Speed Keeps Dropping - Massively

[#1]

[#2]

[#3]

[#4]

@NicDB 

Re: "Excessive" data usage.

I notice that you have a couple of "Smart" TV devices: a Firestick and a TV.

Are you certain that when these are not actually being viewed, they are not still streaming in the background?

Could you de-power them last thing at night, (ie not just put them in to standby), then note the data usage recorded on your hub.

Re-check the data usage on the hub in the morning to see if it has increased significantly, then turn the Smart devices back on.

We have NowTV on a SmartTV and a Firestick. My family, when they have finished watching it, have a habit of just switching off the TV, (to standby), without first exiting the app and letting it close properly.

[#5]

@NicDB 

This is VERY helpful and tells us a great deal ... but probably not great news for you!

You only need to post the SNRM (Noise Margin) plots please.  Here is a summary of what the ones you have posted show...

• Sun 22:05 to Mon 19:43
  appears to be "rock steady" but there is insufficient granularity to be of any help
• Mon 20:43 to 22:40
  distinct noise starts at 22:17
• Mon 21:45 to 23:42 (odd overlap!)
  noise at 23:27 onwards is clearly cyclic suggesting something running a service cycle of some description
• Mon 23:46 to Tue 01:42
  really grim noise from 00:00 onwards
• Tue 01:46 to 03:42
  noise ceases abruptly and completely at 02:21
• Tue 05:46 to 07:44
  no noise at all

Would I be correct in inferring that the missing graphs between 03:42 and 05:46 are all "flat lines"?

If that is correct, then I would conclude that you do have REIN arising from the operation of some electrical equipment between 22:17 and 02:21 - close enough to a 4 hour duty cycle.  That equipment could equally be within your home or outside of your home (think street lighting).

I doubt that the BT engineer will find anything specific as sorting these issues often requires a REIN trained engineer and you will not get one of those until TWO ordinary engineers have not found a fault.  That said a bad joint can cause an AC line balance issue which will make a circuit more susceptible to REIN - see AC Balance issue... (kitz.co.uk)

Reducing susceptibility is not an easy task, it first needs an engineer to consider the matter and then (if there is an imbalance) be able to find and rectify the cause.  A far more productive approach would be to identify and eradicate the source of REIN.  See ::. Kitz - REIN, SHINE & RFI .::

Your first approach to this should be to keep running router stats 24x7 this is your most valuable insight to YOUR issue.  Note the use of the telephone handset when noise is observed will be helpful.  Notice when the noise starts and then try to correlate that with activity within the household.  If you can identify something which has "switched" on, then switch it off and see if the noise disappears, either suddenly or gradually.  A portable AM radio tuned to 612kHz will be useful.

The source of the noise could be something which you perceive as always on, but which at times comes under a stress load.  This might be relevant here - my "baptism by fire" with this stuff was eventually traced to the CCTV camera power pack. Current switch mode power supplies are notorious for going pear shaped and throwing out lots of RFI.  What was happening is that as dusk fell, the IR illuminators kicked in thereby drawing far more current putting the PSU under extreme load, giving rise to RFI.  Switch it off and the SNRM flat lined.  Turn it on after it had cooled down and then over 10 or so minutes, then SNRM plummeted.

I note that you have smart cameras in your network, so this is worth a look.

Do you have PIR switched outside lights?

Do you have a HAM radio operator in the area - the lime window is typical of those seeking long distance communication.  I cannot recall - does that version of router stats plot the tone bins?  See https://kitz.co.uk/adsl/rein.htm#HAM

In case you are wondering what is the relevance of this is, it might help to revisit the earlier explanation of how re-sync works.  If there is a noise event sufficient to drop the sync (SNRM hits zero), the line will re-sync at a lower speed to maintain link stability.  The link will only re-sync again if the link drops.  Based on the evidence you have provided, if you find that the link is slow in a morning, I would expect that a router restart will deliver the normal speed stably throughout the day ... until the RFI returns from wherever.  However as you have noted repeated restarts of the router are not ideal ... the DLM will then start to be really aggressive and might chronically band the line.

This might be a long journey, but it might find a suspect power supply which is working outside of its design envelope and in the extreme could possibly be a fire risk.  The search would be worth it for full peace of mind.  See here - https://modwiggler.com/forum/viewtopic.php?p=1047994&sid=8d2cb4d0db5791c06297bc26b03e048c#p1047994 - notably two posts further on.

One extreme approach for avoiding the impact of REIN if it cannot be found and eliminated (particularly if it is outside of your property) would be to upgrade to FTTP if that is available.  RFI does not impact the operation of FTTP services.

[#6]

Thanks for all your help everyone.

The OR engineer came out and could find no fault during the five minute signal tests etc. A good steady 55+Mbps down and 16Mbps up.  As he found no fault he has closed the ticket.  He did say that if the fault comes back then to call him. I have his mobile number so he can come round to check the fault in real time.

He did suggest getting PN to send out a new router to eliminate a router issue.

Can't think of anything that comes on at night to cause interference. I'm wondering if it is the dishwasher. I put it on around 10pm or just after. It runs for about 1.5 hours on Eco mode then goes into a drying mode on low power. Maybe that is a cyclic noise I'm seeing?

I'll keep the Routerstat app running for a few days. I'll make a note of dishwasher on times. I'll have the corded phone in a day or so.

I bet the speeds are down again by Monday.

I'll check for rogue devices using up my bandwidth too.  Shame there is not an app that detects where the traffic is going to on our network.

[#7]

@NicDB 

So the engineer findings were as anticipated.  Did you see my post soon enough to ask about the AC balance?  See BT line AC Balance? (google.com) I suggest that getting this examined should be considered a priority - if the balance is off, fixing it might be enough to reject the RFI adequate to sustain a stable high speed link.

Now that we have a time frame to inspect, you might try turning on your dishwasher at a different time and see if the interference window moves with it.  Simply observing the SNRM plot before and after turning things on (and off at the plug) might help show what is happening..  Does the dish washer really run for 4 hours?

I doubt that a different router will be of any value here.

Remember that if the speeds are observed to be down, restarting the router when the SNRM is clean will restore the line speed.  What we really want to see is the SNRM plot at the time that the router disconnects and then reconnects at a vastly lower speed.  The "crash" before the resync will point to the issue you most need to identify and address.

There was a lot in my previous post - have you thought about PIR switches / security cameras?

To be honest, now that we have a good picture of what we are looking at, a cheap AM radio would be more appropriate than the phone handset ... though phone handset will indicate if there is audible noise and if there are inbound calls causing issues.

Personally I would leave RS running in the background until you find the culprit - in my experience unless there is something obvious, it might take weeks.  This is an environmental issue, which is not going to go away by changing ISPs unless you have the option of choosing one which does not use the BT (or similar) copper pair circuits.

[#8]

So after five days of monitoring this issue it all seems stable and fast at around 55Mbps/16Mbps (down/up).

Monitoring the data usage after turning off some devices and got these figures after a restart:

Time DSL Up:     Sent:         Received:

11hrs                    242MB        210MB

15hrs                    512MB        1.5GB

2Days                   1.6GB         17.4GB

2.5Days                 2.4GB        23.6GB

Does this seem reasonable?

Here are the SNRM plots.

[#9]

[#10]

[#11]

@NicDB I'll leave it to @Townman to fully comment, but what did you turn off? To my untrained eye, those SNR plots look reasonable - apart from the initial one - before you turned 'something' off?

[#12]

The scale kinda changes somehow?

[#13]

[#14]

There's still something going off - it appears to have lost some stability after the 01.23 drop on the 7th - but as I said, I'll leave further comment to the expert.

[#15]

@NicDB 

Great monitoring!!  Something is being elusive here!

As discussed before Tuesday 6th 00:00-02:21 was a tad grim and was "Of the shape" of not good news.  That said it has been fairly clean since then.  Have you run the dish washer again through the period you have been monitoring?

In many REIN situations there are more than one source of noise and there is some evidence of this here.

Tuesday 03:15 onwards there is something switching OFF for a few seconds at a 25 minute interval.  From 03:52 the frequency doubles (12 minutes) - any idea what that might be?

Tuesday 10am - long drop - was the the BTOR visit?

Wednesday 01:25 drop and a loss of speed 62.6mbps down to 57.3mbps - no obvious build up - the link just vanished.  Any thoughts what might have happened at that time?

Wednesday 07:02 drop no speed change - ditto?

Wednesday 23:26 looks like something switched off - ditto?

Thursday 05:30- ditto - ditto?

Is there anything you are aware of that you have not done which you do normally?

The graphing suggests that what blighted your line on Monday evening and reduced the performance to 17mbps has not reoccurred.  Given that the engineer has again found no issue with your line, I suggest that there are fair grounds to conclude that this is REIN (electrical interference) from a source as yet not identified.

I would recommend continuing to run the monitoring.  The fact that whilst there is nothing going wrong, there is no point running the monitoring would not be a good approach.  Keeping the monitoring active will ensure that should something happen, your will have captured the profile of the event.

You might be able to improve the graphs by modifying the axis settings - I think you can do that on the BTHH5a version (you sure can on others).  Set the axis to minimum zero and max 12.

Keep monitoring, if not using the computer leave the graph display on ... should you see the graph "move" have a think / look around to see if there is anything thing you can correlate to the event.

Telephone handset to listen to the line from time to time will be really useful, as will an AM radio tuned to 612kHz.

Data usage - yes I would have said that is excessive - however, I have just checked mine following a restart 3 days ago.  I have used 19.6GB in 3 days 21 hours ... and I do not have a clue what has used it!

In that time, I have not streamed Netflix, catch up TV - purely email on multiple devices (two laptops, two tablets and mobile phone).  Dropbox and OneDrive across the lot.  mobile and one tablet also connected to iCloud.

Then there is the Sam Knows internet performance survey box - who knows what that is up to?  Similarly there are other unknown quantities - 4x smart switches, Smart TV, Cloud connected NAS and Alexia (on Amazon Fire) downloading unrequested advertising, with the requested music interspersed.

So may be your usage is not as horrid as it appeared to be.  Would now like to understand what is burning my broadband.