SNR and weather

[Roderick Stewart]

Winter must be approaching. Downstream sig/noise ratio is back down to
5dB today. It was 7dB during the hottest days of the summer, and has
been steady at 6dB for several weeks, but now the weather is cool and
cloudy it's back to 5dB. It'll be the same next year. I have FTTC with
the final link being an overhead cable from a pole across the road. I
never notice any change whatsoever in the performance, the sync speed,
or any speed tests, but the numbers (from the status page of a
Draytek 2762) really do seem to depend on the weather.

I'm not worried because it doesn't really affect me, but I just think
it's interesting. It suggests there could be some setups where the
fine weather performance is "borderline" so that changes in the
weather could make a difference. Something to bear in mind.

Rod.

[notya...@gmail.com]

The usual cause is water in joints, underground or in connections on poles, so performance reduces after [heavy] rain.

[Roderick Stewart]

Whatever it is, in my case the response time is quite long, i.e. it
doesn't change suddenly with every change in the weather, just slowly
over the seasons. Maybe a general increase in humidity somewhere,
rather than an accumulation of actual water after the rain? Whatever
it is, it doesn't seem to affect performance. Now that it's down to
5dB I expect it will stay the same until next spring.

Rod.

[Graham J]

Roderick Stewart wrote:

[snip]

>
> Whatever it is, in my case the response time is quite long, i.e. it
> doesn't change suddenly with every change in the weather, just slowly
> over the seasons. Maybe a general increase in humidity somewhere,
> rather than an accumulation of actual water after the rain? Whatever
> it is, it doesn't seem to affect performance. Now that it's down to
> 5dB I expect it will stay the same until next spring.

Possibly RF interference in the spectrum (0 to about 30MHz) used by VDSL.

The interference comes generally from European radio stations. The
signals from these stations are reflected by ionised layers in the
atmosphere, and the height of these layers varies with temperature -
hence the slow changes over the year. All these signals combine to
increase the general noise level across the spectrum, so the router
adapts; but the SNR margin it reports reduces. If there's yet more
interference the router stops using some of the frequencies (bins or
tones in some explanations) so the sync rate reduces.

"Sporadic E" affects frequencies of about 100 to 300 MHz and is
exploited by radio hams to communicate over distances much greater than
line of sight. It also affects Band 2 VHF radio, which is why sometimes
stations from Italy or the south of France are clearly audible in the UK
if you tune around.

--
Graham J

[Brian Gregory]

That figure is probably the signal to noise margin rather than the
actual ratio at some particular frequency.

https://en.wikipedia.org/wiki/Noise_margin

If the connection speeds don't change at all then the noise margin will
reflect changing levels of noise. But if the modem changes to a lower
speed the margin will increase even without there being any change in
noise levels.

If your connection speed is limited by line length your modem will
usually connect at a speed that gets some target noise margin which it
thinks will get the best speed without too many errors.

I'm on the cheapest FTTC package my ISP offers (40M down / 10M up) and I
can always connect at that speed. I often see the noise margin change
from day to night. It seems worse during the day than at night. Also can
be worse after really exceptionally heavy rain. Because I'm always
connected at 40down/10up the noise margin will be reflecting actual
differences in noise levels.

Back when I was on ADSL noise levels were worse at night, so much so
that if I switched the modem off and on again during the day it would
often connect at a significantly faster speed that it had been at
overnight, but with a lower, but perfectly adequate, noise margin.
However later on it when it began to get dark it would begin to struggle
and eventually drop down to a lower speed.

--
Brian Gregory (in England).