Draytek tells me that the router always attempts to SYNC at the highest speed - but it appears that the router does not take into account CRC error rate and SNR Margin. This usually means that not long afterwards, the line will be dropped and a new HANDSHAKE takes place until a lower speed is SYNCed resulting in better stability. It would be very useful to be able to set a maximum value for the SYNC speed and allow the user to maintain better stability. THis should not be all that difficult to code. JMN

For better stability use the Modem Code best suited to your line. This will take into account snr etc. If you provide these details for your line to Draytek they will advise the best Modem Code to use.

Thanks, nealuk, for the comment and advice. When I raised this issue with DrayTek, they did suggest in their reply that using a different modem code, namely, 3.3.7.2_211801_A might help. This is recommended for lower quality lines. I did try this but it made no difference. I monitor my router behaviour 24/7 by taking 5 minute snapshots. If you are interested, have a look at http://www.rmnjmn.demon.co.uk/DrayTek/Data_10Jul12. I think the plots are self-explanatory. Notice the poor stability for SYNC speeds around 7 Mbps compared with a lower speed. Notice when the line is dropped around Monday noon when SNR margin is comfortably at 10 dB that the router insists on SYNCing at a much higher rate with obviously poor stability. This is why I want to be able to set a max SYNC speed which, given my known line length to the exchange, I do not expect to see more than 6.5Mbps with good stability.

(By the way, I also suffer from another problem which DrayTek suggests using Annex M to fix: on most occasions, an incoming PSTN call - on my BT line - causes the line to drop out. Very annoying. I have not tried this modem code yet...one problem at a time.)

JMN

I would get the line fiexd first, as this is likely to be adversely effecting everything else. Is there a Microfilter problem causing the dropped calls?

I'd be tempted to get a new master socket and fit it yourself http://www.ebay.co.uk/itm/GENUINE-BT-2012-OpenReach-NTE5a-N-T-E-5A-Master-Socket-Linebox-Latest-IDC-/270984010087?pt=LH_DefaultDomain_3&hash=item3f17e7d967 which is a breach of telco rules, but then it saves arguments about the £140 call out charge.

And can you use a centralised filter? I think until you get the line resolved, looking at the other stats will be somewhat inconclusive and a waste of time.

Apologies for my frankness.

Best regards,


Neal.

Frankness is not a problem. "Fixing the line" is not a solution to DrayTek's strategy of trying to SYNC at the highest speed without due regard to SNR margin history. I have a Thomson Speedtouch router which is better in this respect than the DrayTek on the same line. I need the DrayTek router because it is far superior in other departments to the Speedtouch. By comparing two routers I have demonstrated that the problem is not a function of the line.

What I want is a return of the options provided by DrayTek in the 2800 series, namely, maxdnrate and codinggain both of which were withdrawn for the 2820 series. This gave me the ability to tune the router to my particular circumstances.

The line from my house to the nearest telephone pole is 210 feet in length. This forms a superb long-wire antenna with excellent sensitivity to pickup noise from unfiltered switched-mode power supplies in the neighbourhood, not to mention power-line networking and the rest. In this regard, "fixing the line" requires a much shorter run in the open - not something that can be changed. This would certainly reduce the noise problem considerably.

An NTE5 master plate would not solve the noise problem caused by the long line length. It would have a marginal effect by isolating house telephones from the ADSL line. In any case, disconnecting all the telephones does not have a noticeable effect on stability - the noise level remains the same, of course. The router is situated some distance from the incoming BT line so a new cable would have to be laid from the NTE5 plate. I agree that an NTE5 plate would most likely cure the PSTN interference. I haven't tested whether the Speedtouch is also susceptible to PSTN interference. But this is a minor problem at the moment.

maxdnrate + codinggain please, DrayTek?

John

PS: thanks again for your comments.

i strongly agree with jmn
A feature like a switch should be available for making easy to choice from line stability or higher synchronisation instead loading each firmware
Billion 7800 and fritzbox 7390 are so polular about this feature to adjust the line with the same firmware
I guess this should be bonus to draytek (for sales) however is up to them if they can understand it

To your problem for a more stable line ... i have done some tests with all firmwares here
http://www.forum.draytek.co.uk/viewtopic.php?f=12&p=73378#p73378

I know is from the 2850 instead of the 2820 but are the same firmwares codes because i am sure the 2850 is just the same with add on the vdsl
So after spending time for many tests i would suggest: (until and IF draytek add an adjustable feature)

Load again the 211801 as they suggested you or the 2121501
Them 2 have longer loop attenuator so a lower synchronizing
(HINT) load the one you want day time because night will have more synchronizing
i got 11-13 with them 2 day time and 14-15+ night time (tested i am not talking with out testing so is proven)
when done ... turn manually to G.DMT from the multimode option at WAN >> Internet Access
By doing this will resynchronize by 1-3 mb less

you should have a really reliable line now (i had 100%) for sure but is important to get one of the 2 as they advised you PLUS the G.DMT
Annex M will not synchronise for me either