Many thanks to RobertoS and the other contributors for very helpful comments which have enabled me to make progress during the weekend when I had time to reflect, experiment, observe and check my records.

Most unusually, we have had 48 hours with no rain, light or heavy. During this time we have kept a close watch on the broadband (at least during waking hours), and no disconnections have been observed. So, the main hypothesis now is "heavy rain means broadband disconnections, no rain means no disconnections." I need to check whether light rain means brief disconnections.

I must apologise for misleading you (and myself) by saying that "the connection also drops at about 8-30am and 5-30 pm every day." It doesn't, at least not every day. I think the explanation is that these are the times at which I often switch the main computer on (8-30am) and off (5-30pm) , so I would be downloading and would definitely notice disconnections. I haven't kept notes about all these disconnections, so I'm uncertain about whether there was rain during them. During the day the computers are mostly used for typing and reading, with only occasional downloads, so disconnections could be missed. Observation and recording will be better in future.

As for the telephone, I have used the quiet line test to establish that the line is quiet when the router lights are on, and that it stays quiet when the router is switched on and off. But I need to check whether there is noise on the line during a disconnection. I'm also beginning to wonder whether the phone really does stay connected when the router lights go out. It takes me about 30 seconds to get from the workroom to the router and the phone, and it occurs to me that I may, for example, have noticed a broadband disconnection on the computer, come downstairs to check the router, and picked up the phone after the lights had come back up. So, more and better observation needed here too. The router is in the living room, and it's easy to see the router lights from there and from the kitchen. I have noticed them flickering and going out, and in future will check the phone straightaway, as RobertoS recommends. I take his point about voice faults.

This note is already too long, so I won't give details of my checks for electrical interference (nothing found). Thanks again.

Have you got RouterStatsLite running yet? It is best to leave it running 24/7 in a situation like this, which means leaving the computer on for a few days. Whether or not you do that, put a shortcut to it in the Startup folder and also set it to start on loading.

The point of RSL is that as well as the general monitoring of the line, with the 5-second sampling interval I recommend disconnections are very obvious in both the graphs and the log file. Make sure you set it to save graphs on closing, and to log to file.

Here Roberto I have to confess that for five years or so I have been giving myself an easy life by using Linux (Ubuntu), and have more or less forgotten about "Windows." But I do take your point about RouterstatsLite and the advantage of the monitoring it enables. I still have a Windows licence and a machine that can run it. It's just that I know I won't be able to set it up quickly. I'll do it as soon as I can, but it may take some time. Thankyou very much for the reference and the detailed tips.

PS I hear on the Linux grapevine that using "Wine" can be tricky.

What router do you have?

It is pretty easy in GNU/Linux to poll a router and use gnuplot to create the readable data chart.

e.g. see my post here

http://forums.thinkbroadband.com/linux/t/4214138-tp-...

Let me know if you need a hand.

Nick