Archived from groups: (
More info?)
In the Usenet newsgroup alt.internet.wireless, in article
<49g0j1t135i54rvuthdm7e009525a1jhno@4ax.com>, Jeff Liebermann wrote:
>All of my early 10base5 and 10base2 installation were in "industrial"
>environments. That's where nobody cares about what it looks like as
>long as it works. Cables hanging from the overhead were standard.
>I don't recall ever running ethernet through the walls to wall plates.
Everything was exposed, ugly, and messy. Works better that way.
Our industrial areas were limited, but even there the stuff was
mainly out of sight. Tends to reduce accidental damage.
>I stock 4 boxes of CAT5, where the only difference is the jacket color.
>White, beige, grey, and blue.
We didn't have that much CAT5, and it was almost always a gray, with
some older stuff beige. We also had a limited supply of green, but they
were cross-over cables only.
>I give the secretaries the choice of colors and invariably, they want
>some color I don't stock.
We follow the Ford Mantra - any color you want as long as it's black^Wgray.
>Also, if they're into Feng Shui, run away. I've had to negotiate the exact
>location, direction, color, and termination on the basis of how it affects
>the flow of chi.
I don't know if it's hiring policy or what, but we don't seem to have any
of those.
>2nd phone at 18. 1st phone with radar at 19. (Took me a while to
>find the old licenses).
Got the 2nd + radar, then the 1st - about a month apart. Was in the
service in Denver, and had to fit the tests in between official duties.
>Got tired of RF and dived into computers, where I successfully repeated
>all my previous mistakes.
I was doing computers in the 1960s as part of the system I was tech-repping
in Japan.
>At the first FM transmitter I baby-sat, there was a warning that if
>the VSWR meter (Bird) every moved from the peg, the final would
>probably blow.
I mainly did AM and UHF TV repeaters. I lucked out in not having much
hardware problems, so it was mainly just keeping the logs up, and making
sure the air filters got changed regularly. Chief engineer would peak
the tuning quarterly, whether it needed it or not ;-)
>At a different nightmare, I had to risk my life grabbing radios out of
>the building as blocks of ice the size of desks came down from the TV
>tower because the automagic VSWR meter relay had failed to start the
>de-icers. Wheeee...
I was raised in the North East, but starting in the mid-1960s, spent most
of the time in warmer climes. Yes, I did have one occasion when I was
driving in snow (Sunol pass on I-680 between Fremont and Livermore - a
whole 980 feet above sea level), and there were occasions when I saw
snow on the hills above San Jose, but I've mainly forgotten what snow
and freezing drizzle is. Not missing it one bit, either.
>However, I'm a big fan of using bridges for precision measurments. While
>not exactly precision, that's how the Bird and MFJ antenna analyzers work.
>It's difficult to build a really broadband HF directional coupler that is
>flat from 1.6 to 30Mhz.
More than an octave is always going to be difficult. Nearly all of the
stuff I worked on was narrow band - 10% was wide band. Still stuff like
the slotted lines could cover an appreciable range. The last FAA project
I worked on was MLS, and we had a 1 percent bandwidth. Waveguide everywhere
and dual directional couplers (loop) that had been hand tweaked to provide
46+ dB directionality. I'd hate to be the poor sod who was assembling those.
>Gaak. I charge by the hour so everything was a rush job. Throw
>everything together and hope for the best. Test only when done. I
>don't like doing it like that, but the labour content would have been
>double if I had stopped to test every connection as I went along.
I dunno - the initial install and termination was one step, and as soon
as we had the first drop working, we had a system on the wire to ping.
Our "standard" was 'ping -s 8192 -c 25' which results in 150 packets. If
we saw a single drop (and don't forget that with a -s 8192, that results
in six packets that need to be faultless), we'd try again. A second
"failure" or more than a single drop in the initial ping was rework time.
>That's what I really liked about 10base2. If you made a mistake in
>the cabling, everyone crashed simultaneously. I never had a problem
>with nobody noticing a problem. The entire company would be up in
>arms screaming at me.
A bad vampire install on 10Base5 was the same. Had that happen a couple
of times - no thank you.
>I kinda missed that with 10baseT and switches, where I could totally
>screw up a segment, and nobody would notice.
That's why we wanted that first drop running something to ping. As
soon as the drop was wired, and the far end connected to the central
point - we'd throw a test set on and watch the lights. This detected
the common wiring errors. Then a lap doggy, and a ping test. If that
worked, on to the next mess. If not - find and fix.
>>That's what interns were for. ;-)
>They're not disposable.
<innocent look> They're not??? </innocent look>
>They would have died finding that one. 10base2 run between two
>buildings. Coax grounded at both ends to AC power ground. We lost
>the ground (or neutral) in one of the buildings making the 10base2
>coax the AC power ground return for the entire building.
Wowser! The facility wire bender was trained by Edison or Tesla, or
somebody, and he was rather dogmatic about wiring safety. Every
training course I've ever encountered has demanded one and only one
ground. Actually, between buildings, we ran twisted pair to
isolation transformers on both ends, and used half repeaters. About
1992, that got replace with fiber.
>Medical office database server was trashed when the vendor installed a
>program update while everyone was still logged in.
news://comp.risks/ - that's common place, and people NEVER seem to learn.
>I had to sit there and accept verbal abuse from the entire medical staff
>because it was taking so long.
That's normal. The network can be down for ten minutes, and we hear about
it from Corporate on the East Coast, even if it's midnight here.
>I forgot to mention DLT. That worked fairly well with few errors.
>However, it was rather expensive and the drives needed constant
>cleaning and occasional rebuilds. In general, the tapes were
>transportable between different drives and error rate was very low.
Tape drive cleaning is a given here. I'm no longer involved, but with
8mm drives, we were running about 7 hours a night per drive, so every
Tuesday was cleaning day. Every drive in the place. Tuesday is also the
day that tapes go/come off site, so the guy who collects tapes is the
one who does the cleaning.
>Recommended.
At home, I've recently switched to redundant drive backups. I have two
cheap systems each with two 120 Gig hard drives, that have multiple
partitions - each one mirroring one of the drives on the servers.
Hard drives are getting reasonably cheap - I paid US$50 each for these.
The important stuff still gets burnt to CD, and lives off site.
>The 75ohm BNC's are exactly the same dimensions as the 50ohm
>BNC's in the area around the center pin, PTFE sleeve, and shield
>fingers. However, the cable crimp diameters are different.
I don't use them, but my understanding was that the dielectrics were
different in order to bring the impedance in line, while maintaining
the same intermatability.
>My string of adapters showed almost no VSWR at 450Mhz. However, I
>cheated a bit. I avoided UHF connectors and right angle adapters.
UHFs were never spec'ed for impedance, although they've been used
in RF since before WW2. Right Angles? Yeah, they've never been able
to get a decent VSWR. There was a company (might have been Omni Spectra)
that used to advertise pre-built cables of 141 semi-rigid with
impossibly short radius turns that fit the same space as a 90 degree
connector, but had VSWRs down in the 1.02 + 0.007GHz :1 VSWRs which was
little worse than the guarantee of an SMA straight plug. I've also seen
it done with quarter inch hard line in a type N, but vaguely remember
that the cost was astronomical.
>Another fun test was a long string of BNC "T" connectors in series. The
>short stub cause a small amount of VSWR and possibly some leakage, but
>the effects were rather small. I think I had about 50 connectors in
>series.
As long as you're low enough in frequency, that's true. Things start to
go south on you above 3.0 GHz, when that 16 mm stub becomes a significant
part of a wavelength.
>So far, I haven't had to do drywall rework. The main vertical pipes
>and splice boxes are hidden behind removeable wood panels. Another is
>under the stairs.
Single floor slab, and relatively low attic space (most rooms have
cathedral ceilings). The boss isn't interested in extraneous wood trim
strips - yeah, I thought of that too. I did hide some of the runs to
points on the outside walls behind the base molding.
>However, FNT flex non-metallic tube (smurf tube) would have been much
>easier. However, the stuff is not up to code for many applications
>which might cause problems with the building inpsectors.
If we ever build a new house, there will be some added features. If
you look back, it wasn't that long ago that phone wires went to one
(at most two) places in the house, and the TV lead came through the
window. My neighbor, who is a retired school principal still can't
figure out why we need six computers in the den (neither can I). The
other neighbor is a CIO with a major bank, and she can't understand
why there are three networks in the house. (1 is company, 1 is the
normal one, and the third is for play and backups.) One of the guys
I work with just built a new house. Conduit in every room with two
fibers - the "den" has six drops. The patch room is an air conditioned
closet in the garage.
>>I've run tests with Gigabit copper, but I see a lot (10%) of packet
>>errors.
>That's awful. I don't see anything that bad.
Only tried once - I had to borrow Gigabit gear, as I haven't seen the
need to buy it yet. Our network isn't that busy.
>If you're getting 10% uncorrected errors, then you've got a wiring
>problem. My guess is split pairs or rotten connections. I've also
>had problems where I ran the CAT5e (before overpriced CAT6) next to
>large metal objects (rack rails) or near magnetic interference sources
>(flourescent ballasts, motors, ferroresonant xformers, etc).
Magnetic - no, but the air ducts are steel, and bonded. They're also
huge.
>I also had problems with stranded wire patch cables and gigabit.
I suspect some of the problems are the patch panel, which is in a
closet along with the firewalls and routers. I'm _reasonably_ sure
I've got the pairing correct, and did take care to keep the pairs
twisted as close to the terminations as possible (less than an inch),
but I'm not sure that punchdown blocks go that well with Gigabit.
The 10BaseT performance is OK - virtually nil errors.
Old guy