Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[kitz]

>> Here is another graph from router stats. Any opinions on the condition of my line greatfully received.

At first glance looks like theres a wee bit of noise around tone 60 (radio transmitter?/ is internal wiring ok?).. tone 64 is the pilot tone so thats not in use any how. SNR is okish in comparison, so whatever caused the blip, line has recovered. Later tones show some bins with few bits, rather than smooth drop off, but thats not too unusual on longer lines.

The bit loading is a bit juttery from that graph, but if its been taken after the router has been up a while, then bitswapping can cause that effect.
Bit loading in the earlier tones more ot less follows the (real) SNR but again showing signs of bitswapping probably having occured.

The real SNR is a bit bumpy but nothing of too major concern (check internal wiring)- again wondering if that line has been up for a while and bitswapping has caused increased power output.

I dont know what the actual line stats are from that graph, but I cant see anything major - comparing it with the SNRM fluctuations in the first graph.. it doesnt look too bad overall.   Line definitely seems in much better condition when using the DG834gv4.

When looking at bitloading graphs its always nice to have a graph after first boot up, then compare with one a day later, as its hard to say for sure with just the one as bit swapping can distort things a bit.

[Bigmac77]

Thanks Kitz, the bits & attenuation plot is the first 24 hours since the router was plugged in.
Bear with me on this, you refer to internal wiring in your reply, when the last BT engineer was here after he had finished he plugged in a black box which which carried out a number of test on my line and ADSL connection I watched as the device declared that the line had passed all the tests. The BT line comes into my utility room extension loft space and connects to a length of phone cable the electrician fitted (house was renovated including a rewire 2 years ago) the phone cable then runs through a duct into the main part of the house to the NTE5 BT socket. The duct carries all the twin and earth feeds from my consumer unit plus the central heating control wires. Could this be the cause of my problems?

[kitz]

The internal wiring was just a standard suggestion to see if any improvements could be made.. It wasn't a definite diagnosis as i couldn't see anything glaringly wrong from the graphs provided.

Thanks for confirming that the bit loading graph was after 24 hrs.  After comparing with a  graph showing a shift of snrm then it is likey/possible that the spikes are due to bitswapping that could/would have occurred over the past day.... And you have to look at a larger picture.

For example if u just look at the tones in region 60 in the graph you could say  something wrong there...  But then look at the real snr which is fine.  So it probably means at some point there was a burst type noise at those tones and bit swapping has done its job to keep the line stable.  The noise burst could have been for a fraction of a second, or could have been longer.. its impossible to say from just the one graph.

I'm accessing via mobile device so its harder to put inline links, but theres a section on the main site which explains bit loading and bit swapping if u want to know more



----
Edited to add link

http://www.kitz.co.uk/adsl/adsl_technology.htm#bit_loading

[Bigmac77]

My main concern is do I have a fault on my line? Yes I have found a router that works, will hold sync and produce a decent SNR graph but it's a router no longer available if something happens to this one unless I stock up on v4 DG834g's from ebay I'll be back to square one.

With regard to my previous post about the proximity of my house wiring to the phone line the only way to tell will probably be to move the NTE5 faceplate to the end of BT's incoming line and see if it makes a difference. I hate loft insulation.

[roseway]

The DG834Gv4 uses a Broadcom BCM6348 chipset, and it's most probable that other routers with the same chipset will have similar ADSL performance, so I don't think you need worry about stocking up with routers. There's a list of several routers with Broadcom chipsets here.

The performance you're getting with the DG834Gv4 is quite reasonable for your rather long line, so there's no indication of any line fault (unless it's an intermittent one). I wouldn't change anything more now, but continue to monitor the noise margin, etc. as you are doing.

[burakkucat]

With regard to my previous post about the proximity of my house wiring to the phone line the only way to tell will probably be to move the NTE5 faceplate to the end of BT's incoming line and see if it makes a difference.

Ideally, I would prefer that the xDSL line runs via a separate route and has a significant spacing from any parallel mains wiring. If you had a suitable length of 2-, 4- or 6-core CW1308 specification cable, you could perhaps leave the NTE5/A in situ and just run it from a temporary feed, bypassing the current cable that is in close proximity to the mains wiring?

[Bigmac77]

I can't run a cable via another route because of the layout of the house but I can move the NTE5 box. Where can I buy cw1308 cable from?

[burakkucat]

. . . Where can I buy cw1308 cable from?

Any of the usual suspects! 

Amazon market place, eBay shops, ADSL Nation, Clarity, run-IT-direct and Solwise, to name a few.

I have a reel of three-pair CW13008 specification cable (green, blue and orange pairs), 150 m length, that I recall buying from a small supplier in the Amazon Market Place.

[kitz]

 umm  this has to be said.. and to make sure Im understanding properly...   the nte5 is BT's demarcation point and therefore any wiring/cabling before that point is bts responsibility.   

When I said internal wiring I meant the usual stuff such as plugging the router into the master socket, checking extension sockets, filters and removal of the bell wire.  These simple things can ofter improve line stability.

It looked like you had a problem:- your earlier routerstats graph certainly showed something weird was happening later on in the evening.  But since using the netgear, the graphs that have come back so far show that although there is some slight SNRM fluctuations and bitswapping, these would appear to be within normal parameters for a line of your length.

Unless Ive missed it, we havent seen any actual line stats since the netgear has been on the line..  Is the line syncing at a better rate?  Have their been any other changes in the line stats and it would also be helpful if we could see the error counts.  (eg RS/CRC and errored seconds)

Since using the netgear have the evening dropouts stopped?

[Bigmac77]

I'm 100% sure there is no problem with my wiring, there is no bell wire or extension sockets, I've tried 5 different filters all plugged into the test socket on the NTE5, I have changed the RJ11 - RJ11 cable from the filter to the router, I've tried it with and without my Panasonic DECT phone all made no difference to the previous four routers. The DG834g v4 has performed better but last night it did what all the others had done and lowered the sync speed at around 10.30. I'll post up some error counts later.

[Bigmac77]

Ok I have some stats and graphs.

Downstream

Noise Margin:     7.5   dB
Connection Rate:  3776  Kbps
Line Attenuation: 54.0  dB
Power:            18.7  dBm
Max Rate:         4064  Kbps
 
SuperFrames:      4490961
SF (CRC) Errors:  359
Reed Solomon:     152692686
RS Corrected:     128479
RS Un-Corrected:  7580
HEC:              216
Errored Seconds:  656
Severe ES:        108

Interleave Depth: 16
Bitswaps:         0

Upstream

Noise Margin:     20.0  dB
Connection Rate:  448   Kbps
Line Attenuation: 31.5  dB
Power:            11.8  dBm
Max Rate:         1084  Kbps
 
SuperFrames:      4491025
SF (CRC) Errors:  0
Reed Solomon:     38173712
RS Corrected:     0
RS Un-Corrected:  0
HEC:              0
Errored Seconds:  0
Severe ES:        0

Interleave Depth: 4
Bitswaps:         0

This represents 21 hours of data since the netgear did a resync last night.

The resync can be clearly seen on the graphs. Please can you help me interperate these graphs.

[Bigmac77]

Immediately after I reset the router yesterday the noise margin was 6dB. The sync was 4138 Kbps and last night after it lost connection it re-synced at 3776Kbps at an SNR of 6dB again.

Are there any other graphs which would be of benefit to post up?

[kitz]

>>> Please can you help me interperate these graphs.

Sometime at around 13.30 you line picked up an additional bit of noise but only about 0.5dB and this can easily be tollerated.

At about 9.30pm the line started to became increasingly noisy until the SRNM got so low, and the build up of errors so high, that the line eventually dropped out.

Unfortunately despite having a respite for a few days, those graphs are back to indicating some sort of EMI.

-----


RS Corrected (FEC type errors) are when theres some noise on the line, but the its coping fine due to interleaving and error correction working as it should.

RS Uncorrected (CRC type errors) is when data hasnt been recovered from interleaving and data has had to be re-transmitted.

Too many CRCs in a short time span can trigger more serious errors such as Errored Seconds and SES..  and too many of those cause the line to drop.

On graph 2 you can see a build up of FECs till about 3pm when all goes quiet.  But then just after 10pm it starts again.  The steep drop is because the line has lost sync and obviously the counters have reset.


On graph 3 you can see that things were fine up until about 10pm, when the line is producing lots of CRCs - its when the line gets too noisy and equivalent of the router saying "sorry didnt hear that- can you say it again" due to lots of background noise.


Looking at the overall picture something happens at about 9.30ish and the line starts to get increasingly noisy, at first the line copes due to FEC, then CRCs start happening, but then it gets to the point where it looks like there a sharp noise burst and it cant cope anymore and drops out.


After resync theres still some background noise but it no longer increases and has reached a plateau.

At about 8am the next morning we see a sharp increase in CRCs indicating something similar with noise but not quite as bad or for as long.



Im trying to thing of something that would trgger that type of EMI - increasing buzz over 1/2 hr then a quick burst and back to a background but more stable buzz.  Perhaps the other guys can suggest something, but nothing immediately comes to mind for me what this could be.

Is there anything electrical in your house which follows this type of daily pattern..  um  water heaters, something that uses overnight power-saving etc...  I think you said you have no street lights nearby which would normally be my first guess, but what about external/motion sensor lighting.

[burakkucat]

Having tried my best, I cannot come up with anything to account for that RFI. (I assume that the antique milking machine in Farmer Giles' cow-shed is not used at those times?)

I just cannot feel "happy" knowing that the internal feed to the NTE5/A travels through ducting with cables carrying relatively high current mains supply . . . I would probably make a temporary disconnection at the point where the CW1308 cable (which disappears into the ducting) is joined to the external feed cable, attach a modem/router (powered by a battery pack -- as sheddyian's experiment) and connect it to my laptop computer, running on its battery. Switch off the electricity at the main switch and monitor the line during the crucial time period(s). That should provide some crucial information.

[kitz]

>>> I just cannot feel "happy" knowing that the internal feed to the NTE5/A travels through ducting with cables carrying relatively high current mains supply

Oh I fully understand where you are coming from B*cat and if the EU does some tinkering then its on their own heads..  as long as we have warned about the demarcation point and how anything before that point belongs to BT and shouldnt be touched, then we have done our job too