cancel
Showing results for 
Search instead for 
Did you mean: 

Downstream SNR varying wildly

summers
Aspiring Pro
Posts: 275
Thanks: 50
Fixes: 1
Registered: ‎01-06-2014

Downstream SNR varying wildly

Here is my Downstream Noise Margin for the last day and a bit, my line is currently going through training - but numbers are fairly typical. This this is just what happens on a line with some noise ....
dick:green Post split to own topic for clarity
144 REPLIES 144
Townman
Superuser
Superuser
Posts: 22,918
Thanks: 9,535
Fixes: 156
Registered: ‎22-08-2007

Re: Downstream SNR varying wildly

Hi Summers,
I asked the mods to split off your post because there is something far from optimal about that plot.  First the shape looks grim, but more importantly the 'baseline' is all wrong.  In ideal conditions, your SNRM should be running around 6dB - your plot is circa 14dB.
If you want help to improve this, can you please look at the speed issues thread (link below) and post here all of the information it requests?  The community members might then be able to assist you gain some improvement.
First off though, is there noise on your phone line?  Dial 17070 option 2.  If there is noise, then raise a noisy line fault with your phone provider.  Say nothing of broadband issues (unless the phone provider is PlusNet).

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

summers
Aspiring Pro
Posts: 275
Thanks: 50
Fixes: 1
Registered: ‎01-06-2014

Re: Downstream SNR varying wildly

Hi, yes may be due to line currently training.
Historically there has been noise on my line, and the defaul noise margin was set to 15dB. Openreach came round on 31/11/15, found the fault on the line, and fixed it. Since then I've not had noise on the phone line when I checked.
When fault was corrected, the noise margin was set back to 6dB, and line put into training. Over the last few days the line has been reconnected a few times with a larger noise margin.
Give me a few minutes, and I'll do graphs of the noise margin and maximal rate since the fault on the line was corrected.
mav:quote
Townman
Superuser
Superuser
Posts: 22,918
Thanks: 9,535
Fixes: 156
Registered: ‎22-08-2007

Re: Downstream SNR varying wildly

Based on what you've said, I suspect that there are more faults on your line than the one repaired - that is not uncommon.  Find something to point the finger at, fix it, walk away and look no further - especially if it was a BT subcontractor who is paid by the job no matter how long it takes (or does not take!).  Short sighted economy but who would expect different from BTOR?
The fact that the SNRM has risen again indicates that the DLM is trying to manage the line's stability.  It is a huge shift and the plotted variation looks grim.
Looking at the graph, it looks like you are doing your own monitoring using excel?  Have you looked at any of the standard tools (for example routerstats - see link below) which can plot more than just SMRM?

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

summers
Aspiring Pro
Posts: 275
Thanks: 50
Fixes: 1
Registered: ‎01-06-2014

Re: Downstream SNR varying wildly

And find attached the Down Stream Maximal Rate, and the Down Stream Noise Margin, since the OpenReach engineer came round. I've added current rate as well.
You can see training being switch on about 7/1/12, and noise margin set to 6db. Then over next few days the noise margin is increased, and the maximal rate drops. Then last night, some noise on the line, when alas I didn't check to see if noise on the phone line.
In the current rate, you can clearly see when the line is reset - this is done at the far end, and not in my router. You can see the Noise Margin changes happening.
This is though *far* better than before the engineer came round, then the line varied between 300kbps and down ...
summers
Aspiring Pro
Posts: 275
Thanks: 50
Fixes: 1
Registered: ‎01-06-2014

Re: Downstream SNR varying wildly

Quote from: Townman
Looking at the graph, it looks like you are doing your own monitoring using excel?  Have you looked at any of the standard tools (for example routerstats - see link below) which can plot more than just SMRM?

I've a script running on my NAS, that records the router parameters every 10 minutes. This was set up, when my line was dying last year. It saves everything as a text file, so to produce graphs need to use a tool. Hey in this case I went via libreoffice.
Anyway this is good enough for my needs. Also all computers at home run linux - so  limited what software I can use.
Townman
Superuser
Superuser
Posts: 22,918
Thanks: 9,535
Fixes: 156
Registered: ‎22-08-2007

Re: Downstream SNR varying wildly

That just isn't good - you have still got a line fault - I'd put money on it.
Shame that linx is so useless when it comes to running common industry standard toolsets...  Wink Roll_eyes ...routerstats would have been so much simplers and to do SNRM tracing properly (especially if searching for REIN / SHINE) you need a 10 second sample interval.  Ah well!
In the meantime, you could post the router stats please, check the phone line for noise on a regular interval, especially when the SNRM is shifting around which sadly I suspect that your monitoring method is not going to show that in real time.  And I'd recommend raising another on-line fault report.  Make sure that indicate that BTOR has been out to fix an issue, but that it would appear that there are others not addressed.

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

summers
Aspiring Pro
Posts: 275
Thanks: 50
Fixes: 1
Registered: ‎01-06-2014

Re: Downstream SNR varying wildly

Well at least in linux it was easy to write a script which probed the telnet interface of thr router - which is all that routerstats does. Yes I could do 10s interval, but the log file has been running for 4 months or so, and is already up to 1.5MB. At 10s I'd be up to 100MB.
Yes, I have my suspicions about the line, hence why I'm logging all the stats of the router. I'll update the script to do 10s probing - at that rate it is better somewhat rewritten (viz, I won't disconnect the telnet, but leave it connected - just doing the probe). Oh yes, current log has all the stats on the router - vis Upsteam and Downstream, Currer rate, maximal Rate, Noise Margin, Attenuation, and Power -  but graphs already posted contain the interesting bits.
Hassle with PN/BT is even before when line was as noisy as a parrot in a washing machine, both PN & BT insisted that they was no problem on their hardware. Alas its also hard for me to be home, in hours when an engineer comes round - you know the kind of problem, full time job etc ...
Townman
Superuser
Superuser
Posts: 22,918
Thanks: 9,535
Fixes: 156
Registered: ‎22-08-2007

Re: Downstream SNR varying wildly

Hi,
Some suggestions... Given that this is scripted, you could consider log switching at the end of the day / week / month to manage file size.  There is little point recording nay max line rate figures as they are (to all intents and purposes) meaningless.  They inform what might be available if all things were perfect and there was no need to reign-in the power levels.  If the router proffers error counters, its useful to harvest them as well.  Matching error counts to SNRM variance illuminates just how much real impact the line noise has on usability.

Superusers are not staff, but they do have a direct line of communication into the business in order to raise issues, concerns and feedback from the community.

summers
Aspiring Pro
Posts: 275
Thanks: 50
Fixes: 1
Registered: ‎01-06-2014

Re: Downstream SNR varying wildly

OK, "expect" script written.
Logging onto the router always takes a few seconds - looks like router is configured that way. But keeping the connection up, and can keep polling the info. Will probably do for an hour or two - not sure how well the expect script/telnet session will stay up when polled continuously.
Didn't think I'd seen any errors on the router - I'll dig through the various menus to see if errors are shown anywhere.
Should be able to get 1s resolution, with a gap of a few seconds when I log rotate. Can run faster than 1Hz - but don't see the need.
Don't yet have a method of getting a timestamp on the records - I'll work on that ...
summers
Aspiring Pro
Posts: 275
Thanks: 50
Fixes: 1
Registered: ‎01-06-2014

Re: Downstream SNR varying wildly

Oh yes - in case you are interested in the power of linux - here is the expect script and surround code that does 1s recording:
( expect <<_end
spawn telnet 192.168.2.1
expect "username:"
send "username\r"
expect "password:"
send "password\r"
expect "TP-LINK(conf)#"
while { 1 } {
send "adsl show info\r"
sleep 1
expect "cmd:SUCC"
}
_end
 ) | sed -n 's/upstream\([A-Z]\)[a-z]*\([A-Z]\?\)[a-z]*/US\1\2/p;
                   s/downstream\([A-Z]\)[a-z]*\([A-Z]\?\)[a-z]*/DS\1\2/p'
ejs
Aspiring Hero
Posts: 5,442
Thanks: 631
Fixes: 25
Registered: ‎10-06-2010

Re: Downstream SNR varying wildly

Polling stats at one second intervals is probably excessively short, and it might even start taking up too much of the router's CPU time.
summers
Aspiring Pro
Posts: 275
Thanks: 50
Fixes: 1
Registered: ‎01-06-2014

Re: Downstream SNR varying wildly

I've set it up for 1s output - and started it running. Note that the 1s delay is after I've got the output of "adsl show info", so that should give the router breathing space.
As the timestamp I'm recording has 1s fidelity (its a time since epoch), and can see if any seconds get skipped - and that will show how much time is being used.
Anyway TP Link routers are openwrt based, which means based on linux. In linux the routing operations are performed in kernel space, whereas the telnet cli will run in user space, and so lower priority. I would expect most routers to be similar, even if there are some out there not based on linux.
summers
Aspiring Pro
Posts: 275
Thanks: 50
Fixes: 1
Registered: ‎01-06-2014

Re: Downstream SNR varying wildly

1s recording seemed to work fine, the telnet cli stay up over night. In one night produced more data than had in 3 months previous. But now means I can give graphs with 1s fidelity. Most don't vary, ie US Stream parameters, powers, etc. So only intersing ones are Down Stream maximal rate, and attenuation. For simplicity of presentation here is the Down Stream Maximal Rate, with time ploted as hours since 00:00:00 on 12/1/16. So 22=10pm last night, 34 = 10am this morning.
You can see that the noise is intermittent. It was big last nigh from when I started recording, through to 11pm. In then started up in the morning at 3am, dropping off at 8am. I have nothing in the house with this timing! In the zoom in on the noise, you can see that when the noise comes, it typically has a duration of at least minutes, often longer.
So what would you conclude? Noise from outside the properly? Maybe weather related? Actually its the same pattern as the night before - so can't be weather. However the time period seems hard to interpret. Maybe that the phone line goes over an electricity sub station? Hmmm, maybe I'll take my SDR upto the electricity sub station and scan the airwaves ....
I'll leave the 1s recording of data going for a few days, to see if there is any pattern.
[Note added, actually the Down Stream Noise margin has the same paattern, but added anyway]
ejs
Aspiring Hero
Posts: 5,442
Thanks: 631
Fixes: 25
Registered: ‎10-06-2010

Re: Downstream SNR varying wildly

Is this the TP-Link TD-W8970 v1? If so, then the upstream figures will never change, the device just doesn't bother to ever update them after the ADSL connection is established. Without seeing all the stats (line attenuation, rate and snr margin), it's not really possible to say much about the line's performance.