Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[rjpreston]

Hi,

I swapped out a HG612 for a VMG1312-B10A modem a week ago and got an immediate 30+% speed increase from 27 to 36Mb. At the same time the SNR dropped from 6.8 to 4.2 downstream and from 6.2 to 2.1 upstream. This was much more of an improvement than I'd hoped for and well worth the 40 notes dropped on ebay for the Zyxel device! Well chuffed.

One thing puzzles me. Dslstats shows a downstream SNR drop of nearly 1dB on a clockwork basis between 8 and 10 pm on most (but not all) nights both before and after a changed from the HG612. possibly happened a bit later last night - looks more like it started midnight-ish. What might be the cause of this? I can't think of anything within our house that is used at those times (and for ~2 hours duration) that might be the reason. Could be something in a neighbour's house, I guess, but seems weird that its quite regular and quite well defined in its duration. can anyone suggest what might cause this sort of snr drop?  Graphs of the last week or so attached.

Cheers 

[j0hn]

I can't read those graphs.
It looks like it's showing 12 days worth on a single graph, is that right?

Can you scale the graph so it shows the increase/decrease in SNRM.
Something between 6-12 hours over the entire graph is easier to see the detail.

edit: to add, part of the increase you have seen appears to be a lower SNRM target.
Is it possible the HG612 had an older firmware without G.INP support?

[rjpreston]

Sorry j0hn, my fault - I'd scaled the graph so that it showed about 12 day's worth to show the repeating SNR drop at the same time each day.

Here's the snapshot of the day that I changed the modem from the HG612 to the Zyxel. I understand the speed increase was due to the lower SNR.


On this new graph an example of the nightly SNR drop is at about 9pm to 9.45

[j0hn]

Looking at both graphs that screams out electrical appliance of some sort.

Items on a fixed timer can often be seen to turn on/off at exact times each day.
The timing on your graphs is a little more variable, but still happening at roughly the same times each day and lasting for a similar duration each time.

Tracking these things down can be extremely difficult.
It can be anything near your line between you and the cabinet.

Water pumps, sewage pumps, treadmills, industrial lighting, generators, etc etc.
The list of possibilities is huge.

Fortunately it only causes the SNRM to drop a small amount.
I see others with similar but the SNRM drop is enough to cause their line to resync each time the nightly "event" occurs.

You could try resetting bitswaps per tone just before it starts and see if it effects any particular tones.

[rjpreston]

Thanks for the tips j0hn. After doing some research as you suggested, the frequencies that were causing the most bitswaps are around 7.2MHz. Had a look at Ofcom's frequency allocation and eventually came to the conclusion that the source might well be shortwave radio broadcasts that come from Woofferton- about 20 miles West from where I live (and we're on the West side of a hill so facing directly towards the transmitter.) The drop in SNR I can see at specific times could well correspond with the transmitters broadcasting the 1 or 2 hour programmes on that specific frequency. They're worldwide broadcasts so the transmitters must be pretty powerful.

Not a lot I can do, but the modem is remaining in sync so doesn't do any harm really.