tcstone
join:2002-03-12
Glendale, CA
 ·Packet8
 ·Pacific Bell - SBC
| ADSL2+ at 10,400 feet
Hello:
Anyone with line/provisioning experience, can you offer an opinion for me?
I have a Covad ADSL2+ line just installed. I overheard my loop length is 10,400 feet (though preliminary estimates put it more like 8600 feet).
I don't know if these stats improve with more time connected to the DSLAM, but at first power-on / sync up, here are my stats:
(DS = Downstream, US = Upstream)
DS Speed: 4213 kbps
DS Latency: Fast
DS Output Power: 20 dB
DS Attenuation 50 dB -- (Ouch!)
DS SNR Margin: 6 dB -- (double ouch!)
Trellis Coding: Disabled
US Speed: 1023 (saving grace!)
US Latency: Fast
US Output Power: 12 dB
US Attenuation: 29 dB (that's more like it)
US SNR Margin: 9 dB (hmmm)
Framing Structure: -
Speed test results are about 3450 kbps down and 845 kbps up.
I'll start by asking a foolish question - at 10,400 feet should my DS stats be that bad?
Can the ILEC do some work to 'condition' my line? How could I get them to do that?
If the line is not improvable, are there any other settings that could improve DS speed? I would not want to lose any upload speed in the process though.
Anyway, the new line is much better than the DSL I have with the ILEC which gives speeds around 2500 down 420 up.
Thanks for your thoughts.
-Casey |
|
LBDSL
Lightning Bolt
VIP
join:2002-01-07
Auburn Hills, MI
| said by tcstone  :I'll start by asking a foolish question - at 10,400 feet should my DS stats be that bad? Yes
said by tcstone
Can the ILEC do some work to 'condition' my line? How could I get them to do that?[/BQUOTE :No said by tcstone
If the line is not improvable, are there any other settings that could improve DS speed? I would not want to lose any upload speed in the process though.[/BQUOTE :No said by tcstone
Anyway, the new line is much better than the DSL I have with the ILEC which gives speeds around 2500 down 420 up.
[/BQUOTE :You should lower your package to 3.0/768, if your ISP offers that speed package. --
Lightning Bolt Technologies |
|
Doctor Olds
I Need A Remedy For What's Ailing Me.
Premium,VIP
join:2001-04-19
1970 442 W30
clubs:
edit:
December 20th, @05:03AM
| reply to tcstone
said by tcstone :I don't know if these stats improve with more time connected to the DSLAM, but at first power-on / sync up, here are my stats: No, they actually degrade over time.
Who is your ISP?
--
What’s the point of owning a supercar if you can’t scare yourself stupid from time to time? |
|
tdumaine
join:2004-03-14
Redmond, WA | Answered in his other thread on the same topic as this one (one of my pet peeves)
»what are safe mode speeds on an ADSL2+ connection? |
|
tcstone
join:2002-03-12
Glendale, CA
·Packet8
·Pacific Bell - SBC
| Sorry! I guess in essence it is the same topic on two threads... I perhaps should have edited the title of the topic (rather than post another thread) since the first couple of posts answered the 'safe mode' question (not sure if that's possible though).
-Casey |
|
tcstone
join:2002-03-12
Glendale, CA
·Packet8
·Pacific Bell - SBC
| Update:
DL speeds had dropped to around 1600kbps but uploads still at 1000. Had large fluctuations in SNR and attnuation on the DL side, but the line did stay up all the time.
Covad tech was out today. He did a test where each leg was shorted to ground (separately). Result was that one leg had some inductive fault on it, it did not short properly.
So, hopefully ATT can fix the fault an my line will perhaps sync more consistently.
-Casey |
|
tcstone
join:2002-03-12
Glendale, CA
·Packet8
·Pacific Bell - SBC
edit:
January 8th, @10:02PM
| Re: ADSL2+ at 10,400 feet / ATU48V
Another Update:
ATT tech came out, said the line was fine. Today second Covad visit finds the same problem on the line, an 'inductive fault' on one of the legs.
I'm guessing that this inductive fault is related to the fact that, compared to ground, there is a -48V DC current on one of the legs. From what I've been told, a 'dry loop' adsl connection should have no DC voltage on it.
Covad does not seem to think that they have any control over this voltage when asked to check if they can turn it off. There is no dial tone on the line however.
They have arranged an MPOE (or something like that) for tomorrow, so I'll report back after that.
A few things I learned today from my 'Covad' tech:
He works for 'CSI' a company which Covad has subcontracted to do these installs in this area (Southern California).
My 'Line Stress' is 28db which I guess is not bad.
Finally, looking over the tech's shoulder on the NOSS system which is used to configure things on my DSLAM port, the 'Technology' being used for my line is called ATU48V.
Does anyone know what that means? I've certainly already make the uneducated jump that the 48V on there refers to the line power which maybe should not be there.
Thanks.
-Casey |
|
LBDSL
Lightning Bolt
VIP
join:2002-01-07
Auburn Hills, MI
| Raybro, may be able to pop in and correct me if I'm wrong, But I'm pretty sure that is the DSLAM card you are plugged into, and with ADSL2+, Covad uses 48 port DSLAM cards. which is what the 48 stands for: (Sample from one of our circuits)
DSLAM Trunk Status: active
Technology: atu48v
Card Status: equip
Port Status: is
Actual Port Speed: 15005 kbps Downstream / 1023 kbps Upstream
--
Lightning Bolt Technologies |
|
Raybro
VIP
join:2000-11-04
Back home | That is correct.
That is the Samsung model number for that card. |
|
tcstone
join:2002-03-12
Glendale, CA
edit:
January 11th, @06:13PM
| Thanks.
Raybro, would having -48V DC (approx. voltage) on my non-Earthlink, non-LPVA, dry-loop ADSL2+ line cause any problems?
If you do 'covad direct' support, please see my post there. Thanks.
-Casey |
|
Raybro
VIP
join:2000-11-04
Back home
| I don't do ADSL2 in my markets but if you have a non voice ADSL2 line then there should be no voltage.
It would seem that they may be crossed to a working pair.
--
Statements made by me are not the views or practices of my employer and they cannot be held responsible for any statements I may make. |
|
Splitpair
Premium
join:2000-07-29
Cow Towne
 ·T-Mobile US
·T-Mobile US
·AT&T Southeast
edit:
January 14th, @08:23AM
| reply to tcstone
said by tcstone :My 'Line Stress' is 28db which I guess is not bad. If that was tested on an open pair and is a measurement of longitudinal balance then 28 dB indicates a way out of balance loop. There are many factors affecting balance but a rule of thumb is one wants at least 57 dB of LB.
Finally, looking over the tech's shoulder on the NOSS system which is used to configure things on my DSLAM port, the 'Technology' being used for my line is called ATU48V. Does anyone know what that means? I've certainly already make the uneducated jump that the 48V on there refers to the line power which maybe should not be there. While the -48 battery could be normal for your loop it shouldn’t be there while testing as the pair should have been opened at the DSLAM to be properly tested. The voltage could be foreign battery as in a crossed pair or it could be the tech was into the wrong test point and didn’t have the loop he was testing opened from the DSLAM which would explain the battery and the low LB. I have seen that happen more times than I care to remember.
BTW I do find -48 without dial-tone on many COVAD loops heck I even had a COVAD tech (Brian) tell me the voltage was not supposed to be there and insisted I clear it I had the CO open the loop and no voltage so we then went to the CO and I showed him at the COLO point the voltage was from the COVAD DSLAM. That was an expensive and embarrassing lesson for him.
Wayne
--
Yeah, there's a storm on the loose, sirens in my head
Wrapped up in silence, all circuits are dead
Cannot decode - my whole life spins into a frenzy
|
|
tcstone
join:2002-03-12
Glendale, CA
·Packet8
·Pacific Bell - SBC
| If that was tested on an open pair and is a measurement of longitudinal balance then 28 dB indicates a way out of balance loop. There are many factors affecting balance but a rule of thumb is one wants at least 57 dB of LB.
Perhaps I got my terminology wrong... I think that test was a measurement of line loss when open. I never heard that the line had balance issues.
BTW I do find -48 without dial-tone on many COVAD loops heck I even had a COVAD tech (Brian) tell me the voltage was not supposed to be there and insisted I clear it I had the CO open the loop and no voltage so we then went to the CO and I showed him at the COLO point the voltage was from the COVAD DSLAM. That was an expensive and embarrassing lesson for him.
Wayne
Sadly, I think that may have been repeated here. I was told by at least four people (either covad or ATT people) that there should be no voltage on the line. But even after the MPOE and a trip to the CO by the most competent Covad tech I've encountered on site, the voltage remains. Perhaps these newer Samsung DSLAMs that support ADSL2+ just do that and there's no way to turn it off.
The voltage was more an issue that I picked up on, but the actual reason Covad spent a lot of time getting ATT to 'fix' the loop is that one of Covad's tests kept failing -- the test where you short each leg to ground at the site and the Covad equip measures whether it shorts properly. On this test there was repeatedly an error, an "Inductive Fault" on one leg.
After watching my line stats vary quite a bit and seeing my ADSL2+ downstream speed vary between 1600kbps and about 2500kbps, and feeling the frustration that I could not talk to anyone with advanced equip / ability to engage or create the best port profile for my loop, I found a Dlink ADSL2+ modem at Fry's yesterday.
That modem allowed me (unlike the supplied modem) to specify the type of connection it would support: eg G.DMT, G.Lite, ADSL2, ADSL2+, Annex M and L, etc.
I got the best sync rate set to G.DMT, about 5500 down and 990 up. Overnight (a time that would often cause the ADSL2+ setting to lose speed) the line did stay up, as revealed by the Smoke Ping I have running. I am a bit concerned about the variability of the ping times though.. Take a look:
»/r3/smokeping.···bfad08f7
My guess is that something in the ADSL2 Adaptive Rate feature is causing the speed to plummet, but I'll have to keep watching the line in G.DMT mode to see how happy that remains.
Thanks.
-Casey |
|
tcstone
join:2002-03-12
Glendale, CA
·Packet8
·Pacific Bell - SBC
| Here is what the last Covad tech noted when he was on the MPOE:
quote:
ILEC loop 24 db of stress on f1 18 db of stress on f2, 28 db of stress on
ILEC loop @ 10.4 k feet
I found 9 db of stress inside the cage
I found stress fluctuating at the NID met with ILEC technician he opened the
loop at the F2, the found 18 db of stress on the F2. Called to the frame and
opened the loop at the SPOT frame found 24 db of stress on the F1 when the
circuit was open at the OVC. I opened the circuit in the loop test
application with the heat coils back in at the and saw 43 VDC with 28 db of
stress on the F1.
I arrived at the CO at 1530 able to synchronize my test CPE at speed with
good snr margins.
Next step:
This loop is too long for 8.0 mbps. We need to downgrade the speed to 3.0
mbps, and put up temporary cross connects, and have ILEC cut to new F1
Thoughts?
-Casey |
|
LBDSL
Lightning Bolt
VIP
join:2002-01-07
Auburn Hills, MI
| I think that is what a few of us have been telling you for awhile, you need to downgrade your service. 
--
Lightning Bolt Technologies |
|
Doctor Olds
I Need A Remedy For What's Ailing Me.
Premium,VIP
join:2001-04-19
1970 442 W30
clubs:
| reply to tcstone
said by tcstone :Here is what the last Covad tech noted when he was on the MPOE: quote:
ILEC loop 24 db of stress on f1 18 db of stress on f2, 28 db of stress on ILEC loop @ 10.4 k feet
Next step:
This loop is too long for 8.0 mbps. We need to downgrade the speed to 3.0 mbps, and put up temporary cross connects, and have ILEC cut to new F1
Thoughts? I agree 100% with the Covad Tech and hope that 3.0 Mbps is slow enough to be stable at your line distance and conditions. You may end up with 1.5 Mbps instead if 3.0 Mbps exhibits similar instabilities.
--
What’s the point of owning a supercar if you can’t scare yourself stupid from time to time? |
|
tcstone
join:2002-03-12
Glendale, CA
·Packet8
·Pacific Bell - SBC
edit:
January 20th, @11:40PM
| reply to tcstone
Re: ADSL2+ at 10,400 feet
Covad ADSL2+ line 10,400 feet from CO |
So, as repeatedly advised, my speed was locked down to 3/768. Above is what the line stats look like on my DLink modem at that setting tonight during my final test.
Almost 21dB SNR margin on downstream, and 16dB SNR margin on upstream. 51dB attenuation on downstream, 28dB on upstream. Attainable rate 6240 down, 1184 up. Current rate 3008 down, 768 up.
[side note- stats looked best at G.DMT compared to ADSL2+ or ADSL2 modes, on the DLink I can choose. I also did these server/error tests with the supplied Broadxent modem with the same results]
Should be great, yes? Nope. For a test I'm running an FTP server directly on this Covad line (no router, just right into my server), and I'm downloading on my ATT line that I've had for years.
Checksum of source file:
5c1bb06eaf49c369d110b6dae669e086 *FoD_6m3_StereoFull.zip
Checksum of downloaded file (downloaded three times back to back):
281fa15724890705f9acaeb02113176d *1-FoD_6m3_StereoFull.zip
c58ac0c35a11c358a13c1d0a6bd49c6d *2-FoD_6m3_StereoFull.zip
9303fe1b8c13239136de5c7d7e36bfed *3-FoD_6m3_StereoFull.zip
[note- I renamed each file after the download completed, but that does not affect the checksum.]
That's even worse than other tests I've done where usually only one download out of three or four would be wrong. This test file was a bit over 25MB. I have done other tests using different setups including different files, different computers doing the serving, and different connections doing the downloading.
Obviously, this is the worst possible thing you could have on any connection, especially one you plan to transfer files with.
Another problem that I mentioned in another thread, is that the uploads do not sustain speed - at this connection rate the line will pull a constant approx 0.62 Mbit/sec, however TCP pauses limit this over the course of a few minutes to an average of 0.49 Mbit/sec. This was about the same percentage I experienced when I was synced at 1024 Kbps. When my ATT line (probably synced at 512 on upstream) hosts the server it pulls a constant 0.42 Mbit/sec with very rare interruptions.
My experience trying to get this line happy over the past month or so leads me to the sad conclusion that Covad really is not in control of these new Samsung ADSL2+ DSLAMs -- for example, they apparently have no way to adjust target SNRs as I have read is possible. The NOSS system that the techs have seems not to be really updated to fine tune things -- they can set 15/1, 10/1, 8/1, 6/768, etc speeds but that's about it. I realize that I don't know exactly what I'm talking about, but I'm just surprised that nothing could be done to make this work at any speed.
I certainly hope that others have better luck with this... I'm sure it is working well for some people. I would love to know why there are uncorrectable errors on uploads! That does seem like the kind of issue that could go undiscovered by many customers for a long time.
-Casey |
|
