Topic: ADSL observations and night time margin drop (Read 8543 times)

Nick_R · « **on:** September 17, 2009, 05:12:50 PM »

Hi.

I've been lurking in the forums for some time now, reading plenty of interesting discussions on ADSL and I thought it's about time I contributed to the debate and to see if anyone has anything to add to my little ADSL problem, which is the ADSL is now afraid of the dark. Firstly, here's a brief background of my equipment / line / ISP.

I currently have a netgear DG834v4 wireless router with the latest DG team firmware installed, currently running with no tweak (100% SNR margin). The router is running in wired mode only with the wireless disabled. It's connected to a BT installed faceplate and BT installed cable up to the back bedroom (installed back in the days when BT only did home installations for ADSL).

The phone line is connected to the Queensbury exchange, 2.27km direct, 3.22km by road and about 3.5km in length from when you could use the phone number to check. The line at my house is underground but I guess there is a section that's overground to go from the village down the hill to Clayton where the exchange is.

Below are my current line stats;

# adslctl info --show
adslctl: ADSL driver and PHY status
Status: ShowtimeRetrain Reason: 0
Channel: FAST, Upstream rate = 871 Kbps, Downstream rate = 5115 Kbps
Link Power State: L0
Mode: ADSL2
Channel: Fast
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 9.3 5.8
Attn(dB): 52.0 15.5
Pwr(dBm): 19.0 12.9
Max(Kbps): 6000 872
Rate (Kbps): 5115 871
G.dmt framing
K: 183(0) 27
R: 6 0
S: 1 1
D: 32 1
ADSL2 framing
MSGc: 51 12
B: 182 26
M: 1 1
T: 1 4
R: 6 0
S: 1.1385 0.9818
L: 1328 220
D: 32 1
Counters
SF: 739186 705494
SFErr: 188 2107
RS: 42133648 0
RSCorr: 1839780 0
RSUnCorr: 1567 0

HEC: 120 150
OCD: 0 0
LCD: 0 0
Total Cells: 144692670 143838589
Data Cells: 34102899 2612117
Drop Cells: 0
Bit Errors: 0 261

ES: 4091 311
SES: 0 139
UAS: 112 9052831
AS: 11993

INP: 0.57 0.00
PER: 16.22 17.67
delay: 9.10 0.24
OR: 28.10 8.14

Bitswap: 1172 0

I'm currently with Xilo (LLU) but was previously with plus net for 8 and a half years. I only got connected to Xilo yesterday, so it's still early days in terms of line performance, but with Plus Net my final IP profile was 3500 while at the moment, I can download at about 4.5 meg (and I didn't have to wait 3 days for it).

Over the day that I've had the connection I've noticed a few things. Firstly, the upstream speed has increased from the default ADSL max speed of 448, but so has the SF errors. I've asked for interleaving to be put on by default (currently set to auto) so hopefully the upstream will look a bit better. With plus net, I never had any upstream errors, but then it was half the speed so I'd still expect to get some with increased speed.

I've also had a quick look at all three modulation types, mainly to determine which will give me the best speed. ADSL1 was the slowest at about 4480, ADSL2+ came in at around 5000 while ADSL2 connects at about 100 – 200kpbs higher. Although I know that ADSL1 has a minimum of 2 bits per bin whereas ADSL2 has a minimum of 1 and ADSL2+ has double the bins. I also suspect that ADSL2 / 2+ manages it's bin allocation better than ADSL1. On routerstats, you don't see an SNR per tone line for bins that have no bits allocated, whereas ADSL1 draws a line over tones that don't have quite enough SNR to allocate two bits, or they previously had bits in but got bitswapped out. Maybe this reduces the overall power that ADSL2 / 2+ uses as ADSL1 would have a downstream power of around 19.4dBm, ADSL2 would be 19dBm and ADSL2+ would be 17dBm.

One big thing that I have noticed is that tweaking the downstream margin while in fast path mode (even though the router shows an interleaving depth of 32 for downstream) doesn't seem to produce anything except errors. The default margin is 9db and produces about 20 CRC errors per 15 minutes. I tweaked it down to 6dB and it produced about 5000 CRC errors per 15 minutes while connecting at 5971. One of the last things I did on Plus net on Monday was to see what the 'flat out' speed that I could achieve was. Setting the target SNR to 4.8dB, connecting at 6432kbps (interleaving is on) I got about 200 CRC errors per 15 minutes for the 7 hours that I left it on. Sadly, with an IP profile of 3500 I wasn't able to tell if sync speed would translate to an increased download speed.

I guess I should explain why I had a 3500 IP profile (it started at about 5000 when I first got ADSL Max), but over the years it slowly went down. The main problem started in Mid July. As soon as it would start to get dark the router (then a Draytek vigor 2600we) would re-sync from it's current speed of about 4100kbps and re-connect at 2880kbps. The first night I didn't think nothing of it until it happened again the second night. I then checked in the morning and the SNR margin had gone upto 20dB (from it's default of 12). Obviously my IP profile was shot by this stage, only 2000, so I started keeping a check of when the connection dropped and when the SNR margin would return, and sure enough it turned out to be when it starts to get dark and when it's getting light.

My first thought was a faulty street light, but the lights were not on at the time the connection drops, and they didn't come on for a good 5 to 10 minutes. In the morning the street lights were all off and had been off for a good 10 to 15 minutes. When the lights were on, there was no obvious light flickering or cycling. As my phone line is underground I don't really know which route it takes to the street cabinet, and that's assuming it goes to the one on the high street in the first place. I raised a fault with plus net and they (or BT) raised my target SNR margin to 15dB, which at least overcame the disconnections when it gets dark, but my IP profile only went upto about 2500. Faced with either paying £150+ for BT to come out and declare a 'no fault' or spending less to get a tweakable router to help me get some of that speed back I went for the router and over the last couple of months I got it to where it could withstand a 6 – 8dB drop in SNR margin and still provide me with good download speeds.

So, the error starts off at say 15dB and over a time of about 5 minutes it slowly slides down between 6 to 8dB to it's resting place, where it stays all night before taking another 5 minutes to go back up when it's getting light. I have also noticed that if I sync at too high a speed there can sometimes be a large spike of errors that can cause the router to re-sync, both at night and in the morning. This spike usually happens when the margin is nearly at the bottom of the downward slide and also when it's just started to rise in the morning.

I got my friend who lives up the road from me to check his router and sure enough, he has the same issue. I've set up router stats on his computer to log what happens, so when I get the log back it'll hopefully show the same error as mine. Though in principle, we already know it's the same error.

Eventually, I did notice that one street light across from where I live wasn't working (it would cycle at start up, give up and not try again). Then, I noticed the street light next to it was going out after 5 minutes. It would come on again, but then eventually it wouldn't try again and would stay out for the rest of the night.

I've reported both lights and I hope that when the council come to fix them, they'd give the electrics the once over to make sure they are ok. If this doesn't fix the issue then it's back to the drawing board.

Gosh, this has ended up been far longer than I expected, apologies to all who've fallen asleep.

If anyone wants to see any graphs then let me know and I'll try post some.

Thanks.

Nick.

Nick_R · « **Reply #1 on:** September 17, 2009, 08:24:06 PM »

Hello.

A quick addition to the previous message.

Here are the stats when the SNR margin dropped this evening at around 7:20pm;

# adslctl info --show
adslctl: ADSL driver and PHY status
Status: ShowtimeRetrain Reason: 0
Channel: FAST, Upstream rate = 871 Kbps, Downstream rate = 5115 Kbps
Link Power State: L0
Mode: ADSL2
Channel: Fast
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 2.4 5.8
Attn(dB): 52.0 15.5
Pwr(dBm): 19.0 12.9
Max(Kbps): 4432 872
Rate (Kbps): 5115 871
G.dmt framing
K: 183(0) 27
R: 6 0
S: 1 1
D: 32 1
ADSL2 framing
MSGc: 51 12
B: 182 26
M: 1 1
T: 1 4
R: 6 0
S: 1.1385 0.9818
L: 1328 220
D: 32 1
Counters
SF: 1858817 1773941
SFErr: 15483 2176
RS: 105952596 0
RSCorr: 5997461 0
RSUnCorr: 156762 0

HEC: 14920 181
OCD: 0 0
LCD: 0 0
Total Cells: 363740148 181189230
Data Cells: 34183381 2647144
Drop Cells: 0
Bit Errors: 0 329

ES: 4832 360
SES: 239 139
UAS: 112 9052831
AS: 30159

INP: 0.57 0.00
PER: 16.22 17.67
delay: 9.10 0.24
OR: 28.10 8.14

Bitswap: 4030 0

Sadly, even though the router held out with a SNR margin of around 2.5dB, the internet was effectively dead due to the errors, so, I re-connected, employed a 3dB tweak, to see what the error rates would be this time and here are the stats, complete with 15 minute intervals;

# adslctl info --stats
adslctl: ADSL driver and PHY status
Status: ShowtimeRetrain Reason: 0
Channel: FAST, Upstream rate = 856 Kbps, Downstream rate = 4292 Kbps
Link Power State: L0
Mode: ADSL2
Channel: Fast
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 5.6 5.7
Attn(dB): 52.0 30.0
Pwr(dBm): 19.5 12.7
Max(Kbps): 5056 856
Rate (Kbps): 4292 856
G.dmt framing
K: 124(0) 27
R: 6 0
S: 2 1
D: 16 1
ADSL2 framing
MSGc: 64 12
B: 123 26
M: 2 1
T: 1 4
R: 6 0
S: 1.8339 1.0000
L: 1108 216
D: 16 1
Counters
SF: 131570 117295
SFErr: 41 2331
RS: 4604974 0
RSCorr: 138286 0
RSUnCorr: 257 0

HEC: 35 184
OCD: 0 0
LCD: 0 0
Total Cells: 21379869 186742285
Data Cells: 104772 2675987
Drop Cells: 0
Bit Errors: 0 333

ES: 5488 373
SES: 862 150
UAS: 133 9052845
AS: 2111

INP: 0.34 0.00
PER: 16.04 18.00
delay: 7.33 0.25
OR: 34.89 8.00

Bitswap: 392 0

Total time = 10 hours 50 min 43 sec
SF = 2393027
CRC = 80257
LOS = 0
LOF = 0
ES = 5488
Latest 1 day time = 10 hours 50 min 43 sec
SF = 2393027
CRC = 80257
LOS = 0
LOF = 0
ES = 5488
Latest 15 minutes time = 5 min 43 sec
SF = 21439
CRC = 6
LOS = 0
LOF = 0
ES = 4
Previous 15 minutes time = 15 min 0 sec
SF = 56066
CRC = 13
LOS = 0
LOF = 0
ES = 12
Previous 1 day time = 0 sec
SF = 0
CRC = 0
LOS = 0
LOF = 0
ES = 0
15 minutes interval [-30 min to -15 min] time = 15 min 0 sec
SF = 54251
CRC = 208
LOS = 0
LOF = 0
ES = 20
15 minutes interval [-45 min to -30 min] time = 15 min 0 sec
SF = 55514
CRC = 52754
LOS = 0
LOF = 0
ES = 871
15 minutes interval [-60 min to -45 min] time = 15 min 0 sec
SF = 55453
CRC = 122
LOS = 0
LOF = 0
ES = 90
Showtime Drop Reason: 0
Last Retrain Reason: 0
#

Speed test shows a 3.5mb download speed, which is good.

Thanks.

Nick.

orainsear · « **Reply #2 on:** September 19, 2009, 10:17:27 AM »

Have you continued to see drops around the time when the street lamps are lighting up?

Faulty street lights can definitely cause issues with ADSL.

kitz · « **Reply #3 on:** September 19, 2009, 01:42:04 PM »

From what youve said it does look like the street lamps could well be the problem.

Nick_R · « **Reply #4 on:** September 19, 2009, 03:17:49 PM »

Hi.

Thanks for the replies.

Depending on what my SNR is just before it drops and the street lights come on determines whether the line holds out or re-syncs. On the first night on Xilo (Wednesday) it went down to 1.5dB before crashing (the internet light came on as red, had to umplug the router from the mains to get it back up). Second night it went down to 2.5dB, it held the line but there was effectively no internet due to the error rate until I reset it. Last night it went down to 0.5dB before I re-synced it (to avoid the red internet light again). It hopefully will hold out just fine at around 4dB SNR.

Interleaving has now been turned on and and target SNR is now 12dB, so hopefully that'll hold a decent connection. It's also fixed the upstream errors, below are the current line stats;

# adslctl info --stats
adslctl: ADSL driver and PHY status
Status: ShowtimeRetrain Reason: 1
Channel: FAST, Upstream rate = 748 Kbps, Downstream rate = 3999 Kbps
Link Power State: L0
Mode: ADSL2
Channel: Fast
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 12.8 6.0
Attn(dB): 52.0 30.0
Pwr(dBm): 18.5 12.8
Max(Kbps): 6788 752
Rate (Kbps): 3999 748
G.dmt framing
K: 139(0) 23
R: 16 4
S: 1 1
D: 64 32
ADSL2 framing
MSGc: 52 12
B: 138 22
M: 1 1
T: 1 4
R: 16 4
S: 1.1041 0.9729
L: 1123 222
D: 64 32
Counters
SF: 882990 839974
SFErr: 7 0
RS: 51213454 2868696
RSCorr: 452178 146
RSUnCorr: 88 0

HEC: 7 0
OCD: 0 0
LCD: 0 0
Total Cells: 134701992 498050978
Data Cells: 2215149 4006878
Drop Cells: 0
Bit Errors: 0 598

ES: 120 689
SES: 2 294
UAS: 164 9053430
AS: 14281

INP: 3.64 2.30
PER: 16.01 17.51
delay: 17.66 7.78
OR: 28.98 8.22

Bitswap: 1 0

Total time = 11 hours 59 min 25 sec
SF = 2668212
CRC = 1131
LOS = 2
LOF = 18
ES = 120
Latest 1 day time = 11 hours 59 min 25 sec
SF = 2668212
CRC = 1131
LOS = 2
LOF = 18
ES = 120
Latest 15 minutes time = 14 min 25 sec
SF = 54063
CRC = 1
LOS = 0
LOF = 0
ES = 1
Previous 15 minutes time = 15 min 0 sec
SF = 56193
CRC = 0
LOS = 0
LOF = 0
ES = 0
Previous 1 day time = 0 sec
SF = 0
CRC = 0
LOS = 0
LOF = 0
ES = 0
15 minutes interval [-30 min to -15 min] time = 15 min 0 sec
SF = 56193
CRC = 2
LOS = 0
LOF = 0
ES = 1
15 minutes interval [-45 min to -30 min] time = 15 min 0 sec
SF = 56193
CRC = 0
LOS = 0
LOF = 0
ES = 0
15 minutes interval [-60 min to -45 min] time = 15 min 0 sec
SF = 56256
CRC = 2
LOS = 0
LOF = 0
ES = 1
Showtime Drop Reason: 1
Last Retrain Reason: 1

For some reason, it still says 'fast' instead of interleaved even though interleave depth is 64 down and 32 up, with the increased margin and interleaving I guess I loose a bit of speed, a speed test shows 3.4meg download, so that's good.

I'm also sure that some of those dodgy counters survive a reboot, let alone a re-sync.

I'll post more stats this evening when the SNR drops.

I also hope the council don't take too long to come out and fix the faulty street lights, it's pretty dark on the path where they are, I also hope it does the trick with the ADSL.

Thanks.

Nick.

sevenlayermuddle · « **Reply #5 on:** September 19, 2009, 04:07:10 PM »

Quote from: Nick_R on September 19, 2009, 03:17:49 PM

I also hope the council don't take too long to come out and fix the faulty street lights, it's pretty dark on the path where they are, I also hope it does the trick with the ADSL.

if you haven't already done so, report the faulty lighting as a safety hazard - which it sounds like it is. Council processes sometimes seem to move more quickly over safety issues, whether due to good conscience or fear of litigation.

Nick_R · « **Reply #6 on:** September 19, 2009, 05:46:37 PM »

I reported them last Wednesday, they said 7 working days, but you're right, I'll phone them again on Monday and tell them it could be a safety issue as there's a black spot on the path, hopefully that'll get them out before next Friday.

Nick_R · « **Reply #7 on:** September 20, 2009, 03:26:11 AM »

Hello.

Night time stats, gotta figure out getting graphs the right size before posting though

stats for this sync run from 11:20am on saturday to 3am today.

# adslctl info --stats
adslctl: ADSL driver and PHY status
Status: ShowtimeRetrain Reason: 1
Channel: FAST, Upstream rate = 748 Kbps, Downstream rate = 3999 Kbps
Link Power State: L0
Mode: ADSL2
Channel: Fast
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 5.6 5.6
Attn(dB): 52.0 30.0
Pwr(dBm): 18.5 12.8
Max(Kbps): 5108 752
Rate (Kbps): 3999 748
G.dmt framing
K: 139(0) 23
R: 16 4
S: 1 1
D: 64 32
ADSL2 framing
MSGc: 52 12
B: 138 22
M: 1 1
T: 1 4
R: 16 4
S: 1.1041 0.9729
L: 1123 222
D: 64 32
Counters
SF: 3543927 3346084
SFErr: 47 0
RS: 205547808 1919546
RSCorr: 6364442 234
RSUnCorr: 583 0

HEC: 46 0
OCD: 0 0
LCD: 0 0
Total Cells: 536552441 573232751
Data Cells: 13582183 4907881
Drop Cells: 0
Bit Errors: 0 598

ES: 140 689
SES: 2 294
UAS: 164 9053430
AS: 56884

INP: 3.64 2.30
PER: 16.01 17.51
delay: 17.66 7.78
OR: 28.98 8.22

Bitswap: 5794 0

Total time = 23 hours 49 min 28 sec
SF = 5329149
CRC = 1171
LOS = 2
LOF = 18
ES = 140
Latest 1 day time = 23 hours 49 min 28 sec
SF = 5329149
CRC = 1171
LOS = 2
LOF = 18
ES = 140
Latest 15 minutes time = 4 min 28 sec
SF = 16789
CRC = 0
LOS = 0
LOF = 0
ES = 0
Previous 15 minutes time = 15 min 0 sec
SF = 56193
CRC = 0
LOS = 0
LOF = 0
ES = 0
Previous 1 day time = 0 sec
SF = 0
CRC = 0
LOS = 0
LOF = 0
ES = 0
15 minutes interval [-30 min to -15 min] time = 15 min 0 sec
SF = 56193
CRC = 0
LOS = 0
LOF = 0
ES = 0
15 minutes interval [-45 min to -30 min] time = 15 min 0 sec
SF = 56256
CRC = 2
LOS = 0
LOF = 0
ES = 1
15 minutes interval [-60 min to -45 min] time = 15 min 0 sec
SF = 56193
CRC = 0
LOS = 0
LOF = 0
ES = 0
Showtime Drop Reason: 1
Last Retrain Reason: 1

As you can see, the SNR margin has dropped since my earlier stats, but it seems to be holding up pretty good, download of 3.4meg maintained, again, apoligies for no graph, I'll try post some tomorrow when I get chance to resize them to reduce the space.

Thanks.

Nick.

kitz · « **Reply #8 on:** September 20, 2009, 03:50:54 AM »

>> With plus net, I never had any upstream errors, but then it was half the speed so I'd still expect to get some with increased speed.

Yep - Youre pushing that line a bit more on adsl2+ and its now reached it maximum on the upstream, which is why you will be seeing more upstream errors.

>> I've also had a quick look at all three modulation types

When doing that you may need to repeat the tests several times to be conclusive. Each 3dB of SNR is worth on average around 800kb of sync. Therefore 0.5dB of SNR change could easily be at least 100-200 of sync.

ADSL2+ should in theory give you a bit more sync speed due to the minimum bin allocation.. but this would only apply on those bins which load at 1 bit.. and thats only going to be a few kb anyhow.

Then theres the 'lines approaching 55dB', where adsl1 seems to be more stable than adsl2+.
So its going to be a bit suck it and see.. and what works best on your own line.

The increase in errors when SNRM is lower is normal.

RE your latest stats.. a night time dive of 6dB.

Does it dive suddenly or start a slow decline. Then in the am does it increase suddenly or take a while to creep back up?

orainsear · « **Reply #9 on:** September 20, 2009, 09:30:31 AM »

>>>I'll try post some tomorrow when I get chance to resize them to reduce the space.

A graphs covering the period where your SNR margin drops and another where it rises again would be useful.

Nick_R · « **Reply #10 on:** September 20, 2009, 04:25:25 PM »

Hi.

Graph production isn't working right at the mo, they're either too large or way too grainy, so, thought it would be quicker to fix up the relevent sections of the log file.

First part is when the SNR margin dropped, the entries are in 30 second intervals;

Sat 19 Sep 2009 19:00:24, , Rx-Noise=12.8, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:00:54, , Rx-Noise=12.8, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:01:24, , Rx-Noise=12.8, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:01:54, , Rx-Noise=12.7, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:02:24, , Rx-Noise=12.8, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:02:54, , Rx-Noise=12.7, Tx-Noise=6.1, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:03:24, , Rx-Noise=12.8, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:03:54, , Rx-Noise=12.8, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:04:24, , Rx-Noise=12.8, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:04:54, , Rx-Noise=12.6, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:05:24, , Rx-Noise=13.4, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:05:54, , Rx-Noise=13.5, Tx-Noise=6.1, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:06:24, , Rx-Noise=13.7, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:06:54, , Rx-Noise=13.3, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:07:24, , Rx-Noise=12.7, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:07:54, , Rx-Noise=12, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:08:24, , Rx-Noise=11.1, Tx-Noise=5.6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:08:54, , Rx-Noise=10.3, Tx-Noise=5.3, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:09:24, , Rx-Noise=8.4, Tx-Noise=5.5, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:09:54, , Rx-Noise=6.3, Tx-Noise=5.6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:10:24, , Rx-Noise=6, Tx-Noise=5.6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:10:54, , Rx-Noise=6.1, Tx-Noise=5.3, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:11:24, , Rx-Noise=6.2, Tx-Noise=5.2, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:11:54, , Rx-Noise=6.2, Tx-Noise=5.2, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sat 19 Sep 2009 19:12:24, , Rx-Noise=6.2, Tx-Noise=5.2, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,

As you can see, it doesn't just drop, it takes a few minutes to drop.

Below is the SNR margin rise in the morning;

Sun 20 Sep 2009 06:42:49, , Rx-Noise=6.6, Tx-Noise=5.6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:43:19, , Rx-Noise=6.6, Tx-Noise=5.6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:43:49, , Rx-Noise=6.6, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:44:19, , Rx-Noise=6.6, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:44:49, , Rx-Noise=6.6, Tx-Noise=5.6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:45:19, , Rx-Noise=6.6, Tx-Noise=5.6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:45:49, , Rx-Noise=6.6, Tx-Noise=5.6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:46:19, , Rx-Noise=6.4, Tx-Noise=5.6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:46:49, , Rx-Noise=6.4, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:47:19, , Rx-Noise=6.9, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:47:49, , Rx-Noise=7.4, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:48:19, , Rx-Noise=8.5, Tx-Noise=5.8, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:48:49, , Rx-Noise=9.4, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:49:19, , Rx-Noise=10.3, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:49:49, , Rx-Noise=10.4, Tx-Noise=5.8, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:50:19, , Rx-Noise=11.1, Tx-Noise=6, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:50:49, , Rx-Noise=11.7, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:51:19, , Rx-Noise=12.6, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:51:49, , Rx-Noise=13, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:52:19, , Rx-Noise=13.4, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:52:49, , Rx-Noise=13.4, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:53:19, , Rx-Noise=13.4, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:53:49, , Rx-Noise=13.5, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:54:19, , Rx-Noise=13.6, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:54:49, , Rx-Noise=13.7, Tx-Noise=5.8, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:55:19, , Rx-Noise=13.7, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:55:49, , Rx-Noise=13.7, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:56:19, , Rx-Noise=13.8, Tx-Noise=5.7, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,
Sun 20 Sep 2009 06:56:49, , Rx-Noise=13.8, Tx-Noise=5.8, Rx-Sync=3999, Tx-Sync=748, RxCRCLatest15min=0,

Again, it takes a few minutes to rise in the morning.

Thanks.

Nick

Nick_R · « **Reply #11 on:** October 27, 2009, 08:56:13 PM »

Hi.

Thought I'd post an update to my issue of the SNR margin dropping at night and recovering in the morning.

The council eventually fixed the two faulty street lights, turned out they were not the issue as the interference was still present on the medium wave radio and the SNR was still dropping at night.

Last Saturday both the drop in SNR margin and the interference dissapeared and hopefully *touches wood* will stay away, current night time stats are now as follows;

adslctl: ADSL driver and PHY status
Status: ShowtimeRetrain Reason: 0
Channel: FAST, Upstream rate = 763 Kbps, Downstream rate = 3999 Kbps
Link Power State: L0
Mode: ADSL2
Channel: Fast
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 14.2 6.1
Attn(dB): 51.5 30.0
Pwr(dBm): 18.5 12.8
Max(Kbps): 7004 764
Rate (Kbps): 3999 763
G.dmt framing
K: 139(0) 24
R: 16 4
S: 1 1
D: 64 32
ADSL2 framing
MSGc: 52 12
B: 138 23
M: 1 1
T: 1 4
R: 16 4
S: 1.1041 0.9955
L: 1123 225
D: 64 32
Counters
SF: 524406 498062
SFErr: 8 267
RS: 30415596 3947408
RSCorr: 4425 10845
RSUnCorr: 157 0

HEC: 7 592
OCD: 0 0
LCD: 0 0
Total Cells: 79862141 1912708550
Data Cells: 116720 59046299
Drop Cells: 0
Bit Errors: 0 9354

ES: 31 1638
SES: 0 987
UAS: 61 9099119
AS: 8467

INP: 3.64 2.27
PER: 16.01 17.92
delay: 17.66 7.96
OR: 28.98 8.03

Bitswap: 255 1

Connection has been up 2 and a half hours.

So, margin now 14.2db and the speed is still 3999kbps (this will hopefully be fixed soon). The SNR margin can get upto 16dB in the day and down to about 13dB at night, so it's looking a lot better than when it would drop to around 6dB.

In the end, I don't know what the cause was, but after 3 months of the issue, I hope it stays away.

Thanks for all your help and suggestions.

Will post my stats when the line is running at it's proper rate.

Thanks.

Nick.

Nick_R · « **Reply #12 on:** November 11, 2009, 01:27:02 PM »

Hi All.

After a few weeks of the line working normally and a lil bit of tweaking, below are the current line stats and a DMT screen shot after been connected at 6251Kbps for 7 days.

The SNR margin varies a little from 7dB in the day time right down to 3.5dB at night, though it doesn't normally go that far down, generally it goes to around 4.5dB. Download speeds are about 5.25Mb/s to 5.41Mb/s according to speedtest.net throughout the day and night. So all in all, I'm very happy indeed. I don't think I could get any more out of the line without it becoming unstable.

The real profile is actually 12dB margin as that's the best I could get. I could try for a 9dB profile and apply a less severe tweek, but if it's working then best leave it as it is.

Thanks for everyone's help and suggestions and long may the line live

Thanks.

Nick.

adslctl: ADSL driver and PHY status
Status: ShowtimeRetrain Reason: 0
Channel: FAST, Upstream rate = 749 Kbps, Downstream rate = 6251 Kbps
Link Power State: L0
Mode: ADSL2
Channel: Fast
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 7.0 7.3
Attn(dB): 51.5 30.0
Pwr(dBm): 19.5 12.9
Max(Kbps): 7448 752
Rate (Kbps): 6251 749
G.dmt framing
K: 224(0) 23
R: 14 2
S: 1 1
D: 64 32
ADSL2 framing
MSGc: 51 12
B: 223 22
M: 1 1
T: 1 4
R: 14 2
S: 1.1414 0.9708
L: 1668 206
D: 64 32
Counters
SF: 37942759 36310062
SFErr: 5457 110
RS: 2162737260 536107
RSCorr: 4665719 5338
RSUnCorr: 80398 0

HEC: 4676 193
OCD: 5 0
LCD: 0 0
Total Cells: 511250585 4166830247
Data Cells: 67621631 79688099
Drop Cells: 0
Bit Errors: 0 11717

ES: 3605 1888
SES: 11 1107
UAS: 156 9099499
AS: 617271

INP: 2.14 1.24
PER: 16.26 17.47
delay: 18.26 7.76
OR: 28.03 8.24

Bitswap: 105017 0

Total time = 1 days 23 hours 44 min 33 sec
SF = 69011225
CRC = 7668
LOS = 0
LOF = 0
ES = 3605
Latest 1 day time = 23 hours 44 min 33 sec
SF = 5254745
CRC = 722
LOS = 0
LOF = 0
ES = 369
Latest 15 minutes time = 14 min 33 sec
SF = 53708
CRC = 1
LOS = 0
LOF = 0
ES = 1
Previous 15 minutes time = 15 min 0 sec
SF = 55310
CRC = 0
LOS = 0
LOF = 0
ES = 0
Previous 1 day time = 24 hours 0 sec
SF = 5311658
CRC = 874
LOS = 0
LOF = 0
ES = 422
15 minutes interval [-30 min to -15 min] time = 15 min 0 sec
SF = 55249
CRC = 2
LOS = 0
LOF = 0
ES = 1
15 minutes interval [-45 min to -30 min] time = 15 min 0 sec
SF = 55372
CRC = 9
LOS = 0
LOF = 0
ES = 5
15 minutes interval [-60 min to -45 min] time = 15 min 0 sec
SF = 55311
CRC = 0
LOS = 0
LOF = 0
ES = 0
Showtime Drop Reason: 0
Last Retrain Reason: 0

[attachment deleted by admin]

News:

Author Topic: ADSL observations and night time margin drop (Read 8543 times)

Nick_R

ADSL observations and night time margin drop

Nick_R

Re: ADSL observations and night time margin drop

orainsear

Re: ADSL observations and night time margin drop

kitz

Re: ADSL observations and night time margin drop

Nick_R

Re: ADSL observations and night time margin drop

sevenlayermuddle

Re: ADSL observations and night time margin drop

Nick_R

Re: ADSL observations and night time margin drop

Nick_R

Re: ADSL observations and night time margin drop

kitz

Re: ADSL observations and night time margin drop

orainsear

Re: ADSL observations and night time margin drop

Nick_R

Re: ADSL observations and night time margin drop

Nick_R

Re: ADSL observations and night time margin drop

Nick_R

Re: ADSL observations and night time margin drop