Bicephale
join:2005-09-24
 ·TekSavvy Solutions..
2 edits | reply to diskace
Re: Feed-Back from a grateful customer!
Hi DiskAce,
I've discovered a few things manipulating 'DMT' in
the course of the last week-end, relatively to the
disconnections which followed the profile upgrade:
 http://www.2cent.de/dmt/titanium2/dmt.zip
Perhaps there's no correlation, this situation may
have improved by itself while i thought i was able
to trim my Line Attenuations and SNR Margins using
'DMT' but i believe the noise peeks correlate with
human activity quite well. Test conditions change
while observations are made, most likely, but i'll
try to get snapshots once a day for a week then we
will see how they superpose. If human activity is
shown to be involved because of grouped events, it
will point at noise generated in one of three ways
or a combination of those: voltage variations due
to cooking & late TV viewers, induced interference
caused by these or just higher InterNet traffic...
As i tried to explain at the start, i'm searching;
i use the clues which are available to me, no more
no less. I seem to be able to compensate for line
noise by sacrificing my UpStream bandwidth: there
is an amount of interference created in any power-
stage emitting RF energy and consumer products are
certainly cheaply built to a point where the noise
it spits in the DownStream bandwith may be a tight
tradeoff that tweaks hardly bring back on the good
side, so to speak... Finally, it looks like i can
make the MoDem less susceptible to noise generated
by the phenomenons i mentioned above. That's only
guessing, there are many variables and it may very
well be over before i get a finger on the cause...
Unfortunately, there's no explanation as to how it
was possible to observe one PPP session going down
while the other stayed up. I haven't rejected the
possibility of a compounded problem where external
causes and line noise were mixed; i wouldn't even
totally reject the idea that the transmitter stage
in the MoDem became too noisy itself but i have no
spectrum analyser at hand to investigate this last
idea. At least i didn't even got to spend a dime!
I hope my tweak will hold on, only time will tell!
 |
 .PNG/thumb.jpg "66709 bytes")