Still having problems

D-Dan · Apr 23, 2008, 20:33:17

You may remember about a month ago I had some serious problems with disconnects which reduced my connection to a pitiful 256K for a week before it came up. Before that I'd been rock solid for 6 months at 7,200K sync with a profile of 6000.

Since then, I have not been able to sync at more than 5200K, and the max my profile has been was 4000. But it gets worse. My sync starts to drop now (no disconnects) until I forcibly reboot the router and get a higher sync. Currently, my sync is 2048 - less than a third of my previous stable sync. My SNR is 14 (and that's according to my Netgear router - which always understates the SNR - so it's probably closer to 17).

Last week, I had a new heating system installed, but I can't blame that since the sudden spell of warm weather means my heating isn't on.

I've switched to the Netgear from my speedtouch router in case that was to blame, though clearly not. Apart from that, nothing else has changed internally.

I'm now going to go back to the Speedtouch (since it is generally better), but I'd like my old sync back (or at least something approaching it).

Ideas?

Steve

Edit: OK - with the Speedtouch - sync is now 5120 with SNR 12 (which I trust is more accurate). But I can guarantee that it will be down to around 3k tomorrow night

Sebby · Apr 23, 2008, 20:54:19

Unfortunately, it sounds like you've got a case of an intermittent source of noise surges, which are notoriously difficult to resolve.

Other than having a BT engineer out (after all, it could be a line problem) the only thing you can do is make sure everything on your side is as good as can be. SpeedTouch routers are great, so I wouldn't say it would be worth getting another router.

Have you tried the test socket?

D-Dan · Apr 23, 2008, 21:06:49

I really want to avoid the expense of a BT engineer if I can, though unfortunately, my test socket is especially difficult to get to directly. I have no power outlets anywhere close to the test socket, which means I'm stretching wires to their absolute limit to get there. Do-able - but if it can be avoided, I prefer to.

What's most annoying is that, until the announced engineering works, I had no problems and a rock solid fast connection. I'm still inclined to think BT are responsible, but I don't fancy paying £170 to have them lie to me about it.

Glenn · Apr 23, 2008, 21:12:49

Quote from: D-Dan on Apr 23, 2008, 21:06:49
What's most annoying is that, until the announced engineering works, I had no problems and a rock solid fast connection. I'm still inclined to think BT are responsible, but I don't fancy paying £170 to have them lie to me about it.

Thats when my problems started too

Sebby · Apr 23, 2008, 21:31:07

Quote from: D-Dan on Apr 23, 2008, 21:06:49
I really want to avoid the expense of a BT engineer if I can, though unfortunately, my test socket is especially difficult to get to directly. I have no power outlets anywhere close to the test socket, which means I'm stretching wires to their absolute limit to get there. Do-able - but if it can be avoided, I prefer to.

I'll be honest with you; given the current state of affairs, I'd say it can't really be avoided. We have to rule out internal wiring as the cause, and to do that, we have to connect to the test socket. Otherwise, it's a BT engineer (possible at the expense of £160+) or you just leave things how they are.

D-Dan · Apr 23, 2008, 21:34:05

OK - I just sent an email to IDnet pleading with them (in no uncertain terms) to tell BT to fix the damn problem (since I'm so sure that's where it is).

However, to prove the point, before I go to work in the morning, I'll cable stretch 'til I'm blue in the face and connect to the test socket. Then I'll re-check stats when I get home tomorrow (approximately 11 hours later).

Sebby · Apr 23, 2008, 21:38:16

If at all possible, give it as long as you can in the test socket. We must prove that it's not something inside your property, which at this stage, we don't know. Unfortunately, IDNet can only ask BT to test the line, and if they find no fault (highly likely) IDNet are powerless. That said, it may well have nothing to do with BT.

D-Dan · Apr 24, 2008, 18:54:17

Just an update for the moment, but the ever helpful Miriam has replied to my email, after speaking to a BT engineer for me. He has checked their logs and noted and drop in SNR at around (I don't know if exactly or approximately) 8am most days, which is what accounts for the drop in sync (though no loss of sync).

Now I'm still flumoxed, but I have something more to work with. I've done some brain racking to try and think what might be causing a problem at around 8am, but I can't think of anything that would only happen at that time. Generally, weekdays I leave for work at about then, although having been off work last week tends to rule out something such as turning off the radio, a light etc. since the pattern would have changed completely last week.

Street lighting can't be the problem, since there are no streetlights in close proximity to home, so I guess the hunt is on.

On a more positive note, she got him to reset my target SNR (it had been set high) and I'm currently synced at 6,944, which is closer to what I'm used to, and only slightly below my pre-problems sync. If the 8am phenomenon continues then this may get screwed again, but now I know "when" I'm looking for I have more chance of finding something.

I'll keep you posted.

Once again, IDNet come through, and whilst the problem may not be solved, at least they gave me a whole lot more info to work with that I didn't have.

As an aside, does anyone know of a reliable method of logging SNR so that I can review the patterns myself. I know I could do it with DMT but I'd rather find a method that doesn't require the PC to be on 24/7.

Steve

Rik · Apr 24, 2008, 19:02:26

Only Routerstats, Steve, but that needs the PC to be on. I'm wondering whether the router is failing to log the sync drops?

Sebby · Apr 24, 2008, 20:20:32

It sounds like that could be the case, Rik. RouterStats should catch it regardless.

D-Dan · Apr 25, 2008, 23:14:28

OK - first 24 hours using RouterStats - and no significant change. Sync is stable and speed is where I like it to be. Maybe I'm an anomaly, where the lower target SNR is good for business

We'll see over the next few days

Sebby · Apr 25, 2008, 23:24:04

It does seem strange that things are stable on a lower SNRM, but it could just be that the noise source isn't around at the moment. Still, I hope things continue like this.

D-Dan · May 06, 2008, 20:25:50

An update - and something of a request for advice. Over the last couple of weeks I've been getting the reducing syncs and increasing SNR. A resync generally brings it back up - but still less than 4Meg.

Anyhoo, I've localised the source of the noise, after finally spotting a pattern - and oddly - it's PC2. I turn it on and SNR drops from the current 20 to 9 or 10 almost instantly.

I can rule out the PC itself, since yesterday I replaced the MB, CPU, RAM and, more importantly, PSU.

Which suggests something external to the box.

I have a cable from the monitor plugged into a pass through for the speakers, and finally to the mains, with a second mains straight to the tower. My personal suspicion is that the cable from mains to monitor is the most likely suspect, since it's home grown, but I'd like a second opinion on this before I start rewiring things.

There is very little plugged into PC2 (Just a cam, headset, ethernet, speakers, mouse and keyboard (both Logitech wireless). Power runs to the mouse/keyboard base unit even when the PC is off so I'm tempted to rule out RF interference from that.

Internally there are two IDE HDs, two IDE DVDRWs, one case fan, one UV light, and the graphics card (Oh, also replaced yesterday).

Could someone with a little more experience in noise tracking point me in the direction of the most likely candidate of the noise source before I ring Miriam and ask her to have the SNR target reset (again).

TIA

Steve

D-Dan · May 06, 2008, 20:51:53

Scratch that - I decided to do the disconnect something, test, rinse and repeat test.

After going through everything - I found it.

It's the damn VGA cable - the only cable I don't have a spare of

Unfortunately - the PC is kinda useless without it. Looks like I'm going to PCWorld tomorrow to get a decent quality one.

Wait - I still have the old one that was a bit flakey - but better than nothing

- Not the cable - must be the monitor

Yep - confirmed - hit the power button - SNR jumps up again. Any ideas of a way to shield it? 15 inch Belinea LCD.

Steve

Simon · May 06, 2008, 20:55:57

Not heard of that before, Steve. Might be a stupid suggestion, but is there any way you could move the router away from the monitor?

D-Dan · May 06, 2008, 20:57:31

I've tried putting some distance between them - I'll try for as much as I can get - but I'm limited by the ethernet cable between PC2 and router.

Edit:

The only cable I haven't swapped out is the monitor power cable. Now, I have one somewhere, though I've disconnected the monitor completely and am simply RD'ing into PC2 ATM. However, from people more knowledgeable than me - what's more likely to be causing noise (before I turn the flat upside down trying to find the shiny new cable I have somewhere) - the monitor - or it's power cable?

Steve

Sebby · May 06, 2008, 21:44:25

I no expert when it comes to curing noise issues, but I would suggest changing the power cable, when you can. The current one could be faulty and sending noise into the electrical wiring.

Rik · May 07, 2008, 08:56:33

My best guess would be that it's the inverter for the backlight, Steve. Try moving a de-tuned MW radio around the monitor and see where the noise is worst.

D-Dan · May 07, 2008, 22:41:42

OK - well it's definitely the monitor itself - I've swapped out every cable I have on that PC. Disconnected the monitor and the problem goes away - connection has been rock solid since last night.

That means for the moment the only way I can use PC 2 is via remote desktop - which is not ideal - but better than nothing.

Rik - and if I find the problem - is there a way to shield it? The monitor works fine - just kicks out too much noise.

Steve

Sebby · May 07, 2008, 23:08:12

At least you've found the cause of the problem, Steve. That's got to be worth something, even if it is a major annoyance.

Rik · May 08, 2008, 08:40:23

Quote from: D-Dan on May 07, 2008, 22:41:42
Rik - and if I find the problem - is there a way to shield it? The monitor works fine - just kicks out too much noise.

Almost certainly not, Steve, but it would make it unfit for purpose, so you could claim against the retailer for a replacement.

D-Dan · May 08, 2008, 09:18:11

Well that blows it then. It was an end of line deal from Tesco, purchased for me by a manager at a staff auction for the princely sum of £10, and didn't come with a warranty.

I'm not too concerned with the £10 - I reckon I've had more than money's worth from it. What bothers me is, if I can't source one on freecyle or an inexpensive refurb, is the cost of replacement

C'est la vie.

Steve

Rik · May 08, 2008, 09:21:07

The only other thing, and it's potentially dangerous, is to try shielding the unit with aluminium foil, Steve. Possibly a TV repair shop could isolate and replace the noisy component, but you're then getting towards replacement cost.

D-Dan · May 08, 2008, 17:13:17

As if further proof were needed, after two days with PC2's monitor off, and email to IDNet to ask them to get my SNR target reset - this just happened:

And as final proof, I hooked my my old CRT monitor, and it didn't hit my BB connection one little bit (though it frequently hit me on the side of the head on account of it overhanging my side of the desk too much - so it's back in the cupboard).

Much as I hate the reason, I'm quite proud of myself for isolating this problem and nailing it once and for all

Steve

[attachment deleted by admin]

Rik · May 08, 2008, 17:17:54

Well done, Steve.