Topband: 2Wire, Inc. 3800HGV-B Gateway. RFI --Problem fixed for now..

[Guy Olinger K2AV]

The problems with the 2-Wire corp's 3800 are that even WITH the degree of
suppression afforded by twisted pair, if a 1.5 kW 160/80/40 signal (which
is IN BAND to the 3800) is around, that the depth of forward error
correction is not enough to keep up with the errors that are unavoidable.
 There are many, many signals which get into the signal and cause errors,
lightning static from strikes within a few miles, junk on other telco cable
pairs, on and on.  The 3800 here was getting 100% correction on what the
AT&T guys called an absolutely sh*tty loop from the pole.  It had about a
half dozen splices, zero shield continuity, was terminated in a pole box
with decades old carbon lightning protectors.  My neighbor's line had about
ten splices in it, and was in worse shape.  BUT the forward error
correction took all that cr*p and made the result perfect HD + internet +
IP phone.

1500 watts of 1.825 to an efficient antenna took it down.

They completely repaired the loop and bypassed the old stuff in the cable
termination box. They said they used the "good" stuff for the loop for this
problem since it was so severe.  I don't know what "good stuff" meant.

1.5 kW 160m still took down both our gateways.  The break point was about
300-400 watts. The 3801 fixed it all.

The 3801 gateway has about 100 times the cache and can delay output far
longer, allowing the FEC to work through it all the first time and start
disabling sub-carriers.  When I'm done with the CQ160 SSB, the 3801 has
turned off all the subcarriers from 1.7 to 2.1 and it's not even showing
CORRECTED errors beyond the volume of misc stuff it handles all day long.
 Weekly operation on 160/80/40 will keep a collection of affected
subcarriers turned off, and they will gradually come back if operation
ceases.

Each subcarrier has a data rate that is managed by the 38xx, it knows the
point where each individual subcarrier goes into errors, for whatever
reason, rolloff, noise, interference, AM BC carriers, hams, whatever.  The
management is dynamic and goes on all the time. An error on a subcarrier
causes an individual subcarrier retest.  Massive lengthy pervasive errors
can cause something called a retrain, where the whole thing restarts and
tests all subcarriers.

Pretty clever box if you ask me.  It just heals around stuff that gets in.
 A close lightning strike will cause it to lose sync, because the induced
noise is so broadband.

Four active channels of HDTV + 18 MB Internet + IP phone.  Over crappy
copper pairs.  Awesome technical accomplishment.

73, Guy.

On Sat, Feb 4, 2012 at 1:38 PM, Jim Brown <jim at audiosystemsgroup.com> wrote:

> On 2/4/2012 8:13 AM, ZR wrote:
> > Im running 250' of shielded CAT5 from house to equipment trailer where
> the
> > router resides and then another 120' up the tower to the 5.8GHz link.
>
> The issues, Carl, are not with CAT5, shielded or unshielded, but rather
> with the equipment on either end, which can generate RF trash or be
> susceptible to RF trash. Some commonly used Ethernet products produce
> birdies on the HF bands (a couple of frequencies on 30M, around 14,030,
> 21,052, the low end of 10M, the low end of 6M are places I hear them,
> and there are others). Some digital equipment produces loud broadband
> hash. I've seen some POE stuff used with data links at tower sites that
> were quite noisy on the MF and HF bands. This trash gets radiated
> several ways -- usually most strongly from common mode current on the
> Ethernet and power supply cables (or coax if it's used), and often
> directly from wiring internal to unshielded boxes.
>
> Ferrite chokes can greatly attenuate the common mode part of that.
> Using shielded CAT5/6/7 can help if, and only if, the shield is
> optimally terminated at both ends (both ends are likely generators of
> digital trash). That usually means bonding to the shielding enclosure at
> both ends.
>
> 73, Jim Brown K9YC
> _______________________________________________
> UR RST IS ... ... ..9 QSB QSB - hw? BK
>