Kitz Forum
Broadband Related => Broadband Hardware => Topic started by: jaypeecee on December 21, 2007, 05:58:24 PM
-
Hi Folks,
I've just read an interesting article in the latest (Feb 2008!) copy of PC Pro magazine. In the article, the author starts out by saying that he has the all-too-familiar line disconnect/loss of sync problem. After much investigation, he concludes that the problem is caused by three factors. Firstly, the use of a TI AR7 chipset-based router. The second factor is to be using an ADSL Max-powered service and the third ingredient is for some external factor that increases the line noise.
What I find interesting about this article is that it conflicts with my experience. I'm now using the Netgear DG834G v3 modem/router, which is based on the AR7 chipset. I chose it because the Kitz site specifically states that it performs well under low noise margin conditions. And does it perform well!! I've not had a single disconnect since I bought it three weeks ago. My noise margin is never in excess of 12dB. So, that takes care of two of the factors described in the PC Pro article. As for the ADSL Max service, I'm not sure if that's what I'm on. Because of my distance from the exchange, BT said the best download speed I could expect was 2Mbps. I get 1.9Mbps so I can't complain. Therefore, my question is - why does there appear to be a difference in performance between the PC Pro article and my experience with an AR7-based modem/router? Is it to do with the ADSL Max service?
JPC
-
I think the truth is that there are numerous possible reasons why a particular line may be unstable, and the author of that article is guilty of making too many assumptions based on his own personal experience. We've seen numerous reports from people saying that the DG834 with the AR7 chipset has proved to be the most stable router for them, because it holds sync down to very low noise margins. But there are others who say that it has critical flaws which make it a bad choice.
There's no single single router which is the best choice for everyone. As it happens, for me the Speedtouch ST576 is the best router out of many which I've tried, but I continue to suggest the DG834 as a good option for difficult lines.
It's certainly true that, for some people, going over to ADSL Max has destabilised what was previously a satisfactory fixed-speed connection. But for most people it gives higher speeds and still adequate stability. So once again there's no single answer to suit everyone.
-
But there are others who say that it has critical flaws which make it a bad choice.
And they are most likely the people complaining about it.
-
I think the DG834Gv3 is a very good router. It would sometimes go from 6db right down to zero and stilll hold the line.My DSLAM is TI and it seemed to work fine with it most of the time.
-
I think we are going to see more and more cases of people jumping on the AR7 bandwagon since Zen made that announcement about it.
OK, so there may be some factor relating to the AR7, certain types of line interference, whether the line is interleaved or not, what DSLAM is in the exchange, etc.. that under SOME circumstances means the AR7 does have a problem.
But think about the number of DG834 series routers and countless others based on the AR7 that have been sold... if every single one of those had a problem as Zen were basically alluding to, don't you think we'd have heard a tiny bit more in the way of complaints??!
It does annoy me when people supposedly in a position of trust and authority (i.e. Zen being a very well respected ISP in the UK) put out what is basically little short of scaremongery.
Just to add, my Netgear stayed synced on ADSL Max for many weeks at a time. It's likely the article's author has a problem with some bad external interference and is factoring the Netgear into the blame when it probably is nothing whatsoever to do with it. I bet he didn't try another router first!
-
mr_chris,
Far be it for me to defend the article's author as what he was saying conflicts entirely with my experience using the AR7-based Netgear DG834G v3. With reference to your point about 'I bet he didn't try another router first!', in fact he did - a SpeedTouch 546 wired ADSL2+ router - and this put an end to the line drops and re-syncs. I've just discovered that the article to which I'm referring can be found onlne at http://www.pcpro.co.uk/realworld/144378/route-to-ruin.html.
Out of interest, I was using a BT Voyager 2091 prior to the Netgear and that's why I made the change. The Voyager was losing sync every few minutes and it seemed to coincide with the wetter weather we have been having since September this year. Does anyone know what modem chipset is inside the Voyager?
Finally, it looks as though I'm on an ADSL Max line.
JPC
-
I think there have been some microcode updates* applied - some more effective than others.
I understand that Zyxel took a decision that the chipset was so flawed that it was uneconomical to support - I have a P2602HW which has the AR7 and that got binned VERY fast for the top of the range router (£150 or so) which they sold at the time.
Netgear obviously felt differently. Zyxel (Europe) development/support is in Denmark/Sweden, Netgear is a US company (as is TI) who sell shedloads of modem/routers to US ISPs - I'm not sure if that was a factor or not.
*this is usually where the processor has a hardware design flaw and so there's a patch applied. On AMD cpus it will be a s/w patch that has to be applied each time the machine boots - or more accurately AFTER the cpu boots; on Intel (recent) cpus its possible to patch the processor (sort of) directly.
-
I've answered my own question about the chipset in the BT Voyager 2091. Apparently, it's the Broadcom BCM6348. Having read more stuff on the net, I'm now of the view that there's a lot of opinion out there and not a lot of fact. So, I'll continue to follow the AR7 debate with interest.
In the meantime, I'm delighted with my Netgear DG834G v3 - does just what it says on the tin!
JPC
-
Hi
Theres a couple of other threads on this forum where we have dicussed the "AR7 chipset issue" announcement by Zen.
Since then, I have always asserted that although there may be something in the AR7 chipset theory.. I do believe that there is some other element behind the true cause.
There are far too many users who have reported that the dg834g has helped their connection.
Ive also been in touch with one of the guys from Zen about thishad some correspondence with one of the guys from Zen about this, there are also many more people who do not believe it is soley the AR7 chipset. Even Zen admit it is not the only cause.
If it was then Im sure there would be a heck of a lot more complaints about it - particulary since the netgear router is one of the most common routers out there. Without a doubt the Netgear does peform better on many lines than some other routers. - The old zyxel could be pretty carp anyhow, and there are some users who would have problems with a zyxel yet be fine when they put a netgear on the line.
>> So, I'll continue to follow the AR7 debate with interest.
Me too.
-
@ JPC - my little rant wasn't directed at you, in fact reading back it was a little bit more of a rant than I intended at the time!!
-
naw I think you may be right - particulary
"I think we are going to see more and more cases of people jumping on the AR7 bandwagon"
Ive already seen people accuse their router without checking out the more obvious causes - Ive even seen someone blame the netgear - but it was the dg834gt which doesnt have the ar7 chipset anyhow. :/
-
@ JPC - my little rant wasn't directed at you, in fact reading back it was a little bit more of a rant than I intended at the time!!
Have no fear, mr_chris. I didn't perceive your comments as directed at me, nor did I see them as a 'rant'. You were simply making some good points.
-
Good.. glad about that :)
I'd like to be able to do a bit more investigating on the AR7 problem. I have a feeling it's to do with when the line is interleaved, too. My Netgear wasn't as stable when my line was interleaved, and it didn't hold sync as well as when I got it put back on fast path, so perhaps that is an issue. I still think Zen should have performed some investigations (they are in a good position to do so, or they should be!) before releasing such a sweeping statement.
They should have started small i.e. "it affects AR7 with such-and-such a DSLAM and such-and-such a line profile" and allowed others to expand on it, rather than saying all AR7s are bad, and then everyone trying to cut through the noise to determine exactly under what circumstances!
So the PC Pro guy did try another router... did he say by any chance what make of DSLAM was at the exchange or whether he was on Interleaved or not? Those are the sorts of things that need to determine exactly what the flaw with the AR7 is.
-
Good.. glad about that :)
I'd like to be able to do a bit more investigating on the AR7 problem. I have a feeling it's to do with when the line is interleaved, too. My Netgear wasn't as stable when my line was interleaved, and it didn't hold sync as well as when I got it put back on fast path, so perhaps that is an issue. I still think Zen should have performed some investigations (they are in a good position to do so, or they should be!) before releasing such a sweeping statement.
They should have started small i.e. "it affects AR7 with such-and-such a DSLAM and such-and-such a line profile" and allowed others to expand on it, rather than saying all AR7s are bad, and then everyone trying to cut through the noise to determine exactly under what circumstances!
So the PC Pro guy did try another router... did he say by any chance what make of DSLAM was at the exchange or whether he was on Interleaved or not? Those are the sorts of things that need to determine exactly what the flaw with the AR7 is.
The thing is, it doesn't affect such-and-such a DSLam nor a line profile. What it does effect is a DSL Max line that has a fluctuating SNR. Whilst it is still being looked into, there are other reasons to suggest that the problem is also related to interleaving as this has also been seen to cause the problem.(Kitz was the first one to relate this issue to me)
As for Netgear not doing anything about it, I have been speaking with them at length and although we have not found a solution we are still testing and working with both Infineon (the makers of the AR7 chipset) and BT to find a solution to the ongoing issue.
There was a large discussion about this matter on the thinkbroadband forum and I am sure it will be discussed here at length as well. As for releasing such a sweeping statement, Zen didn't release it, The Register did. Phil D.Long was misquoted throughout that article.
So the PC Pro guy did try another router... did he say by any chance what make of DSLAM was at the exchange or whether he was on Interleaved or not? Those are the sorts of things that need to determine exactly what the flaw with the AR7 is.
The DSLam does not make a difference at all. The modem code and the firmware does however. For now we do not have an answer, but we are getting closer to finding a way to keep the same interoperability of the AR7 chipset with the stability of the Broadcom.
-
One thing that does occur to me with the Netgear (bear in mind I'm theorising here and dont have access to the large resource of differing types of lines/dslams like Zen does).. and Im just chucking a few ideas into the hat... and I could be talking rubbish.
... but the netgear is reknown for holding on to low snr lines when most other routers will have dropped the connection.
Quite often once SNR gets too low then errors start being introduced on the line. For a non-interleaved line obviously these will be CRC/HEC errors. For an interleaved line you can get tons of FEC errors... could there be something in that?
I still strongly believe there is another factor into this problem, but finding out what this is, is going to be the hard part.
Without a doubt some routers do perform better than others on lines.. there are some routers I would avoid... the Netgear DG834x isnt one of those. There really is just too many instances Ive either seen first hand or had reported to me that that particular router has helped stabilise long lines.
It could just be that the ST routers are better than say the Netgear ones. Theres also many many rumours that "pairing" the dslam + CPE chipsets do seem to provide better results.
>> The Register did. Phil D.Long was misquoted throughout that article.
Hmmm yes on reflection that does make some sense, it wouldnt be the first time.
Phil does carry a certain amount of respect and knows his stuff.. hence why some of us were surprised at such a "broad sweeping" statement.
I havent been on the zen own forums recently but perhaps an updated statement by Phil on "Home Turf" could perhaps clarify some issues.
>> The DSLam does not make a difference at all.
Interesting - thanks for that.. I had wondered if there was something in the dslam/MSAN thing.
>> the firmware does however
Yes - which could also be why IMHO say the Netgears are better than say the old zyxels, which I'd seen perform not as well as others.
DSLmax pushes most lines to their limits, and the "roots" of most firmware goes back to pre-max days.
>> For now we do not have an answer, but we are getting closer to finding a way to keep the same interoperability of the AR7 chipset with the stability of the Broadcom.
Thanks for that.
-
Just to add my 2 pence's worth.
I was running a DG834v2 for 9months on ADSLMax and generally had few issues.
My exchange's DSLAM chipset was / is an Alcatel. Another perhaps useful fact was that my distance to exchange (according to Sam Knows) was about 750mtrs. Interleaving on my line was not set.
I recently changed to a Speetouch 585v6. This unit equally has been very stable. I have noticed that browsing is faster though.....
Seasons Greetings
-
Over on Thinkbroadband is a few posts about this issue regarding the AR7 Chipset.
http://bbs.adslguide.org.uk/showflat.php?Cat=&Board=unhappiness&Number=3223070&page=0&view=collapsed&sb=5&o=0&fpart=6#Post3236240
Well over my head, but interesting, Zennops post starts it off, 5th post page 6, hopefully the link will take you straight there.
-
I have a question for the Forum, (taking off my Zen Cap)
Who's responsibility is it when a modem does not work a with a line due to an error in the line itself and/or the DSL code used by the modems chipset?
This is basically the question that has been posed in that discussion.
And to sum it all UP:
Subject Re: Alternative to ZEN?new [re: alext05] [link to this post]
Posted by Standard User SkyFire (isp)
Posted on Tue 15-Jan-08 11:01:34
Hi,
Zen's hardware does not play a part in this issue; the AR-7 issue deals with communication between ADSL Router and DSLAM - which is dependant upon the hardware/firmware the customer has and the equipment BT Wholesale have deployed.
We are in communication with a hardware manufacturer on this issue and we are making progress on this issue; a firmware for the hardware in question that improves/resolves this issue is likely to come about as a result of this work. In testing it does appear to resolve the problem.
regards,
Phil.
--
Phil D.Long
ZeN Technical Support Manager
iMac 24" and MacBook Pro 15.4" - Leopard-tastic
Zen Internet Technical Support are recruiting click here for details.
Please note, I will not respond to unsolicited private messages.
-------------------------IMPORTANT---------------------
The opinions expressed here are my own and do not belong to Zen Internet.
-
As someone who make be suffering due to this issue, here is what I think.
If it is a line fault then it clearly at BT issue. The practical problem is that this has to accessed through the ISP interface, which can be highly tedious.
If the issue if it is DSL/Chipset code is tougher. In an ideal world your line characteristics would be measured, e.g for noise margin variations throughout the day, attenuation etc. Dependent on the characteristics measured then either the ISP or BT should recommend a selection of compatible modems or routers, plus a realistic assessment of what speed a customer can expect. This assumes that the ISP/BT knows how different routers behave with different line characteristics. I would guess that many ISP's do not have this information, or possibly the equipment or desire to acquire it. BT must have the equipment and technical expertise to do this, but they seem reluctant to publish any in-house testing. So, if the customer has a less than perfect line which many clearly do particularly in more rural areas, he may welll find himself in position of having to buy or borrow several routers to see if he can find one that works. This process is likely to result in complaints to ISP's plus BT engineer call outs, to say nothing of acute frustration for the customer, who is likely to end up feeling that he is stuck between a rock and a hard place (ISP & BT), and that neither of them give a F***. For my money BT is the obvious choice for providing this information. They maybe reluctant on confidentiality grounds, but if they could classify the line on some scale or other and then recommend a selection of routers for a particular class of line, then I feel that confidentiality becomes less of a problem. Yes I know it's more work for them, but they'll save on call outs. Just needs a bit of recording kit you plug in to your line for a week before you go onto broadband.
-
Thank you all for your reply's however you have gone a bit off topic.
I will start a thread about routers and get your opinion about them, but for this thread can we stick with the question of responsibility please.
-
Hmm.. the question of responsibility that you raised was a bit of a shift from the original topic. :) We'll stick with it now, but let's ALL try to keep it tidy.
I'll split the router messages into a separate thread (when I've worked out how to do it).
[Edit] Now done, topic entitled 'Comments on routers'
-
>> Who's responsibility is it when a modem does not work a with a line due to an error in the line itself and/or the DSL code used by the modems chipset?
Tough question in that format. AIU when the exact circumstances when the AR7's start misbehaving havent been disclosed.
If its a line fault then BTs responsibility without a doubt.
BT should have also tested that their DSLAMS work with the current CPE's on the market place. If there are certain routers that dont work well with MAX because they dont do A,B or C then the public should have been advised by BT. Max went through a long testing phase and many problems werent ironed out properly.
Whose silly idea was it to do most of the testing on lines that could only previously support 2Mb? More long lines should have been included in the trials!
We arent talking about "Noddy routers" here we are talking some pretty damn big firms and BT should have worked closer with the router manufacturers... we all know that MAX is about pushing your line as much as it can.
If however its a fault in the firmware which isnt responding properly to any DSLAMS reqeusts, then the fault lies with the manufacturers to update their firmware.
-
We arent talking about "Noddy routers" here we are talking some pretty damn big firms and BT should have worked closer with the router manufacturers... we all know that MAX is about pushing your line as much as it can.
Isn't ADSLmax an international standards-based service? If that is the case, if a bit of kit works in most/all the world except here, I'd say it was BT's fault.
I do not believe that Max is about pushing a line as much as it can per se, it is about doing so whilst keeping max stability. I'd sacrificed some stability for more oomph any day ;).
Cheers
DrT
-
@Azzaka
I don`t blame you for quoting the post as you`ve done.
But I was just trying to broaden the debate about AR7 chipsets.
The posts on TB has some interesting comments, it seems that Zen blames the AR7 Chipset as some sort of De-Facto answer, whilst there isn`t any absolute proof.
There was a mention that perhaps the Netgear may be complying with the specs & that the DSLAMS may not, which might then suggest that others may not?
I fail to see how your quote sums it all up!!
Zen is on record as stating that the AR7 chipset is the problem, but hasn`t any real proof that it is the real problem.
Maybe as a layperson, I`ve misunderstood things???
-
Personally i would like to know where Zen said that the fault is definitely the AR7 chipset and never said that the lines have a lot to do with symptoms. The reason the AR7 issue came to light was because Zen customers started being charged by BT for having a piece of kit that did not work on the line. Yes it could be a DSLAM issue, however when the AR7 chipset doesn't work however a Broadcom does work, then is this the DSLAM or the DSL code on the chip?
The reason i summed it up was due to Netgear releasing a new firmware to us that does work on a line where the previous versions did not work. At this point i decided to pose the question of blame. This seems to be the point that most people like to make, is that it was easier for Zen as a company to blame the Chipset rather than the DSLAM, which by all means i can see the point, however if changing the modem code allows the AR7 to sit stable on a line it previously wouldn't do, is this still Zen blaming the Chipset because its easier to do so?
-
The problem is of course that Zen were widely misquoted (or quoted out of context) when they publicly raised the point about the AR7 chipset, and perhaps they didn't make their point clear enough at the time. The impression which most people had was that Zen were simply blaming the chipset.
When the new firmware is released I presume that Zen will give it their public approval.
-
As far a i Know. BT will be giving their approval and Zen are hoping that netgear will also go on record with them.
-
The problem is of course that Zen were widely misquoted (or quoted out of context) when they publicly raised the point about the AR7 chipset, and perhaps they didn't make their point clear enough at the time. The impression which most people had was that Zen were simply blaming the chipset.
I must admit this is the impression that I got too when I first read the el reg article.. and as Ive already said a couple of times... its something that I feel that Zen perhaps should make a clear statement about on "home turf" where it cant be misquoted/misinterpreted out of context in a tabloid type context that it was.
As a result the Netgear routers (or any AR7 router) suddenly became the scape-goat for just about anything wrong with a line and too many people appeared to jump on the bandwagon AR7 issue.
Yes their very well could be something in it, but it is a combination of factors... and not the doom and gloom blanket that all Netgear router are bad. This and other similar forums have far too often seen the DG834G perform well on some long lines and stabilise some lines which other routers couldn't.
What seemed to happen was practically overnight the Netgear was automatically assumed to be the baddy, without people trying any of the other normal diagnostics that should be performed first. :/
-
I agree with Kitz and she is well aware, Zen have a wiki article about the AR7 chipset that states we will raise the Fault to BT as it could be the line at fault and not the chipset and in most case is the line.
Zen specify the chipset as a precaution as we do have customers who have been charged by BT for this exact issue. As Kitz and many others have said before the problem is not clear cut and dry and can not be pin pointed by Zen, which is why Zen never did release a statement about it. The Reg article had taken points from other sites and then made its own mind up, however a thanks has to be said to the Reg because without them Netgear would not have contacted me, nor would i have been in contact with BT or infineon.
-
FWIW here's a quote from Zen on the Zen Support forum:
We recommend and supply Speedtouch hardware as we have a good working relationship with them and can provide the optimum support on their kit and know from experience that it works very well with Zen and for our customers. Not only that but if an issue arises they work with our technical staff to resolve to our satisfaction.
On the other hand we do have many other customers who will swear by other makes and models.
What we will always do is provide where reasonably possible as much information we can to our customers about what we know and see day to day when troubleshooting faults and I hope our customers will continue to work with us to do that.
In respect of the AR7 chipset concerns this came about after our technicians had been troubleshooting intermittent connection faults on a rate adaptive service and where the SNR was experiencing minor fluctuations. In their observations they found that with any of the customers who used speedtouch kit or non TI AR7 chip devices the sync remained solid holding the line but anything that appeared to have the AR chipset dropped the sync and had a detrimental effect on the performance.
The chipset apeared to be the only common theme and was raised as it "Could" be a cause not "it definitly was".
http://forum.zensupport.co.uk/1/28529/ShowThread.aspx
-
There will be a new Press Release shortly regarding the AR7 issues.
-
Interestingly Netgear's latest routers have moved to the Broadcom chipset, that's sums it all up for me ;)
-
They moved there after the issue with the Version 3.
They have since then, fixed the issues and i would not be surprised if they moved back to AR7 with the new design that is being released.
-
I think this has the potential to be very damaging for the AR7 chipset. It's a VERY good ADSL chipset, from my experience, arguably every bit as good and reliable as the Broadcom.
Will be interesting to see what the DG834Gv5 may bring ;) (There's nearly as many sequels to the DG834 as Police Academy :lol:)