Charged $220 incorrect call out fee.

Connection issues, drop outs or speed related faults for ADSL and ADSL2+ services
seaeagle
Posts: 112
Joined: Mon Nov 05, 2012 3:30 pm
Location: Burnie

Re: Charged $220 incorrect call out fee.

Post by seaeagle » Sun Feb 14, 2016 5:06 pm

I have tested the use of of the printf statement crumbs it does not like CRLF as end of line. I really struggled and still do to understand why I was unsuccessful converting to linux end of line lf. Initially it was because I was a bit hazy about what was going on where but later I was aware of this CRLF being a problem. To me it seems that extracting from the modem the data I can not avoid geting CRLF in the BPT matrix. Linux inside the modem, linux outside and yet the file I save (also specified to have lf only end of line, still result is CRLF). In despearation I have aone liner using dos2unix and finally got the proper eol.

I have been wasting time I think today moving the G604T modem to the first entry point, but still poor speed, basically just dial up. Then I discovered that I have some data files with individual figures so I gathered up a small set of QLN, for some I had this feeling that I should look harder at the QLN data. I have put it in spreadsheet and will attach.

What is hows is remarkable, the QLN floor has been lifted, I assume this why the TP Link W8960N for some reason seems to let me down badly, yet the G604T from the BPT, the adsl only data I can get the other files seems empty. As the D-Link G604T happily runs with much better BPT values I am not too bad off though my upload speed of 35 kB/is pretty poor against my 256 kB/s expected.

It is a bit hard to see but apparently as time goes the QNL has increased until of late it really hit. Interesting thing is we did have a might thunder and lightning going past here in our little valley. The lightning struck the pole transformer on the other side of the road, blew the HT fuse inks of course. Though to the credit of the installation nothing failed here only 20m away. We were without power for close to 24 hours and I fully expected both our desktops to have blown, freezer/fridges, microwave own, door bell unit. Everything just started up again when power came on. Phone landline working. Just coincidence that the QLN seems to have hit it highest point!!

It is a bit crude but I have added the dates, international YYMMDD notation in the second picture, too hard to work out how to modify the spreadsheet.

I am no longer so sure that the modem is at fault, I am notw planning to take it over to my local pc wizard who has a much closer location to the exchange I believe and give it a check out.
Attachments
160214-38c-adsl-line-test-exetel-speed-test-QLN-data-2013-2016-W8960N-D-Link.jpg
legend for the graphs YYMMDD cells
160214-38c-adsl-line-test-exetel-speed-test-QLN-data-2013-2016-W8960N-D-Link.jpg (13.31 KiB) Viewed 3930 times
160214-38b-adsl-line-test-exetel-speed-test-QLN-data-2013-2016-W8960N-D-Link.jpg
QLN samples dec 2013 to feb 2016
160214-38b-adsl-line-test-exetel-speed-test-QLN-data-2013-2016-W8960N-D-Link.jpg (93.56 KiB) Viewed 3930 times

Dazzled
Volunteer Site Admin
Posts: 6007
Joined: Mon Nov 13, 2006 1:16 pm
Location: Sydney

Re: Charged $220 incorrect call out fee.

Post by Dazzled » Sun Feb 14, 2016 9:02 pm

The shell has an easy way to deal with substrings. Here it's an unwanted \r found in the input to a printf. D-Link must have thought they were helping Windows telnet users. Have a look at http://www.linuxtopia.org/online_books/ ... ation.html near the bottom under "Substring Replacement". That site isn't a bad way to bone up on the power that's available.

If the variable is followed by a slash (/), the first occurrence of the pattern following the slash is replaced. Following the pattern, there is a second slash and the replacement string. If the variable is followed by two slashes, all occurrences of the pattern are replaced by the replacement string.

The expression being printf-ed can therefore be substituted on the fly. After the output format, put the input string as a substitution - ${STRING//\r/" "} and every CR is instantly a harmless space!

If we go back to the beginning of this largest thread ever we talked about how the modem could detect radio interference at your location. From an instrument maker's manual -
HLog data can indicate physical copper loop conditions, such as the presence of taps or a capacitive problem. QLN data can show noise events, including crosstalk and radio frequency interference. SNR data can show time-dependant changes in crosstalk levels and loop attenuation due to temperature changes or moisture issues.

Correlating the BPT data with QLN will easily show how noise spikes impact bit loading. Correlating BPT with HLog data will also show how impairments, such as bridged taps and poor contacts, in a premises impact bit loading. Analyzing HLog, QLN, and SNR helps users to determine why data rates fail to reach the maximum possible rates.

seaeagle
Posts: 112
Joined: Mon Nov 05, 2012 3:30 pm
Location: Burnie

Re: Charged $220 incorrect call out fee.

Post by seaeagle » Fri Feb 19, 2016 5:12 pm

Dazzled
First I had some good fun with your bits of information but it does stretch me to get my head around the statements. I also went back and tried a bit of modifications to your use of 'expect' and really got myself into trouble making assumptions. I will set these exercises aside for a bit because I have now emailed support with some information about my connection. On a daily basis it seems I do not know where it will be from failing to make connection to run at a reasonable level. I have a feeling that something/somebody is playing around with the lines in our street.

Support
Now summarising both for myself and for 'support' which I have pointed to this link in my email to save doubling up as well as this topic is my main records of problems (sorry to the original poster, I really hijacked this well and truly)

The TP-Link W8960N is the modem that allows me to do the line analysis of SNR, QLN, HLog and BPT.

1.. Some month ago I had a really good connection, a typical analysis of the line would be as the picture 150614- shows. Then a few times for the rest of the 2015 year clearly something was messing with the lines up our street, for a day or two.

2.. January 2016 was a bit up and down and in the beginning I did not have time to check because of our daughter and family visited but at the end of the month I started really taking an interest. The 160230- picture shows how TP-Link W8960N suddenly struggles.

3.. Today the 19th of Feb 2016 the modem is just about giving up. Picture 160219- Notice my SNR almost nonexistent relative to my 'good days'
Attachments
150614-spectrumplot-isp-connection-2-recheck.jpg
150614-spectrumplot-isp-connection-2-recheck.jpg (61.23 KiB) Viewed 3885 times
160130-04-adsl-line-test-no-phones-combine-version-2.jpg
Bad day on 30th Jan
160130-04-adsl-line-test-no-phones-combine-version-2.jpg (62.86 KiB) Viewed 3885 times
160219-m5-adsl-exetel-speedtest-application-ext-phones-12.png
Bad run today can only use modem to analyse will not connect
160219-m5-adsl-exetel-speedtest-application-ext-phones-12.png (71.58 KiB) Viewed 3885 times

seaeagle
Posts: 112
Joined: Mon Nov 05, 2012 3:30 pm
Location: Burnie

Re: Charged $220 incorrect call out fee.

Post by seaeagle » Fri Feb 19, 2016 5:15 pm

(continued)
As I can only get BPT values out of our spare modem, D-Link DLS G604T , the BPT from today from the spare modem. Not good but at least a slow connection. I will show what we had from D-Link DLS G604T yesterday as well.

I struggle to understand why the D-Link DLS G604T despite the line conditions seems to do so much better than the TP-Link W8960N, did lead me to think that some damaged to the modem may have occurred during a severe thunderstorm that went past about 2-3 weeks ago. In an effort to check the modem I have taken it to another part of town and tested operation and line conditions. The result was not as clear cut as I was hoping, it only confirmed that the ADSL connection worked and provided the expected speed according to the owner who regularly checks with Speedtest.net.

I have done quite a number of speedtests using the application under My Exetel section, hope this is recorded and can be accessed by exetel support. I do have copies of a number of them but certainly not all. :D
Attachments
160218-m5-adsl-exetel-speedtest-application-jrf-no-phones-05.png
TP-Link W8960N test on another location in town
160218-m5-adsl-exetel-speedtest-application-jrf-no-phones-05.png (70.8 KiB) Viewed 3885 times
160218-41b-adsl-line-test-exetel-speed-test-G604T-modem-ext-phones-BPT-graph.jpg
BPT from D-Link DLS G604T
160218-41b-adsl-line-test-exetel-speed-test-G604T-modem-ext-phones-BPT-graph.jpg (69.66 KiB) Viewed 3885 times
160219-42-adsl-line-test-exetel-speed-test-G604T-modem-ext-phones-BPT-graph.jpg
BPT from D-Link DLS G604T
160219-42-adsl-line-test-exetel-speed-test-G604T-modem-ext-phones-BPT-graph.jpg (53.4 KiB) Viewed 3885 times

Dazzled
Volunteer Site Admin
Posts: 6007
Joined: Mon Nov 13, 2006 1:16 pm
Location: Sydney

Re: Charged $220 incorrect call out fee.

Post by Dazzled » Fri Feb 19, 2016 7:39 pm

Expect scripts are written in Tickle (Tcl) not Bash! The differences aren't great, see https://www.tcl.tk/man/tcl8.5/tutorial/tcltutorial.html.

I haven't looked hard at your charts yet, but there is a pronounced low Hlog roll-off - increased attenuation at lower frequencies, possibly capacitance or corrosion. QLN shows noise spikes which may be the radio transmitters we spoke about, and might be cross talk. Look at the effect on SNR and BPT. SNR which is a margin, is lousy to the point of non-existent at higher frequencies, and could be a line in need of maintenance. Has it been wet?

Your friend's result is much better, but there is also a roll-off, and note the effect of the dip at 90-100. I'd check your script code for the 16 bits result in that chart. This one is getting a high frequency signal, and it makes a difference to the total bits available for the analogue to digital conversion.

Has support done a line test for you?

seaeagle
Posts: 112
Joined: Mon Nov 05, 2012 3:30 pm
Location: Burnie

Re: Charged $220 incorrect call out fee.

Post by seaeagle » Fri Feb 19, 2016 11:45 pm

Ouch, forgot about 'expect' not being in bash script. Will have a look at that later.
QLN noise spikes, yes not surprising, accross the valley direct line of site we have a significant tower system so I assume radio stations, yes I think our local AM station in the broadcast band figured nicely

Hlog high frequency roll off, well I thought that fitted nicely with my long distance from the exchange. Similar my friends place though I thought his distance must be much shorter. The low frequency part I had not really thought about, it looked roughly like I had enough bit loading to give me the 256 kb/s upload I am paying for. Of late days I was starting to wonder, especially when I noticed that I am loosing quite a few tones completely - I actually thought that was deliberately done by the ISP.

My friend and I discussed the SNR, which I regarded as real SNR and he thought perhaps 'margin' in the end we decided not to worry about it too much. I now realised I made a mistake taking the margin to be the difference between QLN and SNR using the SNR scale :oops: :oops: I have top look again at the UK link you gave a bit earlier.

The BPT I really have no idea why I see yo-yo like behaviour. It bothers me and perhaps still could be the TP-Link W8960N although normally the spare modem confirms the connection speed yo-yo.

Support has given a ticket number so far. Ah, there is now an email from support on the other computer, better study that.

seaeagle
Posts: 112
Joined: Mon Nov 05, 2012 3:30 pm
Location: Burnie

Re: Charged $220 incorrect call out fee.

Post by seaeagle » Sat Feb 20, 2016 7:32 pm

Last night I a small amount of testing after support advised that they had rebuilt some profile in the exchange which indicated to me that something was showing up allright. Doing a powercycle and a quick few speedtests was disappointing in that my linespeed was even lower.This morning however it was quite another story, my download speed registered over 3000 kb/s and rather solid. SNR recovered and back to what I would like to see. Followed it during the day and it looke good on the spare, D-Link DLS G604T, modem.

The TP-Link W8960N was a different story, via this I can see the SNR is back where I like to see it but the BPT does no longer match my expectations. I think there has to be something wrong with it. I reloaded the firmware, reset to default factory and set up the details from previous runs again.

No change, then bad luck hit, I think I accidentally bricked the the D-Link, meaning I no have no wireless modem and will be falling back on my older NetComm NB5Plus4 while I lick my wounds. I was doing a factory reset and held the reset button in for too long I believe. I can now no longer make contact with the modem, the lights seem to be doing the correct thing but I can not log in and it does not show up on the LAN when running 'nmap'. Neither can I use 'telnet' not looking good.

Spent a few hours trying once more to understand why the Billion 7404VGPX I borrowed from my local pc wizzard would not connect to the internet. I can get everything to work except the making a proper Internet connection. I managed to see a couple of messages, one was "Incorrect IPCP configuration" a Google landed me in Whirlpool topic showing that probably that modem was a causing some frustration in many cases. :evil:

The BPT is the last picture from the spare modem, D-Link DLS G604T, which now unfortunately looks bricked. The speedtest about the same time.

Today my NetComm V85 telephone died I am starting to wonder if there is something of a delayed breakdown of equipment after the thunder and lightning went past a few weeks ago.

Added later
I have now installed the older NetComm NB5Plus4 and run an ehternet cable to my wife's desktop this should give me some time to get a new wireless router sorted out.
Attachments
160220-adsl-line-test-exetel-speed-test-G604T-modem-after-support-reset-09.jpg
160220-adsl-line-test-exetel-speed-test-G604T-modem-after-support-reset-09.jpg (77.46 KiB) Viewed 3848 times
160220-adsl-line-test-exetel-speed-test-G605T-modem-after-support-reset-06.jpg
160220-adsl-line-test-exetel-speed-test-G605T-modem-after-support-reset-06.jpg (57.55 KiB) Viewed 3848 times

seaeagle
Posts: 112
Joined: Mon Nov 05, 2012 3:30 pm
Location: Burnie

Re: Charged $220 incorrect call out fee.

Post by seaeagle » Mon Feb 22, 2016 12:21 am

Well linespeed dropped to half late Sunday evening down 1600 kb/s I am now using the Netcomm NB5+4 model and had a peak in the program, bit of a surprise I have found the BPT table but not any others so far. Happened to be /proc/avalanche/ which caught my eye. Did a quick rewrite of the script and I think I have trimmed the correct lines out, seems the tables have oodles of room in this modem. It is late so I will just upload the graph.

What I noticed instantly was significantly more tones no loading in the UP direction, compared with the faulty (??) TP-Link W8960N. Though the missing tone loadings may be part of the supression of bitloading in the ADSL DOWN direction on this modem. Anyway I like the look of this graphs. Pity I can't get the QLN. I really miss my bricked D-Link, when looking at the BPT graphs above, did surprisingly well.
Attachments
160221-adsl-line-test-exetel-speed-test-NetComm-NB5+4-phones-ext-06.jpg
160221-adsl-line-test-exetel-speed-test-NetComm-NB5+4-phones-ext-06.jpg (61.13 KiB) Viewed 3838 times

seaeagle
Posts: 112
Joined: Mon Nov 05, 2012 3:30 pm
Location: Burnie

Re: Charged $220 incorrect call out fee.

Post by seaeagle » Tue Feb 23, 2016 12:40 am

Dazzled
Any suggestions or two for a new TP Link modem considering that about 12 months may see a fiber node about 50m from my locations, VDSL connection would suit me fine I think :D

Subjectively I think my DOWN linepseed wanders between 1600 and 3000 kb/s, for example here at midnight I just noted a low value and in the morning today high values. Had no time to do any scripting.

Dazzled
Volunteer Site Admin
Posts: 6007
Joined: Mon Nov 13, 2006 1:16 pm
Location: Sydney

Re: Charged $220 incorrect call out fee.

Post by Dazzled » Tue Feb 23, 2016 10:20 am

You should be able to pick up a TP-Link TD-W8960N locally for $40ish. The line low at night might be RF. A cron job to read the modem at intervals and list the result would nail down the times, but I would guess the source is out of your control. Gnome-schedule is the easy way to set this up. That is different from low data transmission speed which reflects traffic elsewhere and is common at peak times.

seaeagle
Posts: 112
Joined: Mon Nov 05, 2012 3:30 pm
Location: Burnie

Re: Charged $220 incorrect call out fee.

Post by seaeagle » Tue Mar 01, 2016 10:53 pm

New modem arrived yesterday great result. Looking good sofar, crossed fingers. Not had time even to check through the modem settings.

The QLN looks very choppy, I have really missed that in the last 10 days to show perhaps some explanation for the BPT varying a lot.
Attachments
160301-adsl-line-test-exetel-speed-test-TP-Link-TD-W8960N-phones-ext-05.jpg
160301-adsl-line-test-exetel-speed-test-TP-Link-TD-W8960N-phones-ext-05.jpg (46.84 KiB) Viewed 3777 times
160301-adsl-line-test-exetel-speed-test-TP-Link-TD-W8960N-phones-ext-06.jpg
160301-adsl-line-test-exetel-speed-test-TP-Link-TD-W8960N-phones-ext-06.jpg (51.36 KiB) Viewed 3777 times

Dazzled
Volunteer Site Admin
Posts: 6007
Joined: Mon Nov 13, 2006 1:16 pm
Location: Sydney

Re: Charged $220 incorrect call out fee.

Post by Dazzled » Wed Mar 02, 2016 1:21 pm

That's a nicer result than you were used to.

It would be worthwhile putting back the scale on the upper graph marked in kHz rather than tone numbers, to help locate the sources of the noise spikes, as these are costing you bits, and therefore download speed.
set x2range [:] (to cover the frequency of the highest tone)
set x2tics nomirror (list) (inserting the values of great interest. 7BU is at 558 kHz)
The values you put in the list will be ticked proportionately in the given range.

There isn't a lot you can do about the 7BU AM station, but some may be RF under your control, like power supplies. Wide spikes can be cross talk energy.

PS Chart titles?

seaeagle
Posts: 112
Joined: Mon Nov 05, 2012 3:30 pm
Location: Burnie

Re: Charged $220 incorrect call out fee.

Post by seaeagle » Wed Mar 02, 2016 9:33 pm

The extra performance seems to have come with a cost, I did not have time to mention this above. When I run my program I now will get a failure in the form of no graphs at worst and generally at least the BPT will survive, however tonight when I tried to change from tone number to frequency as suggested I got no graphs.

From what I have found it seems if I repeat the test too soon then I will get lost graphs, ie the data files for the graphs are empty and gnuplot recognises this and does not try to plot. :D

Dazzled
Volunteer Site Admin
Posts: 6007
Joined: Mon Nov 13, 2006 1:16 pm
Location: Sydney

Re: Charged $220 incorrect call out fee.

Post by Dazzled » Thu Mar 03, 2016 9:49 am

Are you using expect, which will wait for a reply? Tcl also has sleep and after commands which may help if needed before the expect request. Collect the data on the telnet terminal to get an idea of the speed of the router's response. These devices don't have very fast processors.

Expect is an ideal way to automate interactive tasks. It can talk to multiple programs at once and do things automatically that are difficult even with a command shell. It's now an old technology.

In the plot command file, you have 255 bins. (My code as published here was for a 512 bin setup). Bin 0 is not used. The centre frequency of bin N is (N x 4.3125) kHz, so you know the frequency range for your chart. Set x2range accordingly, and you can add as many x2tics as you wish.

Franpa
Posts: 450
Joined: Thu May 15, 2008 11:44 am
Location: Australia, QLD

Re: Charged $220 incorrect call out fee.

Post by Franpa » Sun Mar 06, 2016 2:58 am

seaeagle wrote:Last night I a small amount of testing after support advised that they had rebuilt some profile in the exchange which indicated to me that something was showing up allright. Doing a powercycle and a quick few speedtests was disappointing in that my linespeed was even lower.This morning however it was quite another story, my download speed registered over 3000 kb/s and rather solid. SNR recovered and back to what I would like to see. Followed it during the day and it looke good on the spare, D-Link DLS G604T, modem.

The TP-Link W8960N was a different story, via this I can see the SNR is back where I like to see it but the BPT does no longer match my expectations. I think there has to be something wrong with it. I reloaded the firmware, reset to default factory and set up the details from previous runs again.

No change, then bad luck hit, I think I accidentally bricked the the D-Link, meaning I no have no wireless modem and will be falling back on my older NetComm NB5Plus4 while I lick my wounds. I was doing a factory reset and held the reset button in for too long I believe. I can now no longer make contact with the modem, the lights seem to be doing the correct thing but I can not log in and it does not show up on the LAN when running 'nmap'. Neither can I use 'telnet' not looking good.
1) The D-Link may have been bricked via the firmware update it self, rather than through fault of your own. For example a firmware update to a Netcomm NB1300+4 Mode/Router would eventually brick the device because the log files from the older version were properly cleaned up causing the device to run of of storage capacity or some such, the company performed product recalls to address the issue and fixed ours for free (Maybe we paid postage) even though we stumbled upon the issue well after the recall period had ended. So back on topic, the D-Link firmware update may have a similar issue or it might not.

2) Make sure that your Gateway IP Address specified on your computers match what the Router is configured to, you may have changed this at one point and the factory reset would've reverted it back to the default value, if the PC's Gateway Address doesn't match what the Router is operating on than the computer won't connect to it. I stumbled in to this issue when messing with my Netcomm NB16WV-02 Modem/Router (Exetel staff pointed this out to me).
Windows 10 Pro x64 | Ryzen 3700X | ASUS Crosshair VIII Hero (WiFi) | 16GB 3600MHz RAM | MSI Gamer 1070Ti 8GB | Exetel ZTE H268 Modem

https://www.speedtest.net/result/8729259527 (during low neighbourhood congestion)

Post Reply