Mystery - the main PC on my Home Net makes the ADSL line drop when it starts up!

[LesD]

I am new to this forum but am a member of many others although Tiscali have just shut me out of their's for getting a MAC and joining IDNET 

I am posting because I have a mystery with the main machine on my two machine home network in that the ADSL line drops when the main one is starting up! The second PC on my network does not do this no matter which one you power up first.  I use a Netgear DG 834 ADSL Modem/Router and this did not happen when I had a 2 Mbps fixed rate broadband connection. Both machine run XP Pro with all updates that are available.

With TiscaliMax my IP profile was 3000 kbps but in the evening my downstream speeds were sub 0.3 Mbps. As soon as I was connected to IDNet I was getting 2.8 Mbps downstream speeds all day to begin with but I think because the main machine drops the line each time its boots up this IP Profile is now down to 2500 kbps and my downstream speeds have dropped by 500 kbps in sympathy. 

I know that I really should wait the 10 days BT "training period" but this afternoon curiosity got the better of me and I re-jigged things always round without any explanation coming to light other than that my main machine and to a lesser extent its monitor are generating noise in a way I do not comprehend. Mains born seemed to be one possibility.

At one point this afternoon I had the second machine syncing at 5684 kbps and a 6 db noise margin! I had to look three time as I did not believe what I was seeing having never connected previously at more than 4000 kbps as viewed from the main machine As soon as I switch on and start the other machine, however, the ADSL line drops (the ADSL LED goes out and come on again a few seconds later) and the modem re-syncs at a lower speed that was typically 3500 kbps but at what is now this evening, a dire 2592 kbps! Yes I have made matters worse! 

I am presently using the second machine that is upstairs to see if the BT training will "recover" while this machine is on alone with these numbers:

Connection Speed 3904 kbps 448 kbps
Line Attenuation 40 db 13 db
Noise Margin 10 db 20 db


This afternoon I untangled the downstairs network cables from the mains cables that the main machine uses, run this main PC and its monitor on an extension cables from a socket way up the hall miles away from where the Netgear Router is located, re-situated the router on the floor about 2'6" below where it usually sits and it was here that I first saw the 5684 kbps Connection Speed using the second machine. This may have just been a coincidence that this freaky high speed occurred when I moved the router. I did see connection speeds above 5100 kbps a number of times with it on the floor. It was 5150 before I went for my tea with a 3db noise margin but when I looked after tea it had reconnected at 3808 kbps so it must have succumbed to noise but the router was still on the floor and the main machine was still off.

At one point I even wrapped the dc input supply wire to the Netgear router from the wall-wart once through a "ferrite" core but no benefit was perceptible.

Anyway with extension cables and wires here there and everywhere I decided to give up and put things back as they were but not as tightly bundled as before. Like this as soon as the downstairs PC starts up the line drops and the connection (sync) speed is now a dire 2592 kbps! I have tried a couple of times and there is no difference. So who knows what the BT line training has made of todays antics! At the moment this upstairs PC is connected through Router port 1 and the main machine is in router port 2, which is the opposite way round to how I previously had them as another process elimination.

Every time I try some things and the line drops a few times BT dock me 500 kbps from my IP profile and then my throughput speeds are that much slower.

I have just run the BT speed tester from the second machine and this is what it gives me this evening:

    IP profile for your line is - 2500 kbps
    DSL connection rate: 448 kbps(UP-STREAM)  3904 kbps(DOWN-STREAM)
    Actual IP throughput achieved during the test was - 2284 kbps

I did have an IP profile of 3000 kbps before I started investigating the problem with my main machine!  So I am at a loss as to how to proceed. Any help or ideas would be very welcome. 

[J!ll]

Welcome, Les, sorry I can't help! ex Tiscali here too  am sure someone will be around soon to help 

Karma for you 

[Sebby]

I think you've as good as proven that the main PC is causing the router to lose sync, probably due to noise. Perhaps it has a faulty PSU. Do you connect it via a surge protector? That might help.

Telephone extension wiring is generally very prone to noise pickup. You could try removing the ring wire from each socket to see if this stops the issue from occurring. For every socket on the line, open it up and pull out the ring wire (always connected to terminal 3, and probably orange in colour). The other thing that is probably wise is to get a high quality RJ11-RJ11 lead to connect the router to the phone line, as the ones supplied with routers are of poor quality and will pick up noise.

With the number of times a day this issue is probably occurring, your profile is never really going to have a chance to recover. The training has nothing to do with it; the only reason there is a training period is to establish a MSR (Maximum Stable Rate) and from that a FTR (Fault Threshold Rate). The service is constantly rate-adaptive. Therefore, we need to work on finding/eliminating the cause of the re-syncs, then you'll have a stable connection and the profile will recover.

I hope this helps as a starting point.

[LesD]

Hi Sebby,

Thanks for the reply and the advice .

The bell wire and the wire on termianl 4 have been removed at the matser Socket and I tried a CAT5 cable with RJ11's on each end from an ADSL filter in the inner test socket of my BT Master Socket directly to the Router when struggling with Tiscali's sub 0.3 Mbps evening peak time down-stream speeds! My internal telephone extension wiring was done years ago for voice telephony but my tests with the CAT5 cable eliminated it as a major issue and of course when the second PC is on alone all is well. The max. my local exchange is reportedly able to let me reliably sycn at is 4500 kbps so I know that I will not see any really high speeds.

All sorts of things are in my mind about the main PC tonight.
Fans for one, as one case fan has a speed controller. Was it a simple resistor one I can't remember. It is a couple of years since I built this machine. The power supply was new not long before Christmas and an overly generous 650 watt one at that but it could still be noisy I guess.

In Stafford we suffered a nasty series of mains supply interruption one evening last summer that caused people problems at home and at work and my main machines' LAN port on the Motherboard failed not long after this event so I put a PCI LAN card in and it does seem to have been doing the business OK ever since. I put an in-line mains surge arrester in series in the PC mains lead shortly after that and have tried with it in and out today! That said something is amiss tracking it down what it is, is the challenge.

I was concerned about some locking down on speed that I have read about that BT do if they determine that your line is bad but I am out of my depth here. 

[Sebby]

The ring wire is terminal 3, Les. Also, you must remove it from every socket; removing it only from the master socket won't help.

That aside, I just want to clear something up. Speed and sync are very different. With Tiscali, you were suffering speed issues, due to their over-subscribed network. On IDNet you won't suffer that, but your sync is obviously going to limit the possible speed. Also, it's not your exchange that supports 4.5Mb; that's your line (and it's only a rough estimate, though you should be able to achieve at least that). Your exchange supports up to 8Mb; it's your line between the exchange and your house that determines the rate you can sync at. In addition, it looks like there's something going on inside your property that is causing the quality of your line to deteriorate more.

I'm not following what you tried with Cat5. Are you saying you moved the second PC to another room and ran a Cat5 network cable to it from the router in its original position?

Usually, I'd advise people having noise issues to try the test socket, but I think we've already seen that when you connect away from the PC, the sync is much improved.

For now, I think we should start by getting the ring wire removed from each socket so we know that's not picking up unnecessary noise along the way, and take it from there.

[Lance]

Hi, Les, and welcome to the forum.

Thought I would point out that you won't have another training period, due to being on a MAX product prior to migrating.

One thing you could try is detuning a tranny radio and walking around with it or holding it next to the pc. You'll be able then to here if it is emitting lots of noise.

[Simon]

  Les!  You're in good hands here, and I'm sure these guys will get to the bottom of your problem. 

[Inactive]

Welcome Les, have a Karma.

[Ray]

Les,  and Idnet have another from me.

[Rik]

Hi Les

Welcome to the forum.

What Sebby and Lance have said is good advice. My own instinct is the PSU, and the tranny radio will confirm if that's the case. It's probably worth mentioning it needs to be on the MW band and de-tuned, then move about with it listening for an increase in white noise or static.

[cavillas]

Welcom Les.  Did you turn off the mobo lan from the bios?  Just a silly thought but it can sometimes be overlooked.

[LesD]

Well Folks one and all I don't know where to start but never short of a word as you may have noticed here goes.

The welcome to the IDNet forum is overwhelming thanks to everyone.
At my time of life I am not losing sleep over the matter so there is time to bottom it out.

Today was another day and it brought another puzzle the "noisy" machine isn't noisy today"
With only it working this is what my Netgear Stats say right now:

ADSL Link Downstream Upstream
Connection Speed 2816 kbps 448 kbps
Line Attenuation 39 db 13 db
Noise Margin 12 db 21 db

I got up this morning with an urge to have this machine apart disconnect a couple of fans and swap out the New PSU for the old one that still works but needs a "kick start" sometimes to get it started.

The Noise Margin of 12 db made me hold fire even though the Connection Speed was low compared to what I have been seeing.  The only difference today of any significance is that I have the in-line mains surge arrester out of circuit but I am sure I have tried without it before.

I have not tried the second machine today other than to boot it up to the XP Welcome screen while watching the Netgear Stats on the first/main machine and they did not waver. The second machine is shut down again now.

To elaborate on what I did with my length of CAT5 cable with RJ11's on it:
My BT Master Socket with the removable bottom half of the face plate behind which is what I refer to as the inner test socket is in my hall at the bottom of the stairs.
The second PC that work as I believe it should is at the top of these stairs in a space on the landing.
The first/main machine in eight feet down the hall and round a corner in an "alcove".
The distribution for three CAT5 RJ45 network cables that run into two bedrooms and to the second machine on the landing all start from the back corner of this alcove. The fourth port on the Netgear router serves the main machine that shares the top of my desk with the router but they are at opposite ends about 3 feet apart with the monitor in between.

There is a telephone extension socket in this alcove as well (with its bell wire removed too) that the Netgear Modem is connected to via an ADSL filter. My relatively new CAT5 network cables and the old telephone extension wiring are all in the walls and under the floor upstairs out of sight.

To eliminate my telephone extension wiring I made up a longish length of CAT5 cable with a couple of RJ11 plugs on it.

I removed the lower face plate of my Master socket thereby disconnection my extension wiring, plugged an ADSL filter into the BT inner test socket and ran my newly made up CAT5 cable from the ADSL RJ11 socket in this filter straight into the ADSL RJ11 input socket on the Netgear Router. So there was nothing connected to the incoming BT pair other than my router.

Like this the Connection (Sync) Speed and the Noise Margin as displayed in the Router Stats were not significantly different than with the connection made through my old telephone extension wires.
In fact no different as far as I could determine.

I do realise that the sub 0.3 Mbps peak time speed I had with Tiscali was due to contention in their system but the ex-Tiscali folks reading this will know that Tiscali's first line of defence is to attack and that means always blame their customer kit/installation even in the face of irrefutable evidence to the contrary! Hence all my check to be sure as I could that it was indeed them and not me. I also appreciate the difference between Connection Speed and the Downstream Speeds speed testers like the thinkbroadband one measure. I did think that my exchange was limiting me to 4.5 Mbps max but can understand when told that it is a combination of things including my line. I am 1.62 km from the exchange as the crow flies.

Today I am relaxed and using the main machine just as it is. I will give the transistor radio check a go. This last time I booted the main machine up I had the Netgear router switched off (I normally leave it on all of the time) so the line negotiation was after whatever it is that the PC does to drop the line during its start up process but I still only got a Connection Speed of 2816 kbps whereas until I frigged with things yesterday it was 3500 kbps + but never with a noise margin of 12 db. 6 db was the highest I had seen until recently.

Oh bye the way support at IDNet told me I would undergo another 10 day BT trialling period when I asked if the one I had done with Tiscali was it.

I don't see how but can BT influence the Noise Margin my Netgear measures?  That is a does a lower connection speed equate to a higher noise margin and vice versa?


Thanks again for such a lot of replies and if I have missed something I will try and catch up later! 

[Rik]

Hi Les

Max starts with a target noise margin of 6db, and then gets the line to go as fast as it can and remain stable. If it can't achieve stability, it will try interleaving, and if that doesn't work, it will increase the target margin in 3db increments, to a maximum of 15db. It looks to me as if the line management software has seen the instability and reacted, increasing the target margin which, in turn drops your sync speed by 5-700k per 3db. Unfortunately, the only way to recover from this situation is to maintain a stable sync for 14 days (15 to be safe) after which you should be able to re0sync and find the target margin is reduced by 3db. Repeat until you reach 6db again.

Obviously, you need to resolve the problem in order for this to work for you.

[Sebby]

The Noise Margin of 12 db made me hold fire even though the Connection Speed was low compared to what I have been seeing.  The only difference today of any significance is that I have the in-line mains surge arrester out of circuit but I am sure I have tried without it before.

As Rik describes, BT set your target SNRM to 6dB initially, and this can increased in 3dB increments to stabilise your line. Two things could've happened here. Firstly, your target could have been upped to 12dB. But, I don't think you're experiencing enough disconnects for this to happen. More likely is that, due to the noise bursts that your line is suffering, you connected at a noisy time with a SNRM of 6dB, and since then the noise has subsided, and so the current SNRM is quite high. Probably if you reconnect now, you'll sync higher at 6dB, but then the connection will drop again once the noise resurfaces.

To elaborate on what I did with my length of CAT5 cable with RJ11's on it:
My BT Master Socket with the removable bottom half of the face plate behind which is what I refer to as the inner test socket is in my hall at the bottom of the stairs.
The second PC that work as I believe it should is at the top of these stairs in a space on the landing.
The first/main machine in eight feet down the hall and round a corner in an "alcove".
The distribution for three CAT5 RJ45 network cables that run into two bedrooms and to the second machine on the landing all start from the back corner of this alcove. The fourth port on the Netgear router serves the main machine that shares the top of my desk with the router but they are at opposite ends about 3 feet apart with the monitor in between.

There is a telephone extension socket in this alcove as well (with its bell wire removed too) that the Netgear Modem is connected to via an ADSL filter. My relatively new CAT5 network cables and the old telephone extension wiring are all in the walls and under the floor upstairs out of sight.

To eliminate my telephone extension wiring I made up a longish length of CAT5 cable with a couple of RJ11 plugs on it.

I removed the lower face plate of my Master socket thereby disconnection my extension wiring, plugged an ADSL filter into the BT inner test socket and ran my newly made up CAT5 cable from the ADSL RJ11 socket in this filter straight into the ADSL RJ11 input socket on the Netgear Router. So there was nothing connected to the incoming BT pair other than my router.

Like this the Connection (Sync) Speed and the Noise Margin as displayed in the Router Stats were not significantly different than with the connection made through my old telephone extension wires.
In fact no different as far as I could determine.

Okay. I was under the impression that doing this improved things. I do wonder if the tests you've done so far have been a bit misleading. As I mentioned above, your SNRM is probably fluctuating a lot, and so reconnecting the router away from the computer could have just caused your sync to increase if there was not a lot of noise around at that point.

I do realise that the sub 0.3 Mbps peak time speed I had with Tiscali was due to contention in their system but the ex-Tiscali folks reading this will know that Tiscali's first line of defence is to attack and that means always blame their customer kit/installation even in the face of irrefutable evidence to the contrary! Hence all my check to be sure as I could that it was indeed them and not me. I also appreciate the difference between Connection Speed and the Downstream Speeds speed testers like the thinkbroadband one measure. I did think that my exchange was limiting me to 4.5 Mbps max but can understand when told that it is a combination of things including my line. I am 1.62 km from the exchange as the crow flies.

I'm a similar distiance from the exchange - in fact a little more (with attenuation of 46dB) - and I actually have a target SNRM of 9dB (set manually by BT for stability) and sync at around 4.5Mb, so I'd expect you to sync at this figure or higher consistently. Tiscali's network is not helping your speed, but until we sort out the stability, your bRAS profile is going to be all the over the place anyway, so Tiscali's network is not the end of the world at the moment.

Today I am relaxed and using the main machine just as it is. I will give the transistor radio check a go. This last time I booted the main machine up I had the Netgear router switched off (I normally leave it on all of the time) so the line negotiation was after whatever it is that the PC does to drop the line during its start up process but I still only got a Connection Speed of 2816 kbps whereas until I frigged with things yesterday it was 3500 kbps + but never with a noise margin of 12 db. 6 db was the highest I had seen until recently.

I think I've pretty much covered this above. But I'll just add that with Max, you should be looking to have a pretty consistent sync at 6dB. The fact that it's 12dB is as I mentioned already - either your target has increased (I think probably unlikely) or noise has just reduced since you sync'd.

I don't see how but can BT influence the Noise Margin my Netgear measures?  That is a does a lower connection speed equate to a higher noise margin and vice versa?

They only control it to a degree. They control the target figure, which is their way of ensuring that your line syncs at a rate that it will be stable at. They don't control the actual noise; some of this will be unavoidable (i.e. external to your property) and some will be controllable by your (e.g. filtered faceplate, etc, as we mentioned last night).

If you use the normal cable that came with your router and connect to the test socket, can you always sync consistently high or does is fluctuate wildly even without extension wiring in the equation?

[LesD]

I have had an element of success with the transistor radio technique. 

First off there was so much noise apparently everwhere that I was in despair but soon I realised that some of it round the router was most likely the ADSL so I turned everything off and started them up one at a time. To cut a long story short with the mains supplies running to the left of my "alcove" and the router and its power supply on the floor (to get it as far away from the monitor as I can) at the right of the "alcove", the main PC will boot without dropping the ADSl line.

Thanks so far I know I would have got to this point without you guys!

It was because the main PC was dropping the ADSL line as it came up that I started changing things round but before I did I was syncing at circa 3500 kbps with a noise margin of 6 db pretty consistently as seen from the main machine. Now, however, I sync round 2800 kbps with a 10-12db noise margin so if I understand what you are telling me I have managed to get BT to reduce my sync speed and increase the noise margin to try to restabilse things after I started upsetting the apple cart!

What was a complete mystery to me is beginning to make some sort of sense now thanks to you guys but one thing that still puzzles me is that with the main machine turned off whatever BT has done in terms of sync speed and noise margin, the second PC upstairs still syncs at over 3500 kbps with the 10-12 db noise margin! 

What BT do surely cannot be PC dependent! 

[Rik]

The sync speed should be the same regardless of which PC you use. However, if the main machine is emitting noise, and you re-sync while it's on, it could mean a lower speed, with it off, a re-sync will not see so much noise, so will pick up the extra speed. Does that fit with what you're doing?

One thing to watch, don't re-sync more than 10 times in an hour, that will be seen as instability and will result in a dropped profile.

[LesD]

The sync speed should be the same regardless of which PC you use.

That is what I thought

However, if the main machine is emitting noise, and you re-sync while it's on, it could mean a lower speed, with it off, a re-sync will not see so much noise, so will pick up the extra speed. Does that fit with what you're doing?

The main machine is emitting noise, the transistor radio picks it up in the mains cables, which is why I moved them as far from the router as I could. I do not seem to be re-syncing inadvertently and this maybe because my Netgear is more noise tolerant than some other router or so I read on the Kitz site.

One thing to watch, don't re-sync more than 10 times in an hour, that will be seen as instability and will result in a dropped profile.

I reckon that I have already fallen foul of this one at least twice. Last Weds/Thurs when I dropped from connecting at 3800ish to 3300ish to now after yesterday and today to 2800ish!

I have probably got an IP profile of 2000 kbps by now. 

I will try and leave things as they are for a few days to see if my sync speed improves. 

I don't know whether it is best the leave the router permanently on, which has been my usual practice or to switch it off when the PC's are not in use. The main PC was causing a line re-negotiation each time it booted and that was at most twice a day but that was before I started my investigations I could easily have exceeded 10 re-connections in an hour while trying various alternatives!

[Rik]

Netgears are strange on noise, they fluctuate wildly in the margins they report, but they seem to hold the line regardless, down to about -2db in my case, which is technically impossible. The new v4 seems to be more stable in this respect.

Did you determine which component of the main machine is giving off noise, eg monitor or main box? Is there any swapping you can try to see if things change?

I always leave my router on 24/7 unless I'm away or there is lightning about. In your situation, it might help to power it down overnight, just to make sure you don't get any extra re-syncs, but it's a marginal thing.

[LesD]

Welcom Les.  Did you turn off the mobo lan from the bios?  Just a silly thought but it can sometimes be overlooked.

Hi cavillas,
Yes I did and when I rebooted I recall that good old Bill Gates announced that my XP was invalid as it thought it was in another machine due the some points score they have for a LAN and said I must go on line and re-activate it within 48 hours. Such a friendly chap to kick me when I was down. How was I supposed to do that with no LAN connection. So I nipped out to PC World for the PCI one and got connected with it and re-activated my Genuine XP I am pleased to say.  

[LesD]

Netgears are strange on noise, they fluctuate wildly in the margins they report, but they seem to hold the line regardless, down to about -2db in my case

I have seen 0 db and the line not drop but I don't think mine displays negative values. There is what looks like gobbledegook when the line does drop!

Did you determine which component of the main machine is giving off noise, eg monitor or main box? Is there any swapping you can try to see if things change?

Both the monitor and the PC were emitting noise. The PC or most likely its PSU seemed the worst of the two but the soap-on-a-rope printer PSU was noisy too. I still have not powered the printer up again. Despite thinking that I had bought a relatively good PSU just before Christmas I am beginning to have some doubts. I still have the old one it had become just a wee bit reluctant to start every time so if needs must I could put it back and see what sort of difference it makes but that's not at the top of my list right now as I do think I have improved matters.

I always leave my router on 24/7 unless I'm away or there is lightning about. In your situation, it might help to power it down overnight, just to make sure you don't get any extra re-syncs, but it's a marginal thing.

I think I will stick with the 24/7 approach as that's the way I am used to.

[Rik]

Netgears display negative numbers as a very high positive one, two million odd from memory. An experiment you might want to try - connect the old PSU to the mains, see if it is noisy. Easier than changing everything over.

[LesD]

Netgears display negative numbers as a very high positive one, two million odd from memory. An experiment you might want to try - connect the old PSU to the mains, see if it is noisy. Easier than changing everything over.

Ah maybe thats what I described as the gobbledegook! I don't think the PSU I have will run without a connection to a motherboard unless you know better that is?

I have just run my transistor radio over my walls where I know mains cabling is behind the plasterboard and the radio noise increases in the vicinity of cables pretty much everywhere. There must be enough attenuation or slewing by the time it reaches the PC upstairs for it to survive! I guess though that it is not all due to the PC as I know flourescent lights including the new low energy one emit some noise too!

I shall try another re-start shortly and see if my luck holds and the router holds the line!

[LesD]

I shall try another re-start shortly and see if my luck holds and the router holds the line!

Well so far so good that's another re-start from power off the PC and the Monitor to back on with the printer powered up too without the Router losing the line or re-connecting. It has held the line for 2 hours and 18 minutes now so it is looking promising. 

Thank again to all who have contributed and to Rik and Sebby in particular 

Well time will tell if my connection speed recovers!

[MoHux]

It may make you puff ....... but it would be interesting if you swopped over the upstairs/downstairs computers. 

It might prove/disprove something!!

[drummer]

A belated welcome Les.

You've probably gathered by know that I wasn't fibbing (via Tiswas PM) about this being a very friendly and extremely helpful forum.

A karma on the way and another one if you promise to never mention IDNet on the Tisc forum ever again.