Howdy
Some of you may remember my goal of getting a 4Mb TalkTalk ADSL1 connection from being interleaved to the hilt, over to fastpath.
TT supplied a DSL-3680 for this connection. I'm not sure what chipset is in this but something tells me it IDs itself as IFTN but I can't seem to verify this. Browsing through the ADSL commands now, I am none the wiser. What I can tell you, is the syntax matches the TI stuff I've seen in the ZyXELs, so maybe it's one of them? The DSLAM reports IFTN 0X71C6 which appears to be standard for TT LLU. Perhaps someone can advise if they know different?
Some output with the DSL-3680 acting as just a wireless access point if it helps anyone:
D-Link> wan adsl fwv
DMT FwVer: 3.20.19.0_TC3087 HwVer: T14F7_11.2
D-Link> wan adsl vendor
near-end vendorid: bb0c
far-end vendorid: 8043
D-Link> wan adsl nearituid
near end itu identification: 26 0 54 43 54 4e 0 0
So, rewind a few months.
1. The connection used to drop daily
2. Massive errors were always visible on downstream and upstream
I bought an HG612 and threw it on the line to try my luck. I observed the following almost instantly:
1. The synch speed was considerably lower at 6dB (
HG: 4000~ vs DL: 4500~)
2. The reported downstream attenuation was quite a bit higher (
HG: 57dB vs DL: 53dB)
3. The ping was 2-3ms higher with the HG attached (
quite annoying at the time, native ping to bbc.co.uk was 40ms)
4. The upstream errors vanished with the HG (
PhyR magic? Or is the DLink bollocksed in some way?)
Here's an example of the connection stats a few weeks ago. I think it was a fresh resynch but still heavily interleaved:
# xdslcmd info --show
xdslcmd: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 8000
Max: Upstream rate = 772 Kbps, Downstream rate = 4832 Kbps
Channel: Interleaved, Upstream rate = 608 Kbps, Downstream rate = 4160 Kbps
Link Power State: L0
Mode: G.DMT
TPS-TC: ATM Mode
Trellis: ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 6.6 6.0
Attn(dB): 56.0 31.5
Pwr(dBm): 18.3 12.6
G.dmt framing
K: 131(0) 20
R: 6 16
S: 1 8
D: 64 2
Counters
Path 0
SF: 8377 8312
SFErr: 1 0
RS: 569570 70652
RSCorr: 52 0
RSUnCorr: 9 0
Path 0
HEC: 1 0
OCD: 0 0
LCD: 0 0
Total Cells: 1394767 0
Data Cells: 1975 0
Drop Cells: 0
Bit Errors: 392 0
ES: 1 0
SES: 0 0
UAS: 49 49
AS: 136
Path 0
INP: 1.40 0.72
PER: 1.75 1.75
delay: 16.00 4.00
OR: 32.00 32.00
Bitswap: 6 0
#
I am currently running with the faceplate removed but I noticed it was not perfectly flush. Upon further investigation, there appears to be loads of 'connection joints' wedged inside the master socket. They are all now disconnected but it makes me question the quality of the terminations natively to the master...
Favouring brevity, what currently appears to be working for me, although far from completely fixed:
1. Removed faceplate
-Not sure how much this helped or is currently hindering. It appears to be an XTE 2005 Clone, identical to 'our equivalent' pictured here:
http://www.run-it-direct.co.uk/xte2005cloneadslfaceplate.html2. Shielded RJ11 cable
-Bought 3 of them from here:
http://www.tandyonline.co.uk/high-speed-rj11-dsl-cable-0-5m.html3. +1dB extra DS SNRM
-Via the handy option in DSLstats. I love this tool. Pretty sure it was this combined with the shielded cable that got the interleaving level dropped, before I realised the DECT was shafting me.
4. Removed DECT phone from line
-It was permanently robbing 0.3dB SNRM. If the TV was turned on, it became even greedier and took 1.3dB! It was in close proximity to the TV and the master socket, though.
5. xdslcmd configure --CoMinMgn on --i24k off --phyReXmt 0x3 --monitorTone off --dynamicD off --dynamicF off
-Although I couldn't replicate it every time, turning i24k off seems to reduce my interleaving depth. Drastically*.
-I enabled PhyR for both downstream and upstream, mostly for the lols. It was already enabled for upstream and who knows what features these various DSLAMs support these days. Broadcom might have licensed the feature to Infineon for all I know!
-I left CoMinMgn on, can't remember why. Well played my documentation skills!
-I turned the last three off on the basis I have no idea what they do. I also found another post on here with a very similar connection string so robbed most of that. Cheers whoever posted it
My current stats - connection online for 9 days and 11 hours - pretty sure that is a record for this line! Ping is now 24ms to bbc.co.uk at it's best and my gaming ping is considerably lower as a result. I'm still hopeless at all the games I play but meh
xdslcmd info --stats
xdslcmd: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 8000
Max: Upstream rate = 816 Kbps, Downstream rate = 4684 Kbps
Channel: Interleaved, Upstream rate = 640 Kbps, Downstream rate = 3904 Kbps
Link Power State: L0
Mode: G.DMT
TPS-TC: ATM Mode
Trellis: ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 7.5 6.0
Attn(dB): 56.5 31.5
Pwr(dBm): 18.4 12.6
G.dmt framing
K: 123(0) 21
R: 8 16
S: 2 8
D: 8 2
Counters
Path 0
SF: 48101578 48102197
SFErr: 10285 0
RS: 1635453643 408868674
RSCorr: 260454 0
RSUnCorr: 73108 0
Path 0
HEC: 51267 0
OCD: 489 0
LCD: 489 0
Total Cells: 3234241830 0
Data Cells: 474260267 0
Drop Cells: 0
Bit Errors: 4078306 0
ES: 3502 0
SES: 105 0
UAS: 214 214
AS: 816229
Path 0
INP: 0.25 0.69
PER: 1.75 1.75
delay: 4.00 4.00
OR: 32.00 32.00
Bitswap: 1079 1633
Total time = 1 days 12 hours 3 min 14 sec
FEC: 263578 0
CRC: 10292 0
ES: 3502 0
SES: 105 0
UAS: 214 214
LOS: 0 0
LOF: 0 0
Latest 15 minutes time = 3 min 14 sec
FEC: 8 0
CRC: 0 0
ES: 0 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 15 minutes time = 15 min 0 sec
FEC: 200 0
CRC: 11 0
ES: 7 0
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Latest 1 day time = 12 hours 3 min 14 sec
FEC: 6358 0
CRC: 302 0
ES: 156 0
SES: 3 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 1 day time = 24 hours 0 sec
FEC: 53685 0
CRC: 1721 0
ES: 399 0
SES: 23 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Since Link time = 9 days 11 hours 8 min 57 sec
FEC: 260454 0
CRC: 10285 0
ES: 3497 0
SES: 105 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
#
TODO:
1. Find out why I can't negotiate ADSL2 - I appear to be connected to a TT LLU DSLAM.
2. Figure out the weirdness in my HLOG
3. Fastpath time?
Note spikey upstream HLOG and weird slow peak on downstream start tones:
I wonder what my chances are of getting TT/BTOR to check the socket and the PCP?
Hopefully a quick job for someone but if there happens to be a finer strand of copper...
*interleaving did actually back itself off (down to 8 / 2 ) before I started messing with the command line options but shortly after went back up to 32 / 2.