Kitz Forum
Broadband Related => ADSL Issues => Topic started by: SlowConnector on May 21, 2012, 10:33:27 AM
-
Hi all,
Since February 2012, I have been experiencing a connection problem with by ADSL2+ broadband connection, whereby the connection drops 5-10 minutes after I power up my router, but I am having problems with diagnosing what the source of the problem might be. Essentially I need some guidance as to what tests might identify the problem and hopefully lead to a more reliable connection, which I suspect is slowing down and dropping at other times too.
For your information, I have been in conversation with online Talktalk support staff in this thread (http://www.talktalkmembers.com/forums/showthread.php?t=35324&page=3) on their Broadband Support forum, a result of which I have run various tests from the test socket with new routers, filters and computers at different times of the day and night, but the problem is the same. However, the Talktalk team are not really in a position to provide me with a solution until I can diagnose the source of connection drops.
My router is telling me that after it is powered up, my connection speed is 7-8Mbps, but after about 5-10 minutes there is a build up of CRC errors, at which point the connection drops. The router re-establishes a "stable connection", but at a lower speed of 4-5Mbps or lower.
A BT engineer has paid a visit, but did not find any fault with the line when he ran some tests with a handheld Exfo detector. However, he did connect his computer to the test socket via my router, and saw that the router reported rising CRC errors (http://www.kitz.co.uk/adsl/error_correction.htm#CRC), which he said should not be happening. The engineer then said this was the fault of the router, but I explained to him that I have run the same test with another router, and the connection problems were the same.
Obviously the engineer did not detect a line fault, but my router says there is a line fault that is giving rise to CRC errors. My question is, are there any tests that I could run to diagnose the reason why my connection drops after powering up my router?
Here are some sample router stats just before the connection drops while I was connected to the test socket:
System Information 06:58 18/05/2012
-----------------------------------------
Router Description
Product Name Huawei EchoLife HG520b
Physical Address 00:21:63:70:7d:0e
Software Release V100R001B020 TT
Firmware Release 3.10.11.0-1.0.7.0
Batch Number RTC10P0.020.310110
Release Date 2008/06/20
ADSL Description
ADSL State Show Time
Data Path Interleaved
Operation Mode ADSL2+
Bandwidth Down/Up(kbps) 9231 / 736
SNR Margin Down/Up(dB) 9.4 / 12.2
Attenuation Down/Up(dB) 35.0 / 18.7
Power Down/Up(dBm) 0.0 / 11.3
CRC Down/Up 1553 / 1
FEC Down/Up 46885 / 0
HEC Down/Up 6304 / 0
System Up Time 0:17:38
DSL Up Time 0:16:30
PPP Up Time 0:16:30
-
You could try updating the firmware on your router. I'm using the same router as you (albeit on a BT connection) and the Release Date of my firmware is 2009/10/17 where yours is 2008/06/20. Plus my firmware number is higher. Not sure if that would help as I'm having similar issues which is how I ended up here.
Heres my router info to compare:
System Information
Item Description
Product Name EchoLife HG520b
Physical Address b4:82:fe:81:71:bd
Software Release v2.08
Firmware Release 3.10.21.0-1.0.7.0
Batch Number RTC10P0.028.310210
Release Date 2009/10/17
ADSL Description
ADSL State Show Time
Data Path Interleaved
Operation Mode ADSL2+
Bandwidth Down/Up(kbps) 13225 / 423
SNR Margin Down/Up(dB) 6.9 / 4.8
Attenuation Down/Up(dB) 37.0 / 38.9
Power Down/Up(dBm) 0.0 / 12.3
CRC Down/Up 346 / 64524
FEC Down/Up 18215 / 159
HEC Down/Up 2631 / 0
System Up Time 15:04:02
DSL Up Time 2:30:49
PVC Select
PPP Up Time 2:30:48
-
Hi Alzy,
Thanks for taking time to respond. The firmware on my router is old because it lay in storage for about 2-years, but I don't think that this is the problem. The reason I say this is that before I used the Huawei router, I experienced the same connection problem with another router (Linksys WAG54GS): after powering up, the download speed starts out at around 8Mbps, but the connection drops, and then the download speed is reduced to 6Mbps or less. I used the Huawi router to replace my old router in case it was the problem, but since the connection drops regardless, I don't think I have an equipment issue.
Talktalk who are my ISP have sent me a new router (D-Link DSL 2680/TT) on Friday, but the connection still drops when the router is the only equipment connected to the test socket. The good thing about the new router is that I can now monitor the connection over an extended period using Router Stats-Lite (http://www.vwlowen.co.uk/internet/files.htm#routerstatslite), so I can show you a graph that illustates my connection problem. The sudden drop is very difficult to diagnose, because a connection that drops could be due to a multitude of things, and I don't know what tests I can run to isolate the cause, but I suspect it is a contention (http://www.kitz.co.uk/adsl/contention.htm) issue.
-
Hi,
There are many others on here more knowledgable than me, but my view is that if there's one thing it isn't, it's contention.
My understanding is that whilst contention can/will slow down the throughput of data, it won't have any bearing on the speed that your modem synchs at. That ought to remain stable.
Clumsy analogy :
If you and your neighbours all share a one inch water main, and all your neighbours start using their hose pipes at the same time, your tap will take longer to fill a bucket than usual. But the water main is still one inch, you just get less out of it while demand is high. That's contention.
What you're seeing is the "size" of your "water main" actually reducing slightly.
And since the noise margin is dropping, and the modem then resynchs, I'd suggest line noise or line fault.
Ian
-
A good analogy, Ian. I'll have to remember it. :)
Let us consider Slow's graph of the DS SNRM versus Time. We start by making the reasonable assumption that there is no fault with the DSLAM (or MSAN). The parameter plotted on the Y-axis is the difference between the configured target SNR and that of the current measured SNR. The signal level at which the DSLAM is communicating with the modem can be considered to be a constant.
Y-axis = ((target SNR) - (measured SNR)) ~= ((constant1) - ((constant2) / (current Noise)) ~= 1 / (current Noise)
Rearranging the above, we see that --
(Y-axis) x (current Noise) = constant3
Thus for the value plotted on the Y-axis to start to drop, then rapidly drop to zero before recovering must imply that there is an relatively sudden, then overwhelming, noise event on Slow's line.
Identifying and eliminating the cause of that noise event is somewhat of a black-art. :-X
What could be happening in that half hour period? :-\
-
Thanks for responding!
Yesterday, I spoke with my neighbour in the flat across the landing, who is also having connection issues (but with a different ISP), and he rulled out contention as the source of the problem for the same reasons outlined by Sheddy. Previously I had suspected that my neighbour might be the source of the fluctuating download speed, but he explained to me that since we are connected to the BT exchange by seperate lines, contention between neighbours could not happen.
So now we can turn to Burak's diagnosis, which I intially received badly, because a sudden, then overwhelming, noise event could potentially be very hard, if not impossible, to trace, especially if the cause of the noise event was random. To see if I could identify when or why the noise event was occuring, I ran Router Stats-Lite for 24-hours on my work laptop whilst connected to the test socket without connecting any other equipment to the router or phone line.
As a result, I think I have isolated the source of noise, but I can't believe what the evidence is telling me. Whilst connected to the test socket via my laptop, I went into the living room and turned on my personal computer (not connected to the internet), and a few minutes later the noise margin started dropping! I turned the computer off and on several times, and each time the noise margin dropped a few minutes after power up, and improved as soon as the PC powered down. Not only that, but the noise margin dropped in direct proportion to the length of time the PC was left on. This is shown in the attached graphs, from which you can see that the noise margin drops and at one point became overwhealming, at which point the connection dropped and the download speed plunged to zero.
My deduction from this is purely guesswork, so alternative opinions would be most welcome: I suspect that my ADSL2+ connection is being disrupted by a steady stream of noise broadcast by the Power Supply Unit (or some other component) of my personal computer, and this is being transmitted to the phone line via the power supply cables in my flat. Note that my router is situated next to my laptop and the test socket in a box room, whilst my PC sits in the living room across the hall. All these tests were conducted on my work laptop which, together with the router, were the only appliances connected to the test socket at the time.
At this point, I feel like Victor Meldew by saying "I don't believe it!, "I just don't believe it!", but I can't think of any other explaination. If my PC's PSU is the source of the noise, then what should I do?
-
Just a few immediate thoughts on rf noise generated by computers :
What happens if you switch the "noisy" pc on, but leave the monitor powered down? (Is it LCD monitor?)
Are all the connections on the PC good? anything loose? What does it have plugged into it?
Are all the side panels on the PC attached properly? Are there any holes in the case that shouldn't be there (ie gap where expansion card was removed)?
If you are familiar with/happy with changing bios settings, is there an option for "spread spectrum" in the BIOS? If there is, is it switched on or off? What happens to your noise margins if you change it?
Computers are very noisy, but there are various rules intended to keep interference down to a minimum. Yours does seem excessive!
Ian
-
If I understand Sheddy correctly, the Radio Frequency noise he is refering to is the same as or caused by Repetitive Electrical Impulse Noise (http://www.kitz.co.uk/adsl/rein.htm) - please correct me if that is not the case.
I am guessing that my monitor is not a significant source of the RF noise, if only because it is attached to an insulated power cable that has a ferrite sleeve (which acts as passive RF filter), whereas my computer does not have one. There may be other components that are contributing to the noise (such as my speakers, but they were not turned on during the test), however my personal computer is likely to be be the noise source as it has a 700W Power Supply Unit.
I am also guessing that my computer would not have to generate an excessive amount of RF noise to disrupt the ADSL2+ signal because the phone and power supply lines enter my flat at the same place and follow the same path to my box room that contains my test socket and fuse box.
My first step might to buy a new surge protector that incorporates an RF filter, but I am not sure if it would be effective if the filter only keeps outside noise from "getting in" and not stopping noise from "getting out" of my personal computer. Does anyone know anything more about REIN and how to suppress it?
-
From your experimental evidence, I agree with your assessment. That other PC's PSU is the source of the RFI. It could be that there is a component within the PSU that is starting to "break down". The most obvious test would be to perform a direct swap out of the PSU . . . however not everyone has one tucked away "for a rainy day".
If you feel confident enough, it might be worth removing the PSU from the computer and, after allowing sufficient time since it was last powered on for the capacitors to discharge, open it up to closely scrutinise the components for any signs of defect. Ends of capacitors bulging? Signs of arcing present anywhere? Signs of resistors starting to pyrolyse? Anything that just "doesn't look right"?
-
I am guessing that my monitor is not a significant source of the RF noise, if only because it is attached to an insulated power cable that has a ferrite sleeve (which acts as passive RF filter), whereas my computer does not have one. There may be other components that are contributing to the noise (such as my speakers, but they were not turned on during the test), however my personal computer is likely to be be the noise source as it has a 700W Power Supply Unit.
My first step might to buy a new surge protector that incorporates an RF filter, but I am not sure if it would be effective if the filter only keeps outside noise from "getting in" and not stopping noise from "getting out" of my personal computer. Does anyone know anything more about REIN and how to suppress it?
I may have guessed wrong, which means that Sheddy must be right - the source of the noise may be the monitor.
I think I can rule out my PC as the source of the noise because, but when I power up/power down my LCD monitor (17" made by the Neijiang Direction Liquid Crystal Display Co.) the interference comes on/disappears but the PC does not have the same effect. Even though there are ferrite sleeves around both the monitor's power supply cable and VGA connector cable, they don't appear to be having any effect, so my guess is that the monitor itself must be radiating interference which my router is picking up directly.
I have installed a new surge protector (with an EMI/RF filter) which is now connected to the monitor, but this has not reduced the interference, nor has fitting a ferrite sleeve to the router's cables have had any effect either. The only (tiny) improvement I have noticed is when I have moved the router further away from the monitor. However, I live in a really small flat, and no matter where the router goes, it is never going to be more than 25 feet (7.6 meters) away from the monitor.
My next guess is that the only long term solution is for me to eliminate the interference by investing in a monitor with a low EMI/RF emission rating. In the shorter term, I am stuck with net SNR of 12dB, i.e. Gross SNR of 18dB minus the equivalent of 6dB that is taken away by the interference. In the short term, it seems that I can only train my router to achieve a download speed of 5Mbps so long as my net SNR is stuck at 12dB, unless anyone else has any better ideas.
-
Perhaps the next experiment to perform is to see if the interference is being transmitted via the mains wiring from the monitor to the modem/router or if it is being radiated via "free space".
Assuming you could create a suitable lead to connect to the power socket on the modem/router, it might be possible to run that device (for a iimited, testing period) from a suitable battery. If the modem/router requires 12V DC, I would normally suggest borrowing a car battery! However a more "up-to-date" forum member (whose ID has slipped my mind) has suggested using a rechargeable power pack from a modern hand tool -- a cordless drill, perhaps.
-
The fact that the interference lessens on moving the router away from the monitor would tend to suggest the RFI is being transmitted over 'free space'. Although personally, I'd be looking to change the monitor, if this is not possible, then wrapping the modem/router in tinfoil (and its associated wiring) and then grounding it may make for a solution, albeit not a pretty one !!
-
I was going to suggest, as the next experiment in the sequence, wrapping the monitor in aluminium foil! We are obviously following the same thought pathway. :)
-
I wrapped the router in tin foil as suggested, and Router Stats-Lite shows that there was a small increase in SNR of about 0.25dB (i.e. a small reduction in broadband interference), which suggests to me that what this problem is attributable to REIN (http://www.kitz.co.uk/adsl/rein.htm) being radiated via "free space" as Burak and Blackeagle are suggesting. I have also come across a study of EMI in monitors* that suggest the interference is radiated.
Although I can't rule out the possibility that my phone line, modem power supply might or RJ45 cable might be picking up the interference and transmitting it to my router, the addition of a ferrite sleeve to the router connectors does not seem to have any effect on the SNR, so I don't think the interference is being transmitted via the wiring.
I think the only solution is to buy a new monitor. My current (cheap) 15" monitor uses a 15-pin VGA connector to receive analog video signal from the DVI socket of my computer's GPU, but the connecting VGA to DVI cable is well insulated and has ferrite sleeve at both ends, so I don't think the cabling is the source of the interference. I suspect that only by purchasing a digital monitor that uses Digital Visual Interface (DVI) or High-Definition Multimedia Interface (HDMI) connectors at both ends will eliminate this problem.
* "EMC Analysis of an 18'' LCD Monitor", Theodore M. Zeeff et al., (2000): PDF can be downloaded here (http://www.clemson.edu/ces/cvel/pdf/EMCS00-169.pdf).
-
Thank you for the link to the reference text and for the update of your current status.
The one thing that has been slightly troubling me is that you started to experience the problem earlier this year but you have not indicated that it coincided with obtaining and using that monitor. Hence I am wondering if it is the monitor's internal (probably switching) power supply that has become defective? Expanding on that thought, perhaps take a look inside the monitor for bulging capacitors? :-\
It might be a cheap repair, if you are proficient with a soldering iron. ;)
-
...or £60-ish for a new 19" ws if affordable ;)
-
I can confirm that my broadband problems began in February which is when I started using the "noisy" monitor. Previously I had a HP monitor with a DVI connector, but it burnt out, so I had to replace it at short notice by borrowing the "noisy" monitor from work. Little did I know that it would cause me so much grief.
Opening the monitor and examining the capicitors is beyond my competence, so I will pass
on Burak's request, apologies for that.
However, I am planning to buy a new monitor soon (maybe next month), so hopefully the problem will be resolved then. If not, I will return! In the meantime, I offer thanks to everyone who took time out to respond to this thread: your support was appreciated :)
-
Thank you for the latest update. I am now convinced that we have got to the bottom of this issue.
Perhaps that explains the reason why your work was willing to loan you that particular monitor . . . ;)
Do please "drop in" when you are next passing. We try our best to be a friendly community and a happy looking tortoise is most welcome. ;D
-
I bought my current 22" ws XEROX for under £60 from my local CEX shop. They have a web presence here (http://uk.webuy.com/product.php?mode=buy&catid=924).
@b*cat
I would have suggested wrapping the monitor, but thought it rather large and more troublesome to do ! I also considered suggesting a faraday cage although I think this would impede the view of the monitor. It would most likely cure the problem though :D
I agree with you that it's most likely the switch-mode PSU in the monitor causing the issue.