ADSL Line noise problem

[#1]

A wild shot in the dark....

I monitor my line using Routerstats Lite and generally the line is rock solid for connection, and speed, after the settling in period with my new supplier...

however... there is something that is putting a "spike" on the noise /snr graph, that I cannot think where it is coming from. 

It is not 100% regular... i.e. the timing is not exact..but in the region of ever 20 mins, some extras thrown in for confusion sake as well.

here are some graphs...to show that it is not there all the time..

17 May 12:30

17 May 14:50

***********************************************************************************************

17 May 17:05

***********************************************************************************************17 May 21:00

***********************************************************************************************

18 MAY 13:30

***********************************************************************************************

18 May 17:40

Here is an enlarged version  saved in the vertical to read the timing easier...

We do not have anything, that I can think of that would be generating this in our house....

Our central heating is OFF all the time... ( although there is a wifi controller ) ..

we have battery operated smoke alarms in the hall and upstairs....

we have a CO2 monitor ( battery operated ) in the kitchen.

In all the above cases... if they were the culprit(s) then they would show the spike, on a 24 hour basis, which as can be seen from the first graphs is not the case...

Anyone got any ideas ?   

As you can see, from the speed tests, it does not affect the speed, and it does not affect the connection... it`s "just annoying " not knowing what it is, (could be ) and it not being continuous also adds to the "annoyance" ..

we are in a terraced house, in an urban situation, so it is not a farmers electric fence ! ! ..

[#2]

@shutter 

How about the fridge or freezer switching on? These take a big bite of current when they start.

[#3]

@Baldrick1 They do, and those plots suggest it is a switch being thrown somewhere. However, my router and fridge/freezer are on the same circuit, and I certainly don't get that effect.

[#4]

@jab1 

I agree that it shouldn't but I can think up a few reasons why it might be the cause. I also accept that there are many more possible causes.

[#5]

Okay, @Baldrick1 I agree it could be a cause, but a fridge/freezer on a 20 minute cycle - dodgy? Mine is on 50 minutes between cycles, even in the middle of summer.

Anyway, I'll wait for @shutter to come back before saying any more.

[#6]

@Baldrick1  & @jab1   Yeah...but..... that does not happen ALL day long... and there are long periods ( I`m talking hours and hours )... when the trace on the graph moves between 6.0 and 6.2 for the snr... so cant see it being the fridge or freezer...  and the snr does not spike when I boil a kettle, or put the washing machine on...

But thanks for the input... maybe just have to put up with it... as I said, the connection is rock solid ... and the speed is acceptable...

[#7]

Might be something external to you, @shutter . Have any neighbours suddenly acquired something they only use for restricted periods of time?

[#8]

@jab1  Hmm.... A new boyfriend !....but I don`t think that`s causing the problem ! ! ! ..

[#9]

Surely those graphs are a plot of noise margin - (not Signal/Noise ratio) - well it claims to be that.

So you would get a faster line at -3dB than -6dB, but you spend nearly all your time at the slower 6dB noise margin.

[#10]

@VileReynard Yes, it is noise margin, but there is something upsetting it on a regular (albeit sporadic) basis - which shouldn't happen.

[#11]

---

@shutter wrote:

Anyone got any ideas ?   

As you can see, from the speed tests, it does not affect the speed, and it does not affect the connection... it`s "just annoying " not knowing what it is, (could be ) and it not being continuous also adds to the "annoyance" ..

---

@shutter 

I don't have any ideas on what the 'culprit' causing the momentary downward spikes in your SNR margin could be, but although you don't think so, it probably is affecting your speed by causing you to have a slower line that what you have now..

As @VileReynard  has said, if you had a constant 3dB SNR margin rather than 6dB, your line would eventually re-sync to a higher speed.  It seems that those spikes are presently preventing that happening.

If you were on 3dB SNR margin the spikes would cause that to drop to 0dB or below and cause a re-sync at a lower speed, which IMHO is where you are now.

Good luck in tracking down the source!  

[#12]

---

@VileReynard wrote:

Surely those graphs are a plot of noise margin - (not Signal/Noise ratio) - well it claims to be that.

So you would get a faster line at -3dB than -6dB, but you spend nearly all your time at the slower 6dB noise margin.

---

In theory.... however... I am not interested in high speed, otherwise I would have spent more money on FIBRE and a more expensive router/modem.... also I am not in control of the setting of the dB margin/level.

Interestingly, though... IF.. I were to drop to 3db as a "level"... those spikes would probably cause instability on the line... Stability of connection, is more important to me than the "ego" busting line speeds advertised to those who "my line is faster than yours" playground egotists.

@jab1  @Baldrick1   @RobPN   After booting up this morning, and letting it run for a short time... I replaced the Power block on the router... just to see if there would be a different graph.... however, it did not seem to make any difference initially... 

Looking at the top graph, here..

( @VileReynard  take note )  a lower noise margin figure, did increase the "connection" speed, as expected...  only "marginally"  from about 7480 to 7846 ( according to the graph)...but on OOKLA speed test that figure is usually 1meg slower... 

The timing spaces on the top graph, have been changed to every 5 seconds, instead of the first part which was every 15 seconds... I did this to see if I could more accurately time the spikes.

the Lower graph, is on another version of Routerstats ( the full version)  to give a comparison of the shapes of the spikes and the timings.

It is interesting to note,  @VileReynard  ..whilst typing this.. ..  there has been a change in the noise margin level to average of 4.5dB ( putting it in the 3dB to 6dB bracket, instead of the previous higher 6dB to 9dB bracket)  and there was no change in the "connection speed"  and the "spikes" seem to have reversed direction ! ! ! ...

[#13]

I agree that it is noise margin but isn't SNR generally described this way, not in say percentage terms or X:Y as arguably ratios should be?

@VileReynard 

 I would think that if the DLM ihas set the speed to keep the average Noise Margin, if you insist, to 6 dB then any short term increase in background noise or reduction in signal level (it could be either) will result in possibly increased errors but there will be no increase in speed. On the contrary if the DLM thinks that the line should be set to 6dB it will slow the speed to bring it back to 6. Speed will only increase if the DLM decides that the line can support 3 dB continuously with an acceptable error rate.

@shutter 

Regarding the new boyfriend, are sufficient sparks flying?

[#14]

@Baldrick1   There is no QRM or QRN... the walls do not have ears ! ! ! ...

[#15]

Oi   !      ... who pinched my spikes  ? ? ?   

Just to show that the line is "normally" like the above graph(s)  with no spikes.... ! !  ...

Ah well...

Thanks for the interest guys....think this is gonna be one of those "mysteries of life"...

( the boyfriend is not "at home" next door .......  )