Kitz Forum
Broadband Related => ADSL Issues => Topic started by: konrado5 on April 17, 2014, 06:57:37 PM
-
http://www.ist-muse.org/Deliverables/WPC2/MUSE_DC2.1p_V05.pdf
-
Thanks for pointing it out. It is an interesting paper revealing some examples of the variability in lines and discussing some of the issues and possible central diagnostics. I have saved it for future study.
-
It probably explains also why attenuation measurement is slightly depended on power output etc. despire real attenuation is static.
-
Thank you for the link. I, too, have downloaded a copy of the paper . . . :)
-
burakkucat: could you reply?
http://forum.kitz.co.uk/index.php?topic=13847.0
http://forum.kitz.co.uk/index.php?topic=13838.0
-
burakkucat: could you reply?
http://forum.kitz.co.uk/index.php?topic=13847.0
http://forum.kitz.co.uk/index.php?topic=13838.0
Now you are hijacking your own thread! :wall:
I have previously explained to you, in a PM, that I read all the posts that are made to all threads in this forum. I have also explained that I will post to any thread . . . when there is something that I feel I can add to the topic being discussed.
To answer your question "burakkucat: could you reply?", there is absolutely nothing that I can add to either of your two threads that has not already been said elsewhere.
Quite honestly, Eric (roseway) has summed it up precisely in his reply (http://forum.kitz.co.uk/index.php?topic=13847.msg260595#msg260595):
You aren't going to get an answer to that question here. The only answer anyone here can give is "We don't know". The 410-477 gap isn't something we see in the UK, so it may be something specific in Poland, or it may be something which is specific to your ISP. However many times you ask this question, the answer will remain the same: we don't know.
-
burakkucat: could you reply?
http://forum.kitz.co.uk/index.php?topic=13847.0
http://forum.kitz.co.uk/index.php?topic=13838.0
Now you are hijacking your own thread! :wall:
I have previously explained to you, in a PM, that I read all the posts that are made to all threads in this forum. I have also explained that I will post to any thread . . . when there is something that I feel I can add to the topic being discussed.
To answer your question "burakkucat: could you reply?", there is absolutely nothing that I can add to either of your two threads that has not already been said elsewhere.
Quite honestly, Eric (roseway) has summed it up precisely in his reply (http://forum.kitz.co.uk/index.php?topic=13847.msg260595#msg260595):
You aren't going to get an answer to that question here. The only answer anyone here can give is "We don't know". The 410-477 gap isn't something we see in the UK, so it may be something specific in Poland, or it may be something which is specific to your ISP. However many times you ask this question, the answer will remain the same: we don't know.
100% this.
konrado5, accept that there's nothing you can do and move on, please.
-
burakkucat:
The 410-477 gap isn't something we see in the UK, so it may be something specific in Poland, or it may be something which is specific to your ISP.
I know also it is not something specific in Poland. Polish engineer confirmed it is not normal. I'd like to explain why it is not something specific for ADSL in genera. If you don't see it in UK, you know it is not specific for ADSL2+. I'd like to explaining why. lI'd like to know how to explain it is not normal. I'd like to persuade my ISP to search for cause of this. Perhaps my ISP mounted 160m band notch-filter and I want my ISP removed this filter or explained me reason for mounting.
-
We know your reasons.
We don't know the answer.
-
Is it following reply a good reply? It is normal: higher synchronization=slighty higher attenuation=slighty higher interference capacity. On my SNR graph 410-477 tones are significantly more interferenced and tones 478-511 are much small interferenced. It is not normal
-
No, it's not normal. We don't know why it happens.
-
No, it's not normal. We don't know why it happens.
It implicates I formulated my question wrongly. I question about why it is not normal (not why it happens). My english is not so good.
-
I didn't want to respond to this thread because I have nothing new to add that I haven't already covered.
konrado I think perhaps you are getting confused as to what is normal.
From memory your graph displayed some dips that across the adsl spectrum would not be considered normal ie there were gaps.
It doesnt display the tail off you would normally expect to see for long lines - thats because your line isnt particularly long. To me it displays signs of some sort of interference. Now I refuse to discuss that again, because over the past few months, several people have advised you what those possible causes could be. You have dismissed every possible cause - and there have been numerous suggestions made. You dismissed local radio hams, but how do you know there isnt one for sure.. Ive seen a case where a taxi CB radio cause similar symptoms to someones line. My own line on adsl2+ had a notch cut out that was supposedly down to maritime frequencies.
What I think you are missing the point on is that because of the way adsl works it is pretty normal for most lines to see some tones that are affected by something or other. In fact its far more common to see some tones missing or dipped, its rare that there is a straight line decline without some sort of interference or other. It only usually happens on the very short and good lines. Even just normal background noise mush causes peaks and troughs.
Unless you can find a specific cause then you are out of luck... I think even a qualified REIN engineer would tear his hair out trying to diagnose your line & symptoms.. there's too much background noise about irrelevant concerns. We have tried to explain to you many many times all we know and I refuse to keep going round in circles.
If your ISP has notched certain tones then there would be a good reason for it and you wont be able to get it removed. I had a notch at tones 476 - 499 were no bits were loaded. I had to live with it and just be thankful that I had a good line with good performance and a good internet connection.
B*cat has repeatedly shown you his graphs, and how they arent a normal straight line, but the behaviour was normal and it was as good as he is going to get.
We have repeatedly tried to assist you, but you are getting hung up on the wrong aspect of normal. This is why now we feel that we cant help you any further. We cant give you an answer to something when we dont know what exactly is causing it. :(
-
OK,
There will indeed be people who know the answers.
These people work for the DSLAM manufacturers and will be prohibited by their contracts of employment from making any comments about their work on any public forum, or indeed to any person outside the company.
You will not find these sort of answers on the web.
Indeed if you read my contract of employment literally when I worked full time it was doubtful whether I could even tell people who I worked for let alone what I did or what the firm made!
So all of this tech info is confidential propriety information secret to the suppliers.
In the UK the DSLAM manufacturers supply to BT the network operator and BT supply the resulting product to the ISP. I doubt even BT know how it works in great depth. They will supply parameters to the DSLAM manufacturers which they require to be met and the DSLAM makers/designers will meet that spec. How they do it they certainly are not going to share with anyone else.
As regards your ISP,
They are not tech experts - they will not know the answers. They buy in the product and sell it on to you and me.
In the world of bespoke low volume high cost industrial sales yes, the sales people know about the product and are expected to be able to talk technical details to their customers. Invariably both sides will be educated to university degree plus level.
But in the field of mass market consumer sales like this where your customer base may number in the millions of subscribers the world is different.
It is a case of here is the product, we are not going to change it just for you, the world does not revolve around what one subscriber wants, we have no one able to discuss it with you, you are just one of millions of subscribers, and we really do not care if you go and buy it from someone else.
That is how mass market consumer products work and that is why they are cheap.
Contrary to expectations on forums everywhere on all subjects we as consumers do not have some right to know how something works or why some setting is set as it is.
Digressing briefly on onto the subject of attenuation.
I have two different routers of slightly different make from the same manufacturer, The line Hlog curve reported by one is different from the Hlog reported by the other as is the overall quoted line attenuation figure by 2dB. The line does not change and I can swap the routers over back and fro and the reported attenuation changes with the routers.
This is not the line changing its characteristics, it the the calculation/measuring methods used by the different routers internal firmware that is resulting in different attenuation's being reported.
Finally, I do not frequent forums where my own sector of expertise is on rare occasions discussed. It is a total waste of my time. Trying to explain to someone how something is and why it is and the compromises necessary both theoretical and manufacturing to arrive at the best overall solution when that person does not already have a technical background in the subject is impossible. Much as say someone trying to explain to me about postgraduate level particle physics!
-
kitz: I thought burakkucat excluded external interferences becaufe of my QLN graph. Anomalous region is lowered on QLN graph (and not raised). Moreover, I've still didn't get reply if deliberated DSLAM setting excluded because of:
However I notice a slight dip in snr both before and after which could point to radio/EMI interference being the cause.
Moreover, SNR on unused tones is negative. Is it possible with DSLAM delibareted settings? Moreover, sometimes tones 410-415 are used, sometimes not, sometimes tone 478 is used, sometimes not.
Thank you very much for your time
konrado5
-
Moreover, sometimes tones 410-415 are used, sometimes not, sometimes tone 478 is used, sometimes not.
Thank you very much for your time
konrado5
But this is by design. What you describe is exactly how ADSL is supposed to work. If there is noise on the line, some tones (where the noise is located) aren't going to be used, or if they are, the bitloading will be vastly reduced. Whilst the 'notch' in your graph is obvious, nobody here can offer an explanation as to why it is there. We simply don't know.
In any case, even if you resolved that notch, I doubt that you would see much improvement in speed.
As I see it, you have 3 options from here. You can either :-
1) Accept your line for what it is, complete with 'notch', and agree your speed isn't that bad.
2) Assume your ISP has, for some unknown reason, fitted a filter, and change ISP.
3) Keep asking what/why/how/if questions that nobody here (and I know of no other forum with more highly qualified experts than this one) can answer.
I hesitate to say it, but you seem to have a 'bee in your bonnet' regarding the condition of your line. I can sympathize with you but as far as I can see, there is nothing you can do about it. You need to accept this, and move on.
With no ill will intended,
BE
-
I doubt that you would see much improvement in speed.
I'd have 3 mbps raised synchronization rate (I've had 3 mbps more before my ISP changed DSLAM). Changing ISP doesn't satisfy my curiosity regarding to cause of notch.
But this is by design. What you describe is exactly how ADSL is supposed to work.
I know it is the way how ADSL works. I only question is it possible it is delibarately setted on ADSL2+. I suspect only way to delibarately disabling particular frequiences on DSLAM is disabling particular tones eg. 415-477 and not sometimes 415-476, sometimes 410-477.
-
:wall:
-
Apologies - I cant recall that QLN, but having looked at it now, then I'd concur with b*cat in that it very likely excludes crosstalk/RFI/EMI.
That spike smack bang in the middle is peculiar and it almost looks like the surrounding tones could be blocked out to prevent some sort of overspill, but Im only guessing at that.
I do however recall suggesting to you many months ago (and several times) the possibility of your ISP deliberately excluding these tones, but your replies seem to suggest this wasnt the case and/or your ISP denied that it was using any form of masking on these tones.
TBH Ive seen so many graphs and had so many discussions with you that my head spins sometimes.
As others have pointed out, often the ISP themselves may not even have a clue that these masks are in place, its not an area that their front-line staff who we speak to even have any training or knowledge about, as its left to techies who never see the light of day, nor disclose what they have done in public. You probably know (or should know by now) more information about adsl technology than most ISP staff.
I keep mentioning to you how BE (a UK isp) did something similar with tones 476 - 499 which after much debate with them they said it was to protect against maritime transmissions(?) which were affecting some areas. It took years before a proper explanation came forward and was admitted why they were doing it, but we as consumers could do nothing about it - even if we werent in an affected area.
>>> I'd have 3 mbps raised synchronization rate
It would take all those tones to be fully or near fully bitloaded to gain that much, dont forget that over-time you will also be expected to loose a certain amount of speed from crosstalk. This is inevitable as more users join your dslam. Its highly unlikely that tones in the 400+ region would gain full bit loading.
Unfortunately konrado, when we try and help and discuss the gap, you bring up small irrelevancies such at the minute attenuation changes which are normal, then we get distracted and annoyed by you bringing up something which isnt relevant and is normal.
I appreciate that you have the special ability to focus at length on certain aspects & talent to concentrate on problem solving likely far exceeds most of us here put together - but please try to understand that it becomes confusing for us when you focus in the wrong area and it throws the rest of us off the scent, and thats why we get frustrated when you repeat questions about areas that are normal. People on here do try and help, but we do have a limited amount of time and knowledge that we can spend trying to explain things to you as most of us have occupations or real life responsibilities that must come first. There also comes a point when our knowledge is exhausted.
-
I found probable cause:
http://forum.kitz.co.uk/index.php?topic=13856.0
I keep mentioning to you how BE (a UK isp) did something similar with tones 476 - 499 which after much debate with them they said it was to protect against maritime transmissions(?) which were affecting some areas. It took years before a proper explanation came forward and was admitted why they were doing it, but we as consumers could do nothing about it - even if we werent in an affected area.
Was it SNR very low nearly 476-499 tone?
-
but your replies seem to suggest this wasnt the case and/or your ISP denied that it was using any form of masking on these tones.
Because you've written:
However I notice a slight dip in snr both before and after which could point to radio/EMI interference being the cause.
Moreover, there aren't always the same tones not used, sometimes 410-477, sometimes 411-476, sometimes 415-470.
-
Now Im really just theorising on this... so do not take it as gospel. Take the following as a random thought and not fact.
BUT after having seen that spike of yours smack bang in the middle, that indicates to me some sort of disturbance even if those tones are blocked out either side... now what IF there is some noise spread from that tone, that could explain why sometimes the tones at each edge of the block are in use and sometimes not, because they are sometimes affected sometimes not. It could also explain the U shape either side.
However I dont know Im afraid. If Im talking garbage, someone can correct that thought. This is an area where we may never know unless your ISP confirms it. They may be like BE and deny the reasons for a while.
Id also need to see bitloading & SNRm per tone & QLN & Hlog from the same sync to dismiss the theory - but I really dont know if I could bear to look again, because the boy who cried wolf springs to mind. We've previously been distracted too much by you focussing on minute changes in the overall attenuation and bombarding us with things that arent relevant and are normal, and I dont think it would get us any further anyhow.
-
I sincerely apologize for bombarding with minute changes.
I've found identical PSD (probably) graph here with U shape:
http://www.itu.int/rec/T-REC-G.992.5-200305-S/en
On page 26 of PDF. It seems it is PSD mask for 400-483 tones. DSLAM doesn't block tones but it significantly lower power ouput. It explains the U shape of bit-loading and SNR graph.
-
burakkucat: You rather excluded PSD masks. Do you have the same belief after looking at above?
-
I've found identical PSD (probably) graph here with U shape:
http://www.itu.int/rec/T-REC-G.992.5-200305-S/en
On page 26 of PDF. It seems it is PSD mask for 400-483 tones. DSLAM doesn't block tones but it significantly lower power ouput. It explains the U shape of bit-loading and SNR graph.
Taking the above link shows me a screen, as in the first image below. Selecting the PDF file, downloading it and looking at page 26 I do not see a PSD mask (the second image, below).
I really do not know what to think. To be honest, I would prefer to forget about the whole issue. I am so utterly confused. :'(
-
I thought about 26 page of PDF file (18 page of text). Sorry for misunderstading.
-
I attach screenshot.
-
Thank you for the clarification. Looking at page 18 of the PDF file, I now see the curve to which you refer.
Yes, it does seem to be a possible explanation of what you observe. It also ties in with the sudden fall to the noise floor, as seen in your QLN graph.
-
I'm happy that I resolved the crux. Unfortunately the ISP tech staff probably didn't heared about PSD masks. I don't know if they change it.
Thank you very much for your time.
konrado5
-
Ive just spent a bit of time reading and trying to understand what it is saying and yes I agree it does look exactly like what you are seeing.
That specific type of HAM masking isnt something that Im aware of that is in use by any of the UK ISPs (BE did a total block rather than the U-interface mentioned in that document).
Having read the document it would appear to be a specific filter for the HAM band and bearing in mind that one of the guys posted a document just last week showing that your gap was specific to the Amateur Radio (HAM) band for Poland, then I think its fair to assume that what you are seeing is a dslam mask against Amateur Radio broadcasts in Poland.
Well done for finding that document, because Im quite sure that it is an explanation for what you are seeing.
As regards for getting it lifted, Im afraid to be the bearer of bad news in that its highly unlikely it will be lifted. It will have been put in place by the dslam owners, and therefore unlikely that even the frontline staff of your ISP will know its there.
The only thing that is surprising is that none of their other customers have noticed. But then again not everyone looks at their stats in such detail, and from what you have told us, its unlikely many will have routers that have the ability to plot QLN graphs to be able to spot it.
---
>> Warning - while you were typing a new reply has been posted. You may wish to review your post.
B*cat and yourself have since replied since I started reading and typing this, but Im going to leave it as it was.
-
I'm happy that I resolved the crux. Unfortunately the ISP tech staff probably didn't heared about PSD masks. I don't know if they change it.
I dont think they will.. its certainly not the sort of thing thats put in place lightly, and its there for a reason. To have put the mask in place will be there to protect many connections from being knocked out completely.
Looking at your QLN with that spike right in the centre would indicate to me there is still an amount of RFI getting through, therefore that tone would be useless. How much it affects your neighbouring tones is anyones guess, as this would change from various localities.
Because its been set on the dslam, it will affect everyone connected to it in your town. So if they remove it, its possible that connections on say for example the other side of your town will start being troublesome. That filter is possibly keeping a lot of other connections more stable.
-
Looking at your QLN with that spike right in the centre would indicate to me there is still an amount of RFI getting through, therefore that tone would be useless. How much it affects your neighbouring tones is anyones guess, as this would change from various localities.
It's hardly ever seen at my QLN. I've attached other QLN.
To have put the mask in place will be there to protect many connections from being knocked out completely.
I'm almost sure it is mistake. I suspect in Poland there are DSLAMs selled with default HAM PSD mask. My synchronization was lowered about 3 mbps when my ISP changed DSLAM. Moreover, someone in Poland has the same PSD masks and he uses other ISP service and lives 40-50 KM away from me. He got the gap suddenly and also lost 3 mbps.
http://obrazki.elektroda.pl/1012425200_1396541143.jpg
http://forum.kitz.co.uk/index.php?topic=13838.0
His ISP persists: it is implicited by lowering line quality because of time ( I don't know how to translate it. I think about: line quality is worse if the more time elapsed).
-
:wall:
-
<<snip>Digressing briefly on onto the subject of attenuation.
I have two different routers of slightly different make from the same manufacturer, The line Hlog curve reported by one is different from the Hlog reported by the other as is the overall quoted line attenuation figure by 2dB. The line does not change and I can swap the routers over back and fro and the reported attenuation changes with the routers.
This is not the line changing its characteristics, it the the calculation/measuring methods used by the different routers internal firmware that is resulting in different attenuation's being reported.<snip>
A bit OT but so many people do not recognise how roundings and number of decimal places used in calculations can affect mathematical outcomes.
Yet in real life it makes not a jot of difference, so konrado, please get a real life. ::) ;D
and countess of Google "Więc proszę się prawdziwe życie" :D
-
kitz: what do you think about my above replies?
-
I dont think I have anything else to say other than whats already been said.
-----
>>> it is implicited by lowering line quality because of time ( I don't know how to translate it. I think about: line quality is worse if the more time elapsed).
This is usually the effect of what we call crosstalk.... ie the more people that join the cab, then speeds get lower. There isnt anything you can do about that.. Im sure Ive mentioned several times, that when I was 1st on a brand new shiny adsl2+ MSAN I could easily get 24/2.7 Mbps. A few years later it was something like 21Mbps.
If you read the VDSL sections you also see lots of us talking about it. Those of us first on new DSLAMs are finding several months later that we are losing a heck of a lot of speed. BE has lost circa 20Mb, I lost nearly 20Mb in about 9 months. BT is well aware of it, unfortunately the faster speeds mean we lose more :( Its why we are waiting for a new technology called vectoring to come in as this lessens the effects of crosstalk.
-
kitz: I've meant that someone (someone else who also has 410-477 gap) ISP says: 410-477 gap is caused by lowering line guality when time elapsed. I think it is not cause. I know there are crosstalks. I've noticed it when my ISP sometimes reboot DSLAM. Then I get about 1.5 Mbps higher synchronization rate.
-
>>> 410-477 gap is caused by lowering line guality when time elapsed.
I wouldnt think so... too many other causes could do it. Unless its x-talk or interference, then a line doesnt usually degrade in that way over time - its not something Ive seen before.
-
Thank you for reply. 410-477 gap is explained. What do you think about strange phenomenon: there is always immediately after synchronization establishment: SNR margin 5.9 dB, power output 18.8 dBm, however some seconds after: 6.1 dB, power output 18.6 dBm. First strange thing is: there are always the same values. Second strange thing is: there are higher SNR margin when the power output is lower. Third strange thing is: the values are always changed few seconds after synchronization establishment. I don't know if it is typical or rather strange phenomenon on my line. I don't know if somebody checked it. It happens even if bitswap is off.
Thank you very much for your help
konrado5
-
Hi Konrado5 sorry to but in you seem very knowleageable into the workings of Broadband, I have uploaded my tone Graph and you will see there are a number of tones missing would you be able to help me understand why they are missing.
Cheers NS
-
NewtronStar: attach SNR per tone, QLN and Hlog. It seems it is radio interference or bridge tap.
-
Neutron Star's missing tones seem to correspond (well, more or less ) the 40m Amateur and Broadcast bands. Coupled with Konrado's similar blanking of the 160m band (more or less) this supports the idea that these masks are to hide interference. There are better ways of doing this on an individual basis, but to the administrative mind this is unthinkable; IT MEANS ADMITTING THEIR SYSTEM ISN'T PREFECT ! ( Be grateful, in previous ages the suppression of dissidents could be quick and brutal. remember Galileo ! )
I think you are both stuck with it, in the interests of stability, (and think what sticks to your shoes in a stable !)
-
. Coupled with Konrado's similar blanking of the 160m band (more or less) this supports the idea that these masks are to hide interference
NewtronStar Hlog and QLN will determine it.
-
. Coupled with Konrado's similar blanking of the 160m band (more or less) this supports the idea that these masks are to hide interference
NewtronStar Hlog and QLN will determine it.
I have the QLN Graph but you must remember this is taken when the modem first sync's to the DSLAM and its not updated until I force re-sync.
Here is Hlog it looks ok to me, those 5 small blocks that peak down to 98dB are US & DS Band Partitions i think and I don't see those in the HG612_Modem_stats Hlog Graph.
-
It seems it is 160m band radio interference. Are those tones always missing? Perhaps expanding QLN and Hlog graphs will be helpful.
-
It seems it is 160m band radio interference. Are those tones always missing?
The tones are inplace during the afternoon but 1 hour before sunset they start to go missing one by one until a large chunk of tones has gone and they will start to appear very slowly after 01:00 and by 12 noon they are all back again.
-
Therefore, I'm almost sure someone broadcast radio in 160m band.
-
Therefore, I'm almost sure someone broadcast radio in 160m band.
Yes it seems to be ! but how or why are these RFI signals getting inside my modem and I can say 100% that there wasn't any RFI in June 2013 as I have graphs going all the way back to August 2012 (HG612 Modem Unlock day) I remember it well ;D
-
It is worth reading.
http://www.kitz.co.uk/adsl/rein.htm
-
^^ :lol:
-
It is worth reading.
http://www.kitz.co.uk/adsl/rein.htm
Yes it is as four to six months ago I had terrible FEC errors 100,000 + per min and I found two culprits when setting the Radio Scanner to 512Khz what I found was the TP-LINK power adapters was causing 70% of REIN and and old Sony FreeView BOX (Digital TV) was causing 28% of REIN.
after removing those devices my FEC's never go above 14000 per min on a bad day.
thats my REIN fixed and my RFI is a different story :(
-
NewtronStar: serach if someone broadcast 40m amateur radio.
-
I suggest you should consider the problems as Crosstalk, not (external) RFI. In particular, even-order intermodulation at the exchange, would produce tone-like spectral lines at higher, near harmonic, frequencies. This would explain a mask blocking off these frequencies.
Listening with a receiver would show the difference between RFI and Crosstalk, a system designed round xDSL won't; (if you can't find any spectral lines at all, then interference only exists in the xDSL system so must be crosstalk ! )
-
Wondering how crosstalk can be in sync with the Sun set times ? IE: sunset in December is at 15:58 here tones go missing and in April the tones go missing 4 hours later, sounds like radio propagation to me or street lighting.
-
Wondering how crostallk can be related to 1650-1750 tone.
-
kitz and burakkucat: I hope my reply #38 has not missed.
-
kitz and burakkucat: I hope my reply #38 has not missed.
We're back in the land of tiny measurements that make no overall difference.
The short answer... I dont know :(