New ADSL - Modem Training

I've recently taken up adsl (adaptive rate) and understand that there will be a period of 'training' during which the line quality etc. will be monitored and a speed determined.
I'm quite a way from the exchange and the connection rates I am getting are better than expected (BT checker said 500 kbps).

Am I right in thinking that the modem in the exchange will go through a series of disconnections and reconnections to monitor synchronisation speed? The drop-outs appear to be early morning and later evening (8pm onwards).

For this training period, should I leave the modem/router on 24/7 or is it best to turn off when not required (and save the energy)?

Look here

http://www.dslzoneuk.net/adslmax_explained.php

To switch off or not is often asked but there is no one answer. If you are happy leaving it on, leave it on. Personally, I would switch it off overnight.

The dropouts at the times you mention are more probably due to the fact that noise and interference at ADSL frequencies usually starts increasing from when the sun goes down and decreasing again at the end of the night and your telephone line can act like a AM/MW aerial This is because the frequencies that affect ADSL (same as AM/MW radio frequencies) propagate (travel) much better when the sun is below the horizon. To this you must also add the fact that there is also likely to be more noise coming from other ADSL users at these busy use times. The longer your line the more these various noise sources are likely to affect you.

The exchange kit itself, apart from when you yourself manually disconnect, usually only disconnects if the noise is high enough to swamp the ADSL signal, i.e. Loss Of Signal. It will then reconnect at the best possible speed it can at that particular time. Switching your router on or off normally shouldn't make any difference, unless you do it by pulling the plug on it a lot in a short period of time, when the exchange kit could see this as an excessively noisy line and lower your BRAS profile to cope. If this happens, it can take three days for your BRAS profile to recover.

If you haven't done so already, have a look here for information on optimising your own internal wiring for ADSL;

http://yarwell.blogspot.com/2005/08/adsl-tweaking.html

You will see Yarwell's name appear often on this site as he is a font of useful information and is always helping people on here.

If you need help on getting the stats from your particular modem/router, have a look here to see if yours is covered;

http://www.kitz.co.uk/adsl/frogstats.htm

EDIT: Typo

Edited by deleted (Mon 23-Apr-07 02:24:54)

Many thanks for replies. I have avoided using the house internal phone wiring for the modem by using a Solwise SAR605EH homeplug through to PL-85PE adaptor. The modem is close (ish) to the BT face-plate (old style) with the micro-filter directly in the face-plate.

I've used about 5 mtr of Maplin's 4 strand flat data cable with RJ11 plugs crimped on. Obviously this is not twisted pair and so is unbalanced. I routed the cable away from mains wires etc. and is at ground level with no significant vertical rise.

My modem stats are at 9:00 am:
-----------------------------------
Modem Status
Connection Status Connected
Us Rate (Kbps) 448
Ds Rate (Kbps) 5600
US Margin 24
DS Margin 10
Trained Modulation ADSL_G.dmt
LOS Errors 0
DS Line Attenuation 42
US Line Attenuation 23
Peak Cell Rate 1056 cells per sec
CRC Rx Fast 0
CRC Tx Fast 0
CRC Rx Interleaved 7
CRC Tx Interleaved 2
Path Mode Interleaved


DSL Statistics
Near End F4 Loop Back Count 0
Near End F5 Loop Back Count 0
-----------------------------------------

I'd welcome observations - including thoughts on using shielded or unshielded twisted pairs (CAT5E?) instead of the flat cable. I don't think the internal wiring run is long enough to introduce significant capacitance but I haven't researched the cable specs. I may experiment, especially in the evening, with a mains extension lead and put the modem right next to the face-plate with a very short cable and monitor the noise margin. This could identify any internal factors which I can influence.

Twisted pair is always better than flat in such usage, but whether it makes any real difference in your case only a test will show for sure. Much the same applies to using shielded over unshielded, though I have yet to find any significant advantage in using shielded, even in particularly electrically noisy homes. For usually there is no easy way to earth the shielding, unless you rig it up with metal earthed sockets. Though others on here have reported a benefit to using shielded. So if you have any lying around and your intended test of a shorter run shows a real improvement, no harm in trying shielded, otherwise I would stick with standard CAT5.