Yes, I did find stations around that frequency. Do you think it could be the AM radio station(s) causing the drops?

Well that is a common cause of problems and is well documented and why ADSL behaves the way it does

Interesting. I must've thought wrong that DSL was designed to not be affected by radio stations, then. It just seems weird/unsophisticated that an AM station can sometimes cause DS SNRM drops/mass errors/occasional resyncs.

Edited by deleted (Sat 01-Dec-18 21:00:58)

The design is to not use the frequencies where the radio stations impact the ability of DSL to operate which often explains the dips rather than a smooth drop off due to the physics of attenuation

Right, but surely if it's service affecting (errors/resyncs (albeit not many of the latter)), then it's not working as it should?

Edited by deleted (Sun 09-Dec-18 14:59:17)

It is designed precisely to work like that.

If there is a factory chucking out electromagnetic noise, there is nothing xDSL can do about that factory.

That is why the SNRM exists. To allow for varying noise levels or even constant ones, to make the system work. If you didn't have a margin, but let the line connect at the full speed that you could get with the full Signal to Noise Ratio, then ever time you turned the central heating on there'd be a chance of a disconnection and re-sync.

When the sun goes down in the evening electromagnetic noise in the atmosphere goes up. In the morning the noise goes down as the sun rises. The principle is the same.

At some points in the day some noise hits your line for a while, then goes away. But the service continues happily providing you with the speed you are sync'ed at.

Edited by RobertoS (Sun 09-Dec-18 15:17:47)

That feel when you know who wrote a thread before actually looking at it

Well, I'm just interested. That's perfectly ok.

Thanks, Bob. I may as well just put up with it then.

'Fraid so. It is only if it seriously knocks your connection around over quite a while and takes you below the estimates for the line consistently that even a good ISP is going to be interested.

Getting the sort of speeds we do over copper lines is incredible, and the way the system works is essential. Otherwise, to lower error counts and things by the very nature of physics would mean slowing us all down to a tenth of what we get. The fundamental concept has to be for designers to say to themselves:- "this is what noise does to us - how can we work round it and still get high speeds?"

The data your graphs plot are the answer.