A Cautionary Tale

[Tacitus]

A while ago my standard router, a Zyxel 662H ceased to make a DSL connection from power up.  I assumed it had failed and moved over to a spare 2700 on the same line.  Some 3 weeks later this also failed with exactly the same symptoms - power cycle it and it failed to gain a lock on the DSL signal. 

Needless to say I contacted support and we went through the, 'try this that and the other' routines to no avail.  It seemed reasonable to me that for one router to fail was unfortunate, for two to fail in the same way in a relatively short space of time was too much of a coincidence.

I took the Zyxel to my sister's and tried it on her iDNet line.  From power up to sync took just over a minute and I left it functioning for an hour or so before swapping it back.  I've since had it functioning on her line for 3 weeks with no problems and swapped it for the 'failed' 2700 which has now been going for 3 days also without apparent problems.

So, we now have two routers - different makes and chip - both of which function perfectly well, but not on my line.  I should add that both mine and my sister's lines are iDNet and both run Max, not 21CN.

After discussion with support I elected to get BT out.  Apparently I was risking £160+vat.  I didn't like it, but by this time I had, not unreasonably, become convinced the problem was at the exchange.  BT guy arrives plugs in his modem and hey presto it works.  He performs a few more tests and goes away - I should add that I plugged in another borrowed modem whilst he was there and that also worked.

Now to the tragic bit.  Apparently unannounced BT have raised their call-out charges and I have been hit with a bill for £195+vat (=£229.12).  Just what I need in the run up to Xmas!  According to support it would have cost less if I pulled a wire out of the faceplate - circa £30 apparently.

I have now been told by support that they have seen occasions when modems suddenly become incompatible with the line.  Apparently it is rare but it does happen. It might have been useful if they'd mentioned this earlier, although I doubt they imagined it would happen twice in the same place with different modems. 

So, the moral of this tale is that when iDNet want you to jump through hoops then do it.  Although it didn't save me, it might save some of you a serious bill.

[Rik]

Sorry to hear about that, Tac. It's perhaps worth mentioning that the bill will rise to £239 from January, due to VAT increases. At the same time BT is moving to external demarcation points, which remove the ability to 'prove' a line at the test socket. This will mean BT are even more likely to raise a charge.

Coupled with Simon's messages that, after four weeks, BT have still not given IDNet an explanation for the October outages, makes me think that they treat small ISPs with contempt.

[Tacitus]

Coupled with Simon's messages that, after four weeks, BT have still not given IDNet an explanation for the October outages, makes me think that they treat small ISPs with contempt.

I think BT treat everyone with contempt.....

[Rik]

True. In that respect at least, it was better when it was still State-owned. When you consider their failings for Scotland, Northern Ireland and the north of England, and the capacity problems we've seen for early FTTC adopters, just as we saw for early WBC adopters, you are left to ponder what it is exactly we have Ofcom for.

[Steve]

Aren't the 2 Wire and the Zyxel 662H not built around the same chipset i.e the TI AR7?

I know the Zyxel is but I'm unsure of the 2 wire however there is the odd reference to a chipset via google

http://forum.kitz.co.uk/index.php?action=printpage;topic=5505.0   post#2 mentions the TI AR7

If correct it explains why two different routers ended up with the same result.

[Tacitus]

Aren't the 2 Wire and the Zyxel 662H not built around the same chipset i.e the TI AR7?
I know the Zyxel is but I'm unsure of the 2 wire however there is the odd reference to a chipset via google
http://forum.kitz.co.uk/index.php?action=printpage;topic=5505.0   post#2 mentions the TI AR7
If correct it explains why two different routers ended up with the same result.

TBH Steve I don't know.  I'm aware the Zyxel uses the AR7, in fact most of the Zyxel range do, but I wasn't aware the 2700 did.  In any event the post you refer to mentions the 2700 on 21CN and both mine and my sister's line are using Max. 

Sis can go to 21CN, but since she has a solid setup, I'm inclined to leave well alone.   

[Steve]

It was just the mentioning of the chipset in the 2nd post i.e TI AR7 that raised the possibility of why two routers should be incompatible with your line's Exchange DSLAM.

[Rik]

Sis can go to 21CN, but since she has a solid setup, I'm inclined to leave well alone.   

Probably wise, Tac. I have gained 1M from 21CN, but I was lucky enough to be in the trial which allowed IDNet to override DLM. Before that, it was all over the place.

[Tacitus]

It was just the mentioning of the chipset in the 2nd post i.e TI AR7 that raised the possibility of why two routers should be incompatible with your line's Exchange DSLAM.

Both myself and Sis use Zyxel routers, although at present we both have 2700s.  I have one with V6 firmware whilst she is currently testing my 'failed' 2700 with V5.

I might hop over to the DSLReports forums and see if anyone there has any idea.

The thing that bugs me about all this is that despite being hammered for £200+, I still have no idea what caused the problem in the first place.  From what Rik said, BT are turning their lack of maintenance and shoddy service into a business opportunity.

You couldn't make it up....

[Tacitus]

Probably wise, Tac. I have gained 1M from 21CN, but I was lucky enough to be in the trial which allowed IDNet to override DLM. Before that, it was all over the place.

Her attn is around 41dB, so she might gain a little speed.  Given her current sync is usually around 6700, although it has dropped to nearer 6000 with the bad weather, there is little to gain and, potentially a lot of grief.  Frankly not worth the bother, since she doesn't really need the extra bandwidth allowance.

[Rik]

The thing that bugs me about all this is that despite being hammered for £200+, I still have no idea what caused the problem in the first place.  From what Rik said, BT are turning their lack of maintenance and shoddy service into a business opportunity.

You couldn't make it up....

Curious that BT don't get engineers to give the customer a form to sign, agreeing to and detailing the work to be done. When my alarm system is inspected every year, I get a form to sign indicating what work has been done or is necessary.

Maybe we should start a petition to Ofcom that they insist that BT do this, which would also prove that they had actually turned up at the premises. After all, they only raise these charges for work done after the test socket, so if they fix the problem at the exchange or along the way, they don't need to attend the customer's premises do they? I can't remember, in the past, being charged for a voice engineer to plug his phone in and do a line test - what's the difference with ADSL. BT say the demarcation point is the test socket, so establishing whether there is a fault at that point should, surely, be part of the service we pay for. Just as it used to be.

[Rik]

Frankly not worth the bother, since she doesn't really need the extra bandwidth allowance.

I agree, Tac, it's why I'm beginning to think that going to fibre just isn't worth it to me.

[Tacitus]

Maybe we should start a petition to Ofcom that they insist that BT do this, which would also prove that they had actually turned up at the premises. ........ BT say the demarcation point is the test socket, so establishing whether there is a fault at that point should, surely, be part of the service we pay for. Just as it used to be.

I agree Rik.  BT turn up and perform some tests - which for all the customer knows could be complete rubbish - and then go away and do a report.  Nothing is agreed by the end user and the bill is unknown until the time it arrives.  Given it can vary between £30 at one end and £230 (ish) at the other, these are the actions of the average cowboy outfit.  There should be some system whereby the cost is known before any work is done.

I was, albeit reluctantly, prepared to accept the risk of £160 or so, but somewhere along the line it should have been made clear that the actual bill was likely to be significantly higher.  Now whether this is iDNet's fault or BTs is a moot point.  All I know is that I've been stung for £200+ and still don't know the cause of the problem.  In practice of course the answer is keep buying modems....

TBH given my kit seems to work on other BT lines, I can't really see how BT can claim my kit is faulty.  If as Steve says above, it is an AR7 problem, then why does the current 2700 work on my line without problem?

[Rik]

I wish I could give you a sensible answer, Tac, but I can't. Possibly the engineer fixed the fault that didn't exist?

[davej99]

Aren't the 2 Wire and the Zyxel 662H not built around the same chipset i.e the TI AR7?

According to ASL24 the 2-wire is STMicroelectronics and the Zyxel P660 series is TrendChip. This rather rules out the chipset.

[Steve]

I can argue about the P662HW

http://en.wikipedia.org/wiki/TI-AR7

but I've no idea about the 2 Wire  BT 2700HGV chipset. Although the STM is used on the 2 wire home portal.

[zappaDPJ]

The thing that bugs me about all this is that despite being hammered for £200+, I still have no idea what caused the problem in the first place.  From what Rik said, BT are turning their lack of maintenance and shoddy service into a business opportunity.

You couldn't make it up....

I suggested the very same thing elsewhere today and what makes it worse in my mind is that BT are using the ISP as a collection service and a buffer between themselves and the customer. It's an abysmal system.

[kinmel]

The 2700 uses it's own ADSL modem chipset, blown onto an ARES chip.

[davej99]

The 2700 uses it's own ADSL modem chipset, blown onto an ARES chip.

Sorry to take the thread a bit of course, but chipsets are increasingly of interest. I wonder Kinmel, if you have some links we could steal to get more info in this area. Maybe we should all be opening up out of warranty boxes; or maybe posting pics after we have stamped on rubbish routers. Cheers, Dave.

[Den]

Just taking up Riks point about prices going up in Jan due to the VAT rise. The cost then would be £234 not £239 as Rik said, he was using MK maths 

[kinmel]

Sorry to take the thread a bit of course, but chipsets are increasingly of interest. I wonder Kinmel, if you have some links we could steal to get more info in this area. Maybe we should all be opening up out of warranty boxes; or maybe posting pics after we have stamped on rubbish routers. Cheers, Dave.

I opened up a V5 and used Google to identify and rule out each and every chip on the board except one, that chip has a paper label marked "589-005-045 ( ARES ROHS )  and under the label the 2Wire Logo is etched onto the chipface

You can download a large scale photo of the chip and it's markings from  HERE


and this is what the 2700HGV looks like inside

[Tacitus]

The 2700 uses it's own ADSL modem chipset, blown onto an ARES chip.

That bears out what I found here.

[MisterW]

So, we now have two routers - different makes and chip - both of which function perfectly well, but not on my line.  I should add that both mine and my sister's lines are iDNet and both run Max, not 21CN.

I really can't believe that 2 different routers can be faulty in a way that causes them not to work on your line but to work fine on another line. Common sense would say that its an exchange fault, but I guess BT have a distinct lack of that!. Going back to your earlier post regarding the 'Suspicious VXCO offset' report on the 2700, my guess would be that there is possibly a fault on the line card affecting the frequency and that some modems (and BT's test gear?) can cope with the fault better than others...

[Rik]

Just taking up Riks point about prices going up in Jan due to the VAT rise. The cost then would be £234 not £239 as Rik said, he was using MK maths 

I was putting my markup on, Den.

[Tacitus]

I really can't believe that 2 different routers can be faulty in a way that causes them not to work on your line but to work fine on another line. Common sense would say that its an exchange fault, but I guess BT have a distinct lack of that!.

Agree entirely, which is why I went for the engineer visit.  I wouldn't have believed it, but that's what the tests show.  Given iDNet support never mentioned 'incompatibility' until post the engineer visit, they presumably discounted it on the grounds that one might fail but it's unlikely that it would apply to two different routers.

Going back to your earlier post regarding the 'Suspicious VXCO offset' report on the 2700, my guess would be that there is possibly a fault on the line card affecting the frequency and that some modems (and BT's test gear?) can cope with the fault better than others...

Oddly enough the current 2700 (V6) which has been pretty stable for a while now, reports the VCXO offset as being -112.6 ppm.  The old (failed) one reported it as -120 so I wonder if that represents a trigger point.  According to Zyxel their modems don't use VCXO offset, so that would discount it as the source of the problem for the 662H.

The failed 2700 has been running on Sis line since Sunday lunch with no apparent problems - at least I've not had an anguished phone call yet....    

[Rik]

There's only one way to fix this. We all buy one BT share, then go to the next BT AGM and disrupt it by asking pointed questions.

[cavillas]

How about we all contact our respective MP's and put these BT problems to them, at the same time contact TV, Radio and Papers to push the point.  Perhaps greater publicity and pressure will eventually have some effect on BT.

[Rik]

I feel that's the kind of action we need to take, Alf. BT think they are untouchable, we need to make them learn they are not.

[Lance]

Maybe we need to have a template letter written for users to download, sign and post?

[Rik]

It had gone through my mind, Lance.

[pctech]

I think the one BT share each is a good idea.

[davej99]

I opened up a V5 and used Google to identify and rule out each and every chip on the board except one, that chip has a paper label marked "589-005-045 ( ARES ROHS )  and under the label the 2Wire Logo is etched onto the chipface

Found THIS SITE helpful. It confirms:
Ares 3001-000485 [ DMT Modem/Framer/Controller]  2Wire  HomePortal ADSL2+

[MisterW]

According to Zyxel their modems don't use VCXO offset, so that would discount it as the source of the problem for the 662H.

They must use some sort of VCO to allow for slight frequency deviations in the signal from the exchange otherwise it would hardy ever synch at all. I suspect it's just that they don't display any offset like the 2700 does.

The failed 2700 has been running on Sis line since Sunday lunch with no apparent problems - at least I've not had an anguished phone call yet....   

What does it show for VCXO on her line ?

[Tacitus]

They must use some sort of VCO to allow for slight frequency deviations in the signal from the exchange otherwise it would hardy ever synch at all. I suspect it's just that they don't display any offset like the 2700 does.
What does it show for VCXO on her line ?

I can only go by what Zyxel tech support told me, but I think you are probably right.

Regarding the 2700 on sis line, I can't tell you at mo but I'll take a look when I go over there at the weekend.  My own 2700 is currently displaying -112.6ppm.  The old one consistently showed -120 together with a warning that something might be wrong.

[Tacitus]

According to ASL24 the 2-wire is STMicroelectronics and the Zyxel P660 series is TrendChip. This rather rules out the chipset.

According to Zyxel the chip in the 662 (and presumably in the others in the P600 range) is an Infineon AR7.  Apparently Infineon bought Texas so they fall under one umbrella.

[kinmel]

I can only go by what Zyxel tech support told me, but I think you are probably right.

Regarding the 2700 on sis line, I can't tell you at mo but I'll take a look when I go over there at the weekend.  My own 2700 is currently displaying -112.6ppm.  The old one consistently showed -120 together with a warning that something might be wrong.

MY V6 2700  shows  VCXO Frequency Offset:    7.8 ppm    

Is that good, or bad.

[Glenn]

This any use? http://www.analogzone.com/nett0626.pdf

[davej99]

According to Zyxel the chip in the 662 (and presumably in the others in the P600 range) is an Infineon AR7.  Apparently Infineon bought Texas so they fall under one umbrella.

Infineon is the Microelectronics spin out of Seimens AG, who had acquired the DSL product range of TI, I think. So I guess Infineon chipsets are descendants of the old AR7. In the case of 2wire it seems to be an in house design, with firmware originally from TriMedia, a Philips IP spin out and later superceded. This proprietary 2-wire design may be fabbed by STM's foundry in Singapore on a basic design rule set like ARM. The only thing I know about Ares, as on the 2 wire chip, is as a design tool for pcbs and as a flat panel maker. It may be a defunct silicon foundry - don't know. Like most current products, system design, firmware, silicon design and silicon foundry are no longer vertically integrated. So all of the bits of chipset info we have make a kind of sense but we have no way of knowing what it really means as bystanders.

Apologies to Mods. We should be having this chat in the networking section. Will leave it here.

[kinmel]

This any use? http://www.analogzone.com/nett0626.pdf

thanks for that, but a bit beyond me     

[Glenn]

Was for me too 

[pctech]

Infineon is the Microelectronics spin out of Seimens AG, who had acquired the DSL product range of TI, I think. So I guess Infineon chipsets are descendants of the old AR7.

You are correct Dave.

[davej99]

MY V6 2700  shows  VCXO Frequency Offset:    7.8 ppm    

Is that good, or bad.

Now this is an arm waving guess; but I think the ADSL signal demondulation requires a local oscilator, which is kept in sync with the transmitter. A voltage is applied to a local crystal to adjust the local oscillator frequency by up to about 200ppm, but far less in normal operation. I read that 100ppm is high suggesting the demodulator is running on the edge. So 7.8 ppm should be OK. No facts were harmed in the making of this BS!

[Tacitus]

........A voltage is applied to a local crystal to adjust the local oscillator frequency by up to about 200ppm, but far less in normal operation. I read that 100ppm is high suggesting the demodulator is running on the edge. So 7.8 ppm should be OK.....

This begins to make some sort of sense.  The problems I was having were mainly when doing a power cycle.  Unplug the mains and powerup again and neither modem would sync.  Well they might eventually - it took 5hrs on one occasion.  At that point when sync had been gained (VCXO -120 so it was probably on the edge) I could do a software reboot and it would resync with no problems.  It was mainly when booting from cold that the problems occurred.

What you say about a voltage being applied to the crystal, suggests the modem was not applying the correct voltage until it had warmed up, possibly combined with a fault at the exchange which meant the target frequency range was too high.  Mind you that argument fails, since it only took minute or so from a cold start to gain sync on sis line, so why should it be any different on mine?

Dunno.      I suspect that unless someone very knowledgeable put both modems on a test bench and used a frequency generator and oscilloscope, we shall never know. 

It could be that both modems have developed a similar fault, in which case I've been very unlucky. However, given the nature of the fault I suspect there has to be something going on at my exchange as well.  Getting BT to check that out, is a forlorn hope and, given the new costs imposed by BT it's not a risk I'm prepared to take.  Given iDNet's current relationship with BT, I doubt they would be willing to push it any further and, given I currently have a working line, I can't altogether say I could blame them.

[Rik]

The interesting thing would be to 'scope the signal at your master socket, Tac. I suspect that's one side of 'normal' while your routers are the other side. If you compared the signals at your house and your sister's, you'd likely see a significant difference, imo.

[davej99]

What you say about a voltage being applied to the crystal, suggests the modem was not applying the correct voltage until it had warmed up, possibly combined with a fault at the exchange which meant the target frequency range was too high.

Look on the applied voltage as fine tuning of the crystal oscillator from its nominal figure to demodulate the signal received. If it has to stretch too far. say well over 100ppm that could indicate a faulty local oscillator and it might fail to lock and so demodulation will fail. But on two modems?

Now I am just speculating here, because I have no sure knowledge of the workings of ADSL; but I wonder if it might indicate a faulty transmitter on your card at the exchange, if both modems showed a high ppm, yet a nominal ppm on other lines.. We really need an expert to help us here because I am well out of my depth; sorry. So please do not let me send you off on a tangent.

[Rik]

I agree with you, Dave, I'd bet that the card is out, but that most modems will still work with it so BT will do nothing.

[Tacitus]

I agree with you, Dave, I'd bet that the card is out, but that most modems will still work with it so BT will do nothing.

I agree too Rik, but the only way to find out would be as you suggest, to 'scope' the signal at the test socket.  Unfortunately there's no way to find out what the VCXO reading is for the Zyxel.

I think davej99 is right on the mark.  The card at the exchange is probably sufficiently off for both my modems to fail and yet be sufficiently near for other modems to work.  Given the variety of modems, chipsets and firmware, this is entirely possible.

It really needs a high level engineer (and I do mean an engineer) on the case, but I think Hell will freeze over before that happens.

[Rik]

And thaw again, Tac.

[Glenn]

On my 2700 it reports a VCXO of  2ppm

[Tacitus]

On my 2700 it reports a VCXO of  2ppm

Which, since mine show, -120 (the 'dud' one) and -112.6 (the current V6) suggests that the theory is broadly correct.  -120 seems to be right on the edge since it prompted the 2700 to post a msg saying something looked wrong.  Current 2700 is not far off at -112.6 but is obviously far enough from the danger zone to enable it to work.

The only way we will get to the bottom of it is as Rik suggests - put a scope on the main socket and see what the readings are.  Alternatively BT could look at what's happening in the exchange and try swapping the card, but we've already ruled that out.....

[Rik]

For the record, my VCXO is -9.6ppm.

[Tacitus]

For the record, my VCXO is -9.6ppm.

It all points to an exchange problem.

[Rik]

It does, Tac, but we know BT won't admit that. One day, the linecard will fail and you'll find the number drops dramatically.

[davej99]

Which, since mine show, -120 (the 'dud' one) and -112.6 (the current V6) suggests that the theory is broadly correct.  -120 seems to be right on the edge since it prompted the 2700 to post a msg saying something looked wrong.  Current 2700 is not far off at -112.6 but is obviously far enough from the danger zone to enable it to work.

The only way we will get to the bottom of it is as Rik suggests - put a scope on the main socket and see what the readings are.  Alternatively BT could look at what's happening in the exchange and try swapping the card, but we've already ruled that out.....

If we can get two modems to show normal, IE a few ppm, on a known good line and >100ppm on your line, Tacitus, then I think we have a case, or at least an arguable case, to go back to BT and say, "Sorry, we have grounds to believe the exchange card was/is out of spec and therefore the call-out was justified."  You could ask, "Are you sure you can rule out an unstable remote oscillator causing high local ppm's and my modem to loose sync?"  

There is no hope of using a scope because I think the transmission covers phase modulated tones across the spectrum from 100kHz to over 1MHz and I am speculating somewhere in the mush there is a reference tone to lock in your local demodulator. But I am just guessing. It will take some complex kit to do the job. You might get some insight from RouterStats, which sometimes uses Telnet to give bits per tone, bit swaps per tone and SNR per tone. It might be you will see high bits swaps if demodulation is unstable.

At the end of the day you need a dialogue with a BT and enough bluff to rattle their cage. Perhaps argue, "I have high local oscillator ppm's on two proven good modems, surely there is still a problem." Or maybe, "I saw high ppm's on two proven good modems when I called you out - surely there was a problem?"

Sorry, I can't be much help. I will go away and brush up my understanding of ADSL - but I may be some time. Then I'll write a book "ADSL for BT Dummies." Mayday, Mayday, Mayday. Request assistance. ADSL expert needed. BT need not attend!!

[Rik]

 

The problem always will be that BT are judge and jury. To prove them wrong would almost certainly cost more than the fee they have charged. It might be worth having a word with Trading Standards though.

[davej99]

I think the lesson for the future, as a user group, is to prepare a case before calling in BT. For example if we had got our heads together on VCXO's at the start, proving high ppm's using two routers on two different lines, then it might have been possible to argue the case for a no charge call. In other words, instead of asking BT to tell us if there was a fault, we ask if such and such is a fault.

The call out charge is now becoming a real issue. Is it designed to force customers to live with problems? Is it a way to boosts profits? Is it challenging the customers sense of trust? Does it raise in our minds the notion that we are being had? May be we should start writing to Watchdog because public opinion is one of the few avenues left open.

[Rik]

I suspect it's a way of boosting profits, Dave. Openreach is the one monopoly BT still have.

[pctech]

definitely a way of boosting profits as landline use is falling.

[Rik]

Exactly.

[pctech]

Was a little annoyed to read that O2 awarded a contract to manage the backhaul for the mobile network to BT.

[Rik]

Ironic, really.

[Ted]

If it helps at all, the VCXO Frequency Offset: on my line is  -7.3 ppm. 2700hgv V5, on a good line with 8128 sync.

[pctech]

Ironic, really.

But then of course it was in fact BT Wireless though as I've said before I've had no complaints and been with them over 4 years now.