Introduction These notes describe how we can monitor the performance of a broadband connection to help trace the causes of poor performance. I have two programs, Routerstats and DMT tool, installed on a laptop which is left connected to the broadband router / modem. I then use www.logmein.com to look at and control the laptop from my home, as well as record pictures as shown below. This operation does not interfere in any way with other computers on the remote network. The first picture provided by Kitz and SamKnows on the next page provides details of the Cranleigh exchange. The web sites are:http://www.kitz.co.uk/adsl/adslchecker.php Sam Crawford has allowed Kitz to access his database but you can still go direct if you wish http://www.samknows.com/broadband/search.php
DMT Tools pictures This picture shows the DMT tools V7 results from a Thompson TG585V7 modem when the line was running as well as it ever does. Note that there are three tone gaps with the highest used tone being 95.
All the remaining pictures show DMT Tools V8 running with a Netgear DG 834V4 modem. The following picture shows a good performance with a steady SNR (Signal to Noise Ratio) monitor and up to tone 88 with two gaps providing a sync rate of 576 kbps and with a RCO (Relative Capacity Occupation) of 82 %.
The following picture shows the overnight overall SNR per tone accumulated fluctuation percentages during a noise storm and with the sync speed at only 352 kbps.
Here is a picture showing the start of a “smooth noise” picture with a Relative Capacity Occupation of only 64 % with 71 tones and which was interrupted to re-sync the modem.
Here is the picture immediately after the re-sync at a much faster speed and with tones up to 88 used but with three gaps and an RCO of 87%. Note that if this time is chosen carefully when the noise margin curve is reasonably smooth, a significant improvement in Sync speed is obtained which will last through medium noise storms before crashing into a re-sync.
Routerstats pictures The next picture shows the Signal Noise Margin curves at 15 second intervals. It shows three manual re-syncs with the last one when the curve is in a smooth state which achieves a much higher sync speed of 640 kbps. It follows that the actual noise curve will look very much worse with very wide noise frequency changes. The final curve shows the inevitable two automatic re-syncs with the latter dropping for more than 15 seconds.