Hi all,

We have Draytek2820 router and were experiencing poor ADSL Performance on one particular line. This line barely manages to get above 8Mb/s, whereas our second and much better line is getting consistently 14Mb/s. Both are specc'ed at 20Mb/s. The problematic line is connected into the draytek.

On the problematic first line we've noticed that setting the modulation type has an effect on overall speed:

ADSL2 and G.DMT give the best results and gain about 8Mb/s though this is much less than the good line.
ADSL2+ gives the worst results with sometimes the connection barely managing to pull 1Mb/s

The second-good Line has a modulation type of G.DMT2+ Annex A.

I'm trying to understand whether there is a problem with the Signal to Noise Ratio and/or Attenuation so we can flag the issue with the people that installed our phone system/BT. With the logs on the Draytek I've found entries which are either :

Dec 6 13:00:13 my.router Vigor: ADSL_Status:[Mode=ADSL2(G.992.3) States=SHOWTIME UpSpeed=1099852 DownSpeed=9720370 SNR=6 Atten=27 ]

OR

Dec 7 13:34:08 my.router Vigor: ADSL_Status:[Mode=ADSL2(G.992.3) States=SHOWTIME UpSpeed=1112000 DownSpeed=6648972 SNR=16 Atten=27 ]

The web interface shows

ADSL Status Mode State Up Speed Down Speed SNR Margin Loop Att.
ADSL2(G.992.3) SHOWTIME 1112000 6648972 16 27

Can anyone confirm whether either of these figures are a low SNR and the cause of our problems ?

if you have BT provided line, you can run tests via speedtester.BT.com

A signal to noise ratio of 6dB is not good. 16dB is a lot better. I get about 11dB with a loop attenuation of 33dB which gives a consistent 7.6Mbpsdown and 512kbps up. If you're not with BT as the line provider then you need to find another checker tool.

Why is it using ADSL2 (G.992.3)? That's got a limit of about 12mbps I think - maybe BT are setting it on their end if you've got the router's modulation set to Auto. Try setting the modulation to ADSL2+ (G.992.5) specifically.

Those line stats are decent, try the 211801 or 2471201 modem codes to see if that helps since those can be an improvement from the UK modem code with some exchange hardware.
With a loop attenuation (line length) of 27, you should be seeing more like 15-18Mbps I think, but the SNR values show that the exchange is setting the best possible line speed for the current available signal (SNR), when it's at 6, thats generally the best speed you can get with the line conditions.
When it's at 16, that's probably because it's dropped and the exchange hardware has upped the SNR to help stability - the modem codes should help with this.

Check this page as well, the internal wiring can massively affect the speeds you get so you should try the router in the test / master socket to see if that ups the speed.
http://draytek.co.uk/support/kb_vigor_linefaults.html

Try setting the modulation to ADSL2+ (G.992.5) specifically.

We have tried this before and we get dismal performance. Something like 2Mb/s at best and usually under one. We will though at some point try and set back to ADSL2+ modulation and retest.

A signal to noise ratio of 6dB is not good. 16dB is a lot better. I get about 11dB with a loop attenuation of 33dB which gives a consistent 7.6Mbpsdown and 512kbps up. If you're not with BT as the line provider then you need to find another checker tool.

Thanks for confirming this. I didn't think 6Db was good. I guess its possible we have a very variable SNR ? If the SNR does change quickly does the router change its speeds automatically to cope with the noise?



A bit outside the Draytek product set, but I noticed at home on my Netgear I can get (least what I think is) an SNR per channel using the Telnet command. Does the 2820 have any similar ? My feeling is there is noise on certain frequencies which ADSL2+ is using and ADSL2(G.992.3) does not . Anyone know about the frequcnies/channels between the two modulations.


We are on BT and I'll use the linked BT check from now on, phoning BT, they have said numerous times the line speed is good and tested the line which they say is fine, but actual speeds are nothing like what they cite.