcl2,philabs}!phri!roy
System Administrator, Public Health Research Institute
455 First Avenue, New York, NY 10016

"you can't spell deoxyribonucleic without unix!"