The typical POTS line dc condition is simply the battery feed retard which
is
usually a 200ohm + 200ohm balanced high Z transformer.

Essentially, behind one of the 200ohm windings will be -52V dc and behind
the
other 200ohm winding will be +ve. Remember that the +ve side of the
exchange
battery is connected directly to ETH. So, at the customer premises the +ve
leg
will have 1/2 the line loop resistance plus the 200ohm retard to ETH and
the -ve
leg will have a similar line resistance + the 200ohm retard to -52V.

Thanks for this info. What you are saying is that the DC resistance and the
AC impedence should be the same on TIP and RING.

Your ADSL modem must be connected on the raw line connection and each
telephone
device which relies upon the same POTS line must have a filter (unless you
use a
whole-of-premises filter). If you can get your modem connected directly to
the
line at the incoming connnection point and throw off all other telephone
instruments inside the house you will be able to determine if your internal
wiring is the cause of the problem.

As I pointed out before, the house wiring is totally excluded from the test
I performed, because I plugged the modem directly into the AT&T outside box,
after having disconnected the house wiring from the line. After some further
investigation, I think I found the problem. I have establish that when
everything is working normally the resistance (DC) of TIP is 2,166 Ohm and
the RING is 2,179 Ohm. But, when I am having ADSL problems, the TIP
resistance drops to 146 Ohm. I am speculating that somewhere along the line,
the TIP wire insulator is damaged and it shorts to ground, disrupting in
this way the balance of the line. Because the phone lines to my house are on
poles, I suspect that small changes due to the atmospheric conditions (wind,
temperature, rain, etc.) are creating these sporadic shorts.
I called AT&T and they agree to re-send a technician.

Thank you all.
Gene