dslrocker3
join:2002-05-26
Toronto
4 edits | Help/Advice with my Line Statistics please
Hello. I have always had a problematic line but a few years ago, I was able to get them to cut some wiring dead-ahead to get a stable 1728/640 (interleaved) using my trusty old Speedtouch Home. A few months ago, I asked my ISP, Velcom, to ask Bell to examine my line statistics again because I had a different modem hooked up that was capable of RE-ADSL (reach extended) in the form of a Speadtream 5200 E242. They were able to changed by profile to 2496/800 fast path, with no to very few errors and the RX SNR usually ranged between 10 to 14 decibels.
A few days ago, the phone line "gremlins" have resurfaced. Interestling enough, I have always received better line statistics inside the house than when I hook my modem directly to the jack at the demarcation point (bridge tap at work here? extra 100 or 200 meters inside my house adding more distance making a better connection?). Before Bell changed the profile yesterday to 2496/640 interleaved, I went to the demarc and got about 1850/800 with SS5200 and about 1000/800 with ST516 (this modem really didn't like something at the demarc!)
- but a little higher with the SS5200 and a lot higher with the ST516 if connected inside the house.
You will notice that I have tried 2 different modems to rule that out as the problem. I have tried a different phone wire and have tried different jacks within my house. I have even hooked up to the demarcation point (with even worse results).
In the picture are the line stats from my ST516v6 and in the text below are statistics from my SS5200 E242. SS5200 E242 is a lot more aggressive in connecting at higher speeds but results in more errors. ST516 refuses to connect at higher speeds for the "bad" portions of the signal but that results in much more stability (is more fussy and won't connect at rx snr below 7.0). I have a Bell techncian coming today to look at the line.
Advice/thoughts/insight? Thanks.
ss5200 e242:
DSL Status
Status ATU-C Current Tx Rate
(bits/sec) ATU-R Current Tx Rate
(bits/sec)
UP 1984000 640000
DSL Statistics (accumulated at 15 minute intervals)
System
Time Tx
CRC Tx
FEC Rx
CRC Rx
FEC LOS SEF LOS
(sec) SEF
(sec) Err
(sec) Rx
(blocks) Tx
(blocks) SNR Atten.
02:35:36 0 0 0 173 0 0 0 0 40 18185 18185 5.0 56.5
02:30:28 0 0 0 694 0 0 0 0 115 53078 53078 5.0 56.5
02:15:29 0 68 0 927 0 0 0 0 134 53174 53174 6.0 56.5
02:00:28 0 0 0 818 0 0 0 0 141 53182 53182 5.0 56.5
01:45:27 0 0 0 559 0 0 0 0 80 53183 53183 6.0 56.5
01:30:26 0 0 0 308 0 0 0 0 98 53182 53182 6.0 56.5
01:15:25 0 0 0 331 0 0 0 0 94 53131 53131 6.0 56.5
01:00:25 0 0 0 431 0 0 0 0 94 53175 53174 6.0 56.5
00:45:24 0 0 0 539 0 0 0 0 79 53180 53181 6.0 56.5
00:30:24 0 0 0 127 0 0 0 0 45 53181 53181 6.0 56.5
00:15:23 0 0 0 1279 0 0 0 0 51 53082 53081 5.0 56.5
Totals 0 68 0 6186 0 0 0 0 971 549733 549732 N/A N/A
|
|
noisey
@videotron.ca
| Noise problem.
These things tend to happen in spring and fall on the lines outside.
On a stable line you should be able to get 3-meg interleave, or at worse the 2.5-meg interleave with 640 upload.
Only advice I can give is to tell the tech to make that SNR go up.
Have you noticed if you occupancy has gone up at all since the problem surfaced?
If so, tell him that. its a sign of a wire problem outside (maybe wet wires, maybe a voltage cross etc.) |
|
just noticed
@videotron.ca
| I didn't read the whole post you made, sorry.
yeah I would say there is something wrong with the wiring at the demarc or something wrong with the way its set-up.
doesn't make sense that you get 2.5 in the house and 1 at the demarc.
Some type of short for sure.
Let the tech know. |
|
Bicephale
join:2005-09-24
 ·TekSavvy Solutions..
| reply to dslrocker3
Hi,
This is about the best one can hope for if far away:
Is a higher profile worth the low SN ratio?, SammyJ, 2007-Aug-30
I believe the trick here was to pick a proper profile
which was probably equivalent to Spectral Shaping...
Here's a post where i mentioned some very puzzling
behaviour relatively to internal wiring:
»Re: Optimum phone cable length
...and here's another with more paradoxal results:
»Re: SpeedTouch 516 Tweaking :)
In your situation i'd do serious homework 1st then i
would consider asking for a higher profile with this
tool at hand:
%20.GIF/thumb.jpg)
From the ground up!, Bicephale, 2008-Aug-30
Prove Bell's worker that you can force the MoDem to
connect when needed, the line may very well happen
to be quiet as he pays you a visit and become quite
catastrophic on evenings, which is what i described
in that later thread... |
|
dslrocker3
join:2002-05-26
Toronto
4 edits | reply to dslrocker3
Let's just say that a few years ago, I was told by a field technician that the line was trouble waiting to happen. I think the technician said something about voltage on the line but that they were not able to fix it/not cost feasible. Supposedly, the pair with the strange voltage readings is the one that is able to provide the higher speed connection but I was essentially warned that it could be problematic in the future. The other pair on the pole behind my house was very clean but couldn't even provide a 1Mbps connection.
When things are working properly, the results at the demarc and at a jack inside the house are usually about the same. But when I check yesterday, espseically with the ST516, that's when I noticed an extreme difference (ie. worse connection at demarc). How is this possible? |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
2 edits | If you've been told that you could "bypass" your house wiring
by connecting directly at the demarcation point then now you
learned the truth about it: it's fundamentally wrong because it
doesn't account for the "antenna effect" of your wiring hidden
behind walls - which can amount to a couple hundred feet in a
private home!... Centralized filtering eliminates confusion, long-
term noise curves reveal daily patterns if any and hence point
at potential local issues. Thomson's SpeedTouch 5x6 combined
to TerTech's 'DSL_StatScope' professional-grade utility is quite
the most user-friendly way available to acquire a better overall
perspective of the situation... Beware of SNR Margins as these
can remain stable for days if not weeks and never give a single
clue about what's really taking place. 'DMT' captures are great
for a quick start, "Diagnosis" mode is powerful when it becomes
more serious but 'DSL_StatScope' is much less involving, IMHO.
This is TerTech's 'DSL-StatScope' thread, for your convenience:
%20.GIF)
Yet Another Modem Statistics tool, Bicephale, 2009-Apr-10
Addendum:
Oh, and 'DSL_StatScope' can record logs which other
readers can "PlayBack" and examine as they see fit.
Also, this short thread illustrates the next step:
The customer's own wiring, Bicephale, 2007-Jun-13
Red is bad, green is good... |
|
dslrocker3
join:2002-05-26
Toronto
4 edits | It's interesting how you talk about an antenna effect. When I troubleshooting years ago, I decided to change some of the wiring in the house. I removed some wiring that was causing a problem (maybe a staple in through the wire?) and ran a new line up to my modem.
I wasn't willing to drill through the floors and walls so what I did was bridge from a downstairs jack with cat5e to where I wanted to place my modem on the second floor. The downstairs jack where my connection is bridged acts very strangely. To get the best connection and to get the second floor jack up to par with the other jacks throughout the house, I have to plug in an extension cord into the first floor bridged into jack (and it has to be a certain, specific cord for some reason) and it must be left unfiltered with no device hooked up to it. This adds about 0.5 dB attentuation but gives a better connection. So I guess this all means that less attentuation/least amount of wiring is not always better, depending on many factors. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| When i moved in, my MoDem couldn't be located conveniently
so an HPNA EtherNet extender is the solution i chose to avoid
wiring issues which i just didn't want to address back then. It
would be profitable to remove everything but the DSL device,
electrically speaking - which means using Centralized Filtering
is a must if you haven't tried it yet! Long phone cabling can't
but cause trouble, if it helps that's because something is dead
wrong, as i illustrated in one of the linked threads from above.
In the end, what you should go after is lower noise and hence
fewer disconnections. It doesn't matter what the SNR Margin
is as long as you remain connected... My SpeedStream works
reliably below 6 dB, my SpeedTouch doesn't. Many parameters
may appear to be relative in the DSL world, that's why i prefer
to compare curves these days: i've found better. |
|
dslrocker3
join:2002-05-26
Toronto
3 edits | reply to dslrocker3
Bell didn't even bothered to show up to day for the scheduled appointment today. I know they showed up yesterday unannounced but told a person at my place that because I (the person who uses the internet here) wasn't home, he would just come back tomorrow (which was today, Saturday, during the originally scheduled time). They closed the ticket and said that conclusively declared that I have a defective modem (even though I have 3 of them sitting beside me) without checking anything at the demarcation box or by climbing the utility pole behind my house. I wouldn't be surprised if the technician's log claims that he came at the proper scheduled time, when in fact he didn't.
Oh well, another trouble ticket has been opened. I'm skeptical now what will happen and I have a bad feeling that I will simply be changed to 1728/640 interleaved even though I was stable for 3 months at 2496/800 fastpath with a RCO that wasn't maxed out (SNR varied from day to day from about 10 to 14 dB). |
|
The New Bell
@cgocable.net | reply to dslrocker3
Sounds like my adventures with a no show.
»[Internet] My Bell Odyssey |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| reply to dslrocker3
Hi,
Sorry to hear that!!! Perhaps your DSL device isn't holding
at 10+ dB 100 % of the time and that's how Bell came to
close your ticket, i sort of wonder. This is the problem i'm
having with single snapshots: generally speaking, we can't
be sure.
Look closely at the 3rd (grey) spectral envelope graphic:
...now look at yours:
Based on this i honestly believe that it's reasonable to ask
for an opportunity to try 3008/640 Kbps Interleaved for a
while - if that profile happens to exist, of course. You can
use 'DMT' to force the unit to reconnect but maybe you'd
be better not to mention the "tweak" word to anyone...
Invent a little white lie for them, say you think the MoDem
disconnects when you have something cooking and you're
too busy to mind about it when it ocurs... It's a real issue,
i've demonstrated the "Cooking Hour Syndrome" long ago:
A wire is a wire is a wire, Bicephale, 2007-Sep-8
%20.GIF)
Tweaks, Bicephale, 2008-Jul-7
Then, with a higher profile, it would make sense to fix your
wiring - at least it should be somewhat more motivating...
 |
|
dslrocker3
join:2002-05-26
Toronto
1 edit | reply to dslrocker3
LOL, I guess that I really shouldn't leave the SNR target set to 1dB since there is an open trouble ticket. I was surprised that on a speed test, I was still able to get 2000/500 Kbps even with all those errors.
On a side note, I've noticed that this ST516 is much better at error correction than the SS5200 E242. The E242 seems fine with interleaved turned on but it's horrible on a high error line if you are on fast path, at least from my observations. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| Have you considered disconnecting everything at the
demarcation point to test your MoDem from there? It
is a tweak you're showing here, right? Hoping for some
happy ending, i guess...
|
|
dslrocker3
join:2002-05-26
Toronto
4 edits | I haven't done that since the profile change that turned interleave on.
When the profile was set to 2496/800 fast path, I was getting about 1850/800 with the SS5200 E242 and about 1000/800 with the ST516v6 at the demarcation box's jack (about 2000/800 and 1700/800 inside the house).
Yes, that last picture was with the tweaked settings. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| Oh, and when i write nothing else connected i mean something like this:
By the way, did you have to bring your 'DMT' gliding cursor up or down?...
 |
|
dslrocker3
join:2002-05-26
Toronto
3 edits | I turned down the slider for the target noise margin forcing to modem to pretty much accept any and all signals, regardless of how bad they were.
I haven't tried there yet. With evertyhing disconnected, the only difference from the inside the basement and the grey box would be about 6 feet of thick wire. Could the results actually be theat much better at the bolts in my basement when it is fed by the same jack that I already tested at in the grey box? Long ago, when I was trying to get a stable 1728/640 with a non RE-ADSL modem, I used a phone jack and hooked it up in the basement by using an old ethernet cable that was no longer being used because the clips on the end were broken. I'll have to comptemplate perhaps doing that again - or perhaps "sacrificing" a phone cord by chopping one of the ends of it.
My next step might be to go back to the outside grey box demarcation point and feed an cable like I did before directly to the bolts there, except on Bell side of the demarc making directy contact with the drop wire - although I strongly suspect that Bell really doesn't like it when you do stuff like this. |
|
ex bell
@cgocable.net
| said by dslrocker3  :My next step might be to go back to the outside grey box demarcation point and feed an cable like I did before directly to the bolts there, except on Bell side of the demarc making directy contact with the drop wire - although I strongly suspect that Bell really doesn't like it when you do stuff like this. They aren't lurking out there now are they?
Sneak out quick and wire a jack right to the drop wire.
You can wire on to the drop or bsw for a bit unless there is a storm in the area.
Bell doesn't care as long as you put it back together before you have them come there for a repair.
|
|
dslrocker3
join:2002-05-26
Toronto
4 edits | reply to dslrocker3
Update: Bell just changed the profile again. It started off at 2496/800 fast path (with enough overhead to probaby be fine at 3008/800 or maybe even 3.5Mbps downstream), then 2496/640 interleaved, and now it is at 1728/640 interleaved with a downstream SNR of 9.00 dB, attentuation 55.5 using my 5200 E242 and 1728/640 rx snr 8.0, attentuation 58.0 using my ST516v6. However this profile change didn't really accomplish anything as I could have simply limited the connection rate using my other modem (Speedtouch 516)
I know I'm being cynical now but I have funny feeling that a technician isn't coming this time. On the positive side, my RX corrected errors have dropped form the thousands per 15 minute intervals to only 23 in the last 8 minutes. This still doesn't address the sudden loss of signal quality that resulted in drop of about 1000Kbps in attainable rate but at least it's stable now. |
|
dbsanfte
join:2005-03-15
Montreal, QC
·Colbanet
1 edit | reply to dslrocker3
FECs are just flipped bits that get automagically fixed via the wonders of parity. They don't cause retransmission of frames, and so don't affect speed. It's the CRC errors you need to worry about.
I get somewhere around 7 million FECs per hour, and yet only maybe 20 CRCs in that period, and a solid connection at 3.5dBm. |
|
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| reply to dslrocker3
Hi,
If you have a demarcation box with this type of
connector module then it's perfect for testing:
When in use (with its protective lever aside) the
"testing" RJ-11 outlet is the only active element:
the pair of silver switches below it will disconnect
everything else attached to the orange section.
There should be no paradoxal behaviour observed
when testing from that location. |
|
