| Author |
Message |
Franc Zabkar
Guest
|
 Posted:
Fri Jan 14, 2005 8:03 am Post subject:
Re: Dirty Phone Lines or ...? My Next Step? |
|
|
On Thu, 13 Jan 2005 17:39:48 -0600, LoadHawg <not@chance.com> put
finger to keyboard and composed:
| Quote: | OK I finally got it to stop clearing the modem after a disconnect and
was able to induce the errors and slow speed. Usually the connection
ceases to pass data anywhere around 50 to 100 errors which I can
usually get to in a matter of minutes if I try however this one, while
it was easy to induce the errors... hung on tenaciously passing a
'trickle' of data all the way up to 578 errors ...
|
What kind of errors do you see in your modemlog and ppplog?
http://www.modemsite.com/56k/modemlog.asp
http://www.modemsite.com/56k/ppp.asp
| Quote: | ... at which point it
simply would not do even the simplest thing and no data whatsoever
would pass and I disconnected the session. I immediately grabbed the
&V and #UD data and ran it thru the diag analyzers - below is both the
raw and procesed data. Interestingly enough it appears to be
maintaining the speeds which leaves me scratching my head... ideas?
thanks,
|
It appears that your ISP's equipment terminated the call, perhaps
because it detected an inactivity timeout (???).
Termination Cause : Disconnect Frame Received
Disconnect reason : LINK DISCONNECT
AFAICS, your phone line seems to be quite good. In particular, your Rx
error rate is very low. The only thing that concerns me is the RBS
figure (robbed bit signalling). Ideally it should be zero. Your other
log showed a value of 1. I have no idea whether this parameter is of
significance.
The variation between your two logs could be due to different ISP
equipment. Notice that your other AT&V2 report identified the server
equipment as Conexant/Lucent, whereas there is no such info in the
current log.
Connection time : 00:38:01
TX Rate (Last/Init/Min/Max) : 31200/31200/31200/31200
RX Rate (Last/Init/Min/Max) : 44000/46667/44000/46667
Modulation/Protocol/Compression : V.90/LAPM/V.42bis
Signal Level (TX/Power Drop), -dB : 5/0
RX Signal Level (Last/Min/Max), -dB : 11/11/11
EQM Value (Last/Min/Max/Negative) : 29/19/66/0
EQM Last 10 Readings : 36 35 29 31 29 30 33 29 29 29
RBS Pattern/Alternating RBS Pattern : 04/00
Retrains (Issued/Granted/Fast) : 0/0/9
Renegs (Issued/Granted) : 2/0
Retrans per frame/Frames rejected : 1/25
Total number of REJ sent/received : 25/7
Minutes Since Last Retrain/Reneg : 18
TX/RX flow control : V.24 ckt 106/133 / V.24 ckt 106/133
TX/RX chars sent : 263152903/1873953024
TX/RX chars lost (data overrun): 0/0
TX/RX I-Frame count : 3328768/11614208
TX/RX I-Frame error count : 1792/25
| Quote: | PS: In the process of trying to get the system to stop reseting the
modem so I could retain the diag info I installed the modem as a
"Standard 56000 V90". For kicks I tried it - guess what? For the first
time in weeks I could not induce errors at will.
|
That's puzzling. Yours is a hardware modem (ie no drivers), so there
should be no difference, unless the initialisation string contains
some unusual AT commands.
| Quote: | I don't know exactly
what's different about this driver ...
|
Not a driver, just an INF file.
| Quote: | ... but one thing I noticed was it
showed no options for things like h/w or s/w compression. (I have
turned both of those off at various times before under the Zoom
drivers but no joy).
|
That's because the generic INF file contains no EC or data compression
AT commands. Regardless, the modem will still negotiate these by
default.
The first part of your AT#UD report is missing.
| Quote: | AT#UD
DIAG <2A4D3263 41=80
|
<snip>
| Quote: | DIAG <2A4D3263 60=5F
|
Try this procedure to have the AT#UD report automatically appended to
your modemlog. I'm not sure if it will work for Win2K, but it
definitely works for XP.
http://groups-beta.google.com/group/comp.dcom.modems/msg/3ab4a8762820aa6d?dmode=source
- Franc Zabkar
--
Please remove one 's' from my address when replying by email. |
|
| Back to top |
|
 |
LoadHawg
Guest
|
| Posted:
Fri Jan 14, 2005 10:15 pm Post subject:
Re: Dirty Phone Lines or ...? My Next Step? |
|
|
Franc I wanted to follow up further on a couple of points you
raised...
Franc Zabkar Jan 13, 9:31 pm
OK to follow up I've looked at the modemlogs before and again today -
in short I don't think there's much of anything there that will help.
If you'd like I can post some but as you know it's just capturing
stuff like the AT commands and my own inits and so forth. I'm seeing
nothing special or unique between a 'good' session and one that I
induce to 'go bad' in that log.
The other one, PPP log looks much more promising. Unfortunately I have
W2K so have to start it at the shell - I'm going to have to come back
to this later today when I have a little more time but this one
definately looks like it's worth checking out further...
<SNIP>
| Quote: | That's puzzling. Yours is a hardware modem (ie no drivers), so there
should be no difference, unless the initialisation string contains
some unusual AT commands.
|
A further follow up on the two different modem INFs and the initial
apparent success of teh generic settings: A) Zoom 3048C V.92 vs B)
generic 56000 V.90. I had earlier reported that no errors were
occuring when I switched to this driver. This is only true in part.
A) it does NOT seem to be popping errors like the Zoom setup - stays
at a rock steady 0 or 2 (depending on which ISP - it's always been
that way for years). This is different from the Zoom 3048C .INF.
B) unfortunately it STILL does die the slow death of devolving to no
throughput - just the dead modem w/ only kb here or there w/ extended
periods of silence. This is exactly the same as before.
While both teh issue of popping errors in the DUN window and the
slowing and final lack of throughput manifested at teh same time - it
suggests they may not exactly be connected. Or the errors continue and
the generic settings don't share that info w/ the DUN... At any rate
the problem of the connection dying is identical but less easy to
notice w/o errors being displayed using the generic driver - just
wanted to follow up on that as it was indeed puzzling. Hope that
helps.
Anyway at this point as soon as I have a little time I'm going to
pursue grabbing some PPP logs for both good and bad sessions to
compare and see what can be seen and will share. I already posted a
full pair of good and bad diags (post call, AT&V...) in a separate
thread - it's long but a full set of data is there for 2 sessions.
Thanks! |
|
| Back to top |
|
|
Franc Zabkar
Guest
|
| Posted:
Sat Jan 15, 2005 2:34 am Post subject:
Re: Dirty Phone Lines or ...? My Next Step? |
|
|
On Thu, 13 Jan 2005 21:35:46 -0900, floyd@barrow.com (Floyd L.
Davidson) put finger to keyboard and composed:
| Quote: | Franc Zabkar <fzabkar@optussnet.com.au> wrote:
On Thu, 13 Jan 2005 17:39:48 -0600, LoadHawg <not@chance.com>:
AFAICS, your phone line seems to be quite good. In particular, your Rx
error rate is very low. The only thing that concerns me is the RBS
figure (robbed bit signalling). Ideally it should be zero. Your other
log showed a value of 1. I have no idea whether this parameter is of
significance.
The odd thing about the RBS figure was that it changes. Every
dump he has posted had a different number. That may just mean
there is one RBS link, and the different numbers listed
indicated a different frame number for each connect, which would
be normal.
Note that the Signal Level figures for both TX and RX are
"adjusted". It appears they refer to testtone levels, not data
levels. Whatever, without knowing exactly what they are
adjusted for, it is impossible to say they are good or bad.
There is no modem that will transmit at -5 dBm, and a RX level
of -11 dBm wouldn't likely work either. Hard to say what the
numbers actually mean.
Signal Level (TX/Power Drop), -dB : 5/0
RX Signal Level (Last/Min/Max), -dB : 11/11/11
|
FWIW, my typical results are:
Signal Level (TX/Power Drop), -dB : 10/0
RX Signal Level (Last/Min/Max), -dB : 18/18/18
| Quote: | ...
RBS Pattern/Alternating RBS Pattern : 04/00
Retrains (Issued/Granted/Fast) : 0/0/9
Renegs (Issued/Granted) : 2/0
Retrans per frame/Frames rejected : 1/25
Total number of REJ sent/received : 25/7
Minutes Since Last Retrain/Reneg : 18
None of that look bad at all.
However (not included in the quoted text here), the SNR numbers
were 44 and 43 on every one of these reports. That is
borderline.
|
All my logs over the past several years vary between 44 and 48. During
the last year my Rx rate was a very stable 48000bps. I live about
1.5km from the exchange. My modem's chipset is Rockwell/Conexant ACF2.
| Quote: | I'm surprised that it was able to connect at 44K,
though I suppose the reason is that the "hum" being heard is 60
and 120 Hz, and the reports are indicating that the low end of
the bandwidth being used is 150 Hz.
|
- Franc Zabkar
--
Please remove one 's' from my address when replying by email. |
|
| Back to top |
|
|
Jeroni Paul
Guest
|
| Posted:
Sat Jan 15, 2005 5:00 am Post subject:
Re: Dirty Phone Lines or ...? My Next Step? |
|
|
Hi LoadHawg
Let me suggest two things worth checking:
Try PortMon from http://www.sysinternals.com
Induce the errors and see what it shows. Very simple program requiring no
installation.
Use the ATM2 command to keep the modem speaker on, and induce the errors.
Listen for some discontinuity in the noise, which will indicate that it is
renegotiating due to some noise or quality degradation, or some carrier
disruption.
Good luck
BTW: don't be tempted to install ComSpy from Spywindows. This is garbage and
will corrupt your Windows registry.
LoadHawg wrote:
| Quote: | Franc I wanted to follow up further on a couple of points you
raised...
[...] |
|
|
| Back to top |
|
|
Floyd L. Davidson
Guest
|
| Posted:
Sat Jan 15, 2005 4:38 pm Post subject:
Re: Dirty Phone Lines or ...? My Next Step? |
|
|
Franc Zabkar <fzabkar@optussnet.com.au> wrote:
| Quote: | On Thu, 13 Jan 2005 21:35:46 -0900, floyd@barrow.com (Floyd L.
Davidson) put finger to keyboard and composed:
Note that the Signal Level figures for both TX and RX are
"adjusted". It appears they refer to testtone levels, not data
levels. Whatever, without knowing exactly what they are
adjusted for, it is impossible to say they are good or bad.
There is no modem that will transmit at -5 dBm, and a RX level
of -11 dBm wouldn't likely work either. Hard to say what the
numbers actually mean.
Signal Level (TX/Power Drop), -dB : 5/0
RX Signal Level (Last/Min/Max), -dB : 11/11/11
FWIW, my typical results are:
Signal Level (TX/Power Drop), -dB : 10/0
RX Signal Level (Last/Min/Max), -dB : 18/18/18
|
Those are reasonable! If you are using an identical modem, it
does tend to indicate the modem is correctly measuring levels.
If that 5/0 is an accurate measurement, and the -11 is a dBm for
the received signal, that line is *really* screwed up!
The send level for data should be 13 dB down from "test tone
level". Which means the level from your modem seen at the line
card of the telephone company should be -13 dBm, at most.
Modems usually use -11 or -10 dBm as defaults, and assume a 2 to
3 dB loss on the cable (and they are aware that there is 4 dB of
headroom). That gives very little margin of error for someone
located across the road (or inside) the telco office.
The problem is that digital carrier simply clips at +4 dBm, and
data has peaks that exceed 12 dB over the average. A -10 dBm
modem has peaks at +2 dBm, and if that is connected directly to
the telco line card there is only a 2 dB margin before clipping
starts. That can be too close if there is anything at all in
the system that has too hot a level.
If the modem is sending at -5 dBm... the peaks will be at about
+7 dBM, or 3 dB over the clipping point unless the modem is on
a line with at least 3 dB of loss (perhaps 2 or 3 miles from the
line card).
Typically a telco line design target for received signal is -9 dBm,
so a data signal should be at -22 dBm.
Your -18 dBm is 4 dB hot, suggesting you have a either a short
cable between you and the line card, the line card is putting
out a very hot signal, or the modem is not providing an accurate
reading. (In fact, all of the above are likely to be relatively
true.)
But at -18 dBm, you have a reasonable reading that could be
produced by normal circumstances, and should work.
A -11 dBm receive signal isn't reasonable. If that is true, it
is probably overdriving the modem. And that -5 dBm TX level is
probably clipping at the telco's line card.
If that modem's measurements are anything like correct, there is
something wrong with the telco's equipment. It looks like it
has been configured for a very long line, and is actually on a
very short line. Perhaps the telco tech thought boosting the
level would be helpful, which it is correct up to the point
where it clips, and that is almost instant death for v.32, v.34
or v.90/92 connections.
| Quote: | RBS Pattern/Alternating RBS Pattern : 04/00
Retrains (Issued/Granted/Fast) : 0/0/9
Renegs (Issued/Granted) : 2/0
Retrans per frame/Frames rejected : 1/25
Total number of REJ sent/received : 25/7
Minutes Since Last Retrain/Reneg : 18
None of that look bad at all.
However (not included in the quoted text here), the SNR numbers
were 44 and 43 on every one of these reports. That is
borderline.
All my logs over the past several years vary between 44 and 48. During
the last year my Rx rate was a very stable 48000bps. I live about
1.5km from the exchange. My modem's chipset is Rockwell/Conexant ACF2.
|
Okay. That indicates they are giving very conservative
measurements (modems are *not* test equipment as such, and the
readings are totally relative!). If 44 dB is a typical reading
for you when you get 45+K connections, we need not be concerned
about it then.
The problem is, the other measurements may or may not be any
more "accurate". What might be interesting is to see what that
modem produces on another telephone line. Say the neighbors
line, so that it has much the same characteristics.
Regardless, all of the above is just a guess, and cannot be
relied upon as correct.
--
Floyd L. Davidson <http://web.newsguy.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd@barrow.com |
|
| Back to top |
|
|
LoadHawg
Guest
|
| Posted:
Sat Jan 15, 2005 9:12 pm Post subject:
Re: Dirty Phone Lines or ...? My Next Step? |
|
|
....interesting development...
I did find one modem that seems free of this problem so far. I have TWO old Supra 28.8s. THe first one I tested with
apparently has been flashed or upgraded to a 33.6 - using this one I reported I could duplicate the problem. However the
2nd one I tried overnight - it's only a 28.8 - no errors, zippo, nada and what's more important is no problems w/ the
connection spiraling down to no throughput. I've downloaded extensively w/ it over a period of many hours and hit those
websites that easily cripple my 56K and similar 33.6 throughput. Just an FYI if this helps give any clues. It turns out
my old trusty standby backup I keep in a shoebox is indeed still 'trusty' (provided I use the right one).
Maybe there is a problem w/ teh modem h/w in the 56K despite the fact we don't see a clear error in the logs/diags and
it maintains original connectivity speeds. Again as a reminder, the speed is maintained, but the TX/RX lights trickle
down to only occasional intermittent flickers and so very very little data is let thru that essentially nothing can be
sent or received - not even a small email. The problem has been duplicated across 2 different PCs, 2 different sets of
serial/phone cables, 3 different ISPs and 2 different modems. The second Supra 28.8(33.6) I don't have any history on
and don't know if it was good before. We do get a LOT of lightning strikes out here. I've gone thru many cheapo modems
due to lightning strikes since moving out here before getting the Zoom 3048C (claims to have lightning protection and it
has lasted a fair bit longer than any other).
Anyway I'm kinda wondering if there is a fault w/ the Zoom 56K modem. |
|
| Back to top |
|
|
Floyd L. Davidson
Guest
|
| Posted:
Sun Jan 16, 2005 12:13 am Post subject:
Re: Dirty Phone Lines or ...? My Next Step? |
|
|
LoadHawg <not@chance.com> wrote:
| Quote: | ...interesting development...
I did find one modem that seems free of this problem so
far. I have TWO old Supra 28.8s. THe first one I tested with
apparently has been flashed or upgraded to a 33.6 - using
this one I reported I could duplicate the problem. However
|
Incidentally... I also have a pair of old Supra FAX Modems.
Before I retired I traveled to villages, and often needed to do
extended trunk testing with a modem! (Which is to say, I wanted
to read Usenet news, and could justify the 2-5 hour calls by
saying that when customers asked about trouble dialing out,
which they in fact often did, I would actually *know* how it
worked!)
Until recently there was still some of the old analog SCPC
(Scientific Atlanta and California Microwave installed in the
1970's) satellite equipment being used, not to mention that the
initial digital systems were even worse. So I had a lot of
chance to experiment with various modems over "impaired" long
distance trunks. The Supra modems were significantly better
than anything else that I tried. (I only had one USR Courier,
and never did take it on trips. I assume it would be just as
good.)
| Quote: | the 2nd one I tried overnight - it's only a 28.8 - no
errors, zippo, nada and what's more important is no problems
w/ the connection spiraling down to no throughput. I've
|
In that case it might be interesting to know what happens if you
configure another other modem to limit the connect speed to
28.8kbps, and try them again. There are significant differences
in the spectral output generated by higher speeds... but I'll
be darned if I can imagine just how that would cause a slow
decline from whatever the initial data rate is.
The other Supra modem would be the first to try. If it works
okay at 28.8kbps, try one of the others.
| Quote: | Anyway I'm kinda wondering if there is a fault w/ the Zoom
56K modem.
|
Because you've seen this with such a variety of equipment, but
particularly with the Supra modems as well as the Zoom modem,
I'm still inclined to think there is a problem with the
telephone line. But that isn't certain by any means!
I sure would like to see the diagnostics from a USR Courier on
that line. You don't have a friend that might loan you one???
--
Floyd L. Davidson <http://web.newsguy.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd@barrow.com |
|
| Back to top |
|
|
|
|
|
|
FAQ
Memberlist
Register
Profile
Log in to check your private messages
Log in