Internet drops - due to interference?

[#1]

The above happens pretty much every day - at least once. Bit of a lengthy story behind this - as it's been happening since we moved to a new property in April, but here's where we are up to.

When I first noticed the drops, I contacted customer services - we've had an Openreach engineer out twice - first time around they noticed a lot of line errors, and the incoming line from the street to the outside box was re-wired. No change in drop frequency. Next time around, still a lot of errors - found that ivy was crushing the cable from the outside box to the master socket - so this was replaced and we had a new master socket fitted. Again, no real change in drop frequency. I was advised by the second engineer that REIN or SHINE could be causing a problem.

I let things be for a few months and have just got back round to trying to locate the culprit causing disconnections. At present I am using an Asus RT-N66U router with the Asuswrt-Merlin firmware (latest stable version) and an Openreach modem. There is one cordless telephone connected to the master socket phone port. No other telephone sockets exist - there is only the bare end of a second telephone cable that runs from the outdoor box to the office upstairs - but the socket has been cut off.

I have reproduced the problem with the Plusnet Hub using its built in modem also - so I'm not blaming the Openreach modem or Asus router.

I use a powerline network (using TP-Link adapters, all the same model) to connect my main PC upstairs to the router - but still get the disconnection issue when all my powerline adapters are removed from the sockets. So I don't think Powerline is to blame.

This setup is largely identical to how it was in the previous property - save a few extra devices and powerline adapters. Only major electrical wiring difference is that there are two ring mains from the consumer unit, rather than the one I had at the previous property.

When the connection is up, we get speeds up near what we would expect to get - apart from maybe if we've had a spate of sequential disconnects that I presume have triggered something on the Plusnet line monitoring system leading to a rate limit (probably happens like that every 2-3 weeks or so).

The only pattern I have picked up so far is the following;

When I turn on the upstairs PC for the first time each day (be it morning/afternoon/evening), the internet is pretty much guaranteed to drop, with the log on the router around the time showing this:

Jan 16 20:32:41 dnsmasq-dhcp[286]: DHCPOFFER(br0) <UPSTAIRS PC>
Jan 16 20:32:41 dnsmasq-dhcp[286]: DHCPREQUEST(br0) <UPSTAIRS PC>
Jan 16 20:32:41 dnsmasq-dhcp[286]: DHCPACK(br0) <UPSTAIRS PC>
Jan 16 20:33:15 pppd[369]: No response to 10 echo-requests
Jan 16 20:33:15 pppd[369]: Serial link appears to be disconnected.
Jan 16 20:33:15 pppd[369]: Connect time 1363.7 minutes.
Jan 16 20:33:15 pppd[369]: Sent 211266792 bytes, received 1344479117 bytes.
Jan 16 20:33:15 miniupnpd[4968]: Failed to get IP for interface ppp0
Jan 16 20:33:15 miniupnpd[4968]: SendNATPMPPublicAddressChangeNotification: cannot get public IP address, stopping
Jan 16 20:33:15 dnsmasq[286]: read /etc/hosts - 5 addresses
Jan 16 20:33:15 dnsmasq[286]: read /etc/hosts.dnsmasq - 4 addresses
Jan 16 20:33:15 dnsmasq-dhcp[286]: read /etc/ethers - 4 addresses
Jan 16 20:33:15 dnsmasq[286]: using nameserver 208.67.222.222#53
Jan 16 20:33:15 dnsmasq[286]: using nameserver 208.67.220.220#53
Jan 16 20:33:18 WAN_Connection: Fail to connect with some issues.
Jan 16 20:33:18 stop_nat_rules: apply the redirect_rules!
Jan 16 20:33:21 pppd[369]: Connection terminated.
Jan 16 20:33:21 pppd[369]: Sent PADT
Jan 16 20:33:21 pppd[369]: Modem hangup
Jan 16 20:33:30 rc_service: httpd 288:notify_rc start_autodet
Jan 16 20:33:30 kernel: autodet uses obsolete (PF_INET,SOCK_PACKET)
Jan 16 20:33:32 rc_service: httpd 288:notify_rc start_autodet
Jan 16 20:33:45 miniupnpd[4968]: upnp_event_process_notify: connect(192.168.1.78:2869): No route to host
Jan 16 20:34:06 pppd[369]: Timeout waiting for PADO packets
Jan 16 20:34:11 pppd[369]: PPP session is 6762 (0x1a6a)
Jan 16 20:34:11 pppd[369]: Connected to 10:e8:78:a8:0d:e4 via interface eth0
Jan 16 20:34:11 pppd[369]: Using interface ppp0
Jan 16 20:34:11 pppd[369]: Connect: ppp0 <--> eth0
Jan 16 20:34:11 pppd[369]: CHAP authentication succeeded: CHAP authentication success
Jan 16 20:34:11 pppd[369]: CHAP authentication succeeded
Jan 16 20:34:11 pppd[369]: peer from calling number 10:E8:78:A8:0D:E4 authorized
Jan 16 20:34:11 pppd[369]: local IP address 143.159.38.111
Jan 16 20:34:11 pppd[369]: remote IP address 172.16.18.24
Jan 16 20:34:11 miniupnpd[4968]: ioctl(s, SIOCGIFADDR, ...): Cannot assign requested address
Jan 16 20:34:11 miniupnpd[4968]: Failed to get IP for interface ppp0
Jan 16 20:34:11 miniupnpd[4968]: SendNATPMPPublicAddressChangeNotification: cannot get public IP address, stopping
Jan 16 20:34:11 miniupnpd[4968]: ioctl(s, SIOCGIFADDR, ...): Cannot assign requested address
Jan 16 20:34:11 miniupnpd[4968]: Failed to get IP for interface ppp0
Jan 16 20:34:11 miniupnpd[4968]: SendNATPMPPublicAddressChangeNotification: cannot get public IP address, stopping
Jan 16 20:34:11 miniupnpd[4968]: ioctl(s, SIOCGIFADDR, ...): Cannot assign requested address
Jan 16 20:34:11 miniupnpd[4968]: Failed to get IP for interface ppp0
Jan 16 20:34:11 miniupnpd[4968]: SendNATPMPPublicAddressChangeNotification: cannot get public IP address, stopping
Jan 16 20:34:11 miniupnpd[4968]: ioctl(s, SIOCGIFADDR, ...): Cannot assign requested address
Jan 16 20:34:11 miniupnpd[4968]: Failed to get IP for interface ppp0
Jan 16 20:34:11 miniupnpd[4968]: SendNATPMPPublicAddressChangeNotification: cannot get public IP address, stopping
Jan 16 20:34:11 rc_service: ip-up 24834:notify_rc start_firewall
Jan 16 20:34:11 miniupnpd[4968]: shutting down MiniUPnPd
Jan 16 20:34:11 start_nat_rules: apply the nat_rules(/tmp/nat_rules_ppp0_eth0)!
Jan 16 20:34:11 dnsmasq[286]: read /etc/hosts - 5 addresses
Jan 16 20:34:11 dnsmasq[286]: read /etc/hosts.dnsmasq - 4 addresses
Jan 16 20:34:11 dnsmasq-dhcp[286]: read /etc/ethers - 4 addresses
Jan 16 20:34:11 dnsmasq[286]: using nameserver 208.67.222.222#53
Jan 16 20:34:11 dnsmasq[286]: using nameserver 208.67.220.220#53
Jan 16 20:34:12 miniupnpd[24870]: HTTP listening on port 46700
Jan 16 20:34:12 miniupnpd[24870]: Listening for NAT-PMP/PCP traffic on port 5351
Jan 16 20:34:12 rc_service: ip-up 24834:notify_rc stop_upnp
Jan 16 20:34:12 rc_service: ip-up 24834:notify_rc start_upnp
Jan 16 20:34:12 rc_service: waitting "stop_upnp" via ip-up ...
Jan 16 20:34:12 miniupnpd[24870]: shutting down MiniUPnPd
Jan 16 20:34:13 WAN_Connection: WAN was restored.

The PC is plugged into a 6-gang multi adapter along with the monitor, a wireless charging pad, an amplifier, and the power supply for an LED strip light.

My current wonder is whether the PC is setting off some interference in the electrical wiring that bothers the modem and leads to a disconnection - or potentially another device.

Problem is, we sporadically get disconnects unrelated to the PC being turned on - so potentially it's a red herring - but it certainly occurs more than could be thought of as coincidence.

My question is, where to go from here - I'd looked into interrogating and monitoring the modem signals - but as far as I can ascertain the Openreach modem is now locked down to external management.

Would it be worth getting a surge protected multi-adapter for the PC in question - I realise it would have to be protecting against surges in the other direction from its design, but still...

Any help would be gratefully received!

[#2]

Am I reading this correctly, are you saying that you have an external box with two cables coming off, one going to the master socket and another one with the end cut off? If so the second unused cable needs to be removed from the external box. As this box is upstream of the master socket this should officially be done by Openreach.

Powerline devices are renowned for giving problems, especially if the data is having to be transmitted between two ring mains. Performance is also degraded if they are plugged into the end of extension leads.

[#3]

Yeah that's exactly the set up.

I'll get that cable sorted and see what happens after that.

I have to say that my powerline performance has always been very good - throughput is good - main issue is the sporadic dropouts - which are still happening at the same frequency without any powerline devices.

[#4]

hi. 

Id also investigate fitting a surge protector / mains filter to the bt modem and router supply . Which bt modem is it. ECI or Huawei ?

RF interference could come from any source . A badly wired fridge or freezer is a likely candidate if within range. Also low energy bulbs , especially the compact fluorescent types and or their power supplies if remote. In fact any power supply including that of your router or modem. 

[#5]

Welcome to the community forums @alistairmillar

We are seeing a fair few disconnections over the past 30 days.

Our tests are also showing your router in sync at 22.3mbps below your speed estimates, likely caused by the drops.

Once you've done as much as you can to rule out your equipment causing the interference and if the problem still exists, I'd head on over to http://faults.plus.net to report a fault to us so we can further investigate.

[#6]

The modem and router are both on a surge protected supply already.

It's the BT ECI modem - I hear the Huawei one has been locked down to being monitored for linestats (which I was going to collect to see if there's a SNR issue playing into things) but the ECI one can be potentially monitored via a serial interface if one is willing to break out the soldering iron - is this the case still? If so, are there any tutorials that could be used?

We had a weekend without disconnections though - wasn't using the upstairs PC. Lo and behold, I turn it on and as soon as I do, we disconnect! I'm coming back again and again to that being the issue. Is there some form of power spike being generated in the cable as the PC or associated peripherals switch on - and could that be sorted with a surge protector.

Today's job is to investigate the BT second wire and remove it - just so I can totally rule it out. The bare end does terminate in the same room as the PC that (I think) is causing the problem - could this be linked?

[#7]

OK, an update:

I had a look in the pre-master socket junction box, and it looks like the wire to the room upstairs has been disconnected a long time ago and just left embedded in the wall. The only incoming cable joins to thenew cable to the master socket in the living room. I did notice a nest of cut wires as seen below (forgive the side-ways pic).Essentially only the white and orange wires connect to anything. The rest all dead end and were tucked in the box.

I'm assuming that there's not a problem here given that two Openreach engineers have seen it recently...

If it is, that rules out the spare wire theory.

Looking at the connection log, those disconnect times do correlate with when the upstairs PC turns on (the 5 day up-time from 29th December coincides with when we were away on holiday and the shorter uptimes over a couple of days relate to when I was on nights).

There are sporadic drops unrelated to the PC coming on - but the vast majority are linked.

I've had a listen around the relevant plugs with an AM radio on 612khz - and when the PC comes on there's a massive amount of funky noise from the power supply - but then it's a massive switch-mode power supply so I kind of expected that.

Would changing the PSU be a next step? The PSU in question is a Coolermaster Elite Power model and hasn't caused this problem in a previous property. Other things I can think of include the socket - which I'm swapping today just in case (although it was swapped out a few months back to a 2-gang one).

Moderator's note by Mike (Mav): Post released from Spam Filter.

[#8]

Cheers for the update. I’ll admit it’s a bit of an odd issue, but it does look like to be related to your PC or its PSU.

If you can test with a different power supply it’ll probably further narrow down the cause of the problem.

Keep us posted, I’ll be interested to know how this is resolved.

[#9]

Hi @alistairmillar

it does sound as if something's not right with your PC or it's power supply. 

Presumably your PC has a network card for the Ethernet connection. Ensure it is seated correctly on the PC motherboard . With the mains power cable removed, unseat the network card and reset on the motherboard.  

Check for any stray cabling in the PC , 

is the actual Ethernet network cable in good condition? Is it a standard Ethernet cable with factory terminated plugs at each end ? 

I would also look at the PC earthing situation. Is the supply cable correctly earthed and terminated in the mains plug? And have you earth, live and neutral continuity from your mains circuit distribution board to the mains socket your PC is plugged into ? 

[#10]

When you say it drops when you switch on the pc do you mean this or is it when you switch on the multi-adaptor? If it's the latter then unplug everything except the pc then try switching on the pc. If this does not cause a trip then try reconnecting each bit and trying again in turn. I would be suspicious of the led power supply.  If it really does trip only when the pc is turned on then remove the Ethernet cable and see if that fixes it. This should at least confirm that this is the source of the issue. Also try using a different Ethernet port on the router to see if you have a faulty LAN port.

[#11]

Thanks guys.

I managed to replicate the problem with the ethernet cable unplugged and only wireless connection to the router. The ethernet card is integrated on the motherboard also. So it rules out the ethernet cable.

Yesterday I got a new, surge protected multiway adapter and swapped the socket itself out for a different one. All the earth connections are sound.

Good news is that the internet remained connected when I switched the PC on this morning and last night.

Bad news is that we had a disconnect at 17.30 yesterday - while we were eating tea! I guess that could have been a random coincidence so I'll monitor things for the next few days.

[#12]

Ok so if  you have the option to use Wireless or ethernet on the PC , either one should be switched off when not in use.

I still think it could be the Switch mode power supply. i would be concerend about it makign a lot of RF noise, or some kind of surge curent as they are not meant to in my experience .

What happens if you just use the Sleep or hibernate function on the PC, rather than shut down? 

Another option - if you continue to use ethenet connection, might be to put a small ethernet switch in the  ethernet connection between the PC and the router to act as a protector.

[#13]

Ethernet has been unplugged for the last week or so as I got rid of the powerline adapters that it was using to connect to downstairs. It has always gone to a switch that sits downstairs - never connected straight to the router.

Was still getting the problem up until yesterday when the wall socket was changed out (also cleaned up and stripped back some of the mains wire to the socket to get some fresh copper) along with the old multiway adapter to a new surge-protected one.

Whenever I was connected over ethernet I had the wireless switched off - and vice versa.

Internet usually doesn't disconnect if the PC is restarted or shut down and then powered up with a small time delay - it's usually the first time its switched on after a fair few hours off. I guess this lends weight to the PSU issue - maybe its interference off some capacitors getting charged after a long time off as opposed to a brief gap where they still had residual charge

I'll see if the lack of disconnects continues now the multiway and sockets are swapped out, next step if this carries on is to swap out the PSU. It's due an upgrade anyway...!

[#14]

If your router and PC are on the same mains circuit could it be a poor connection upstream of your router such that when you turn the PC on there is a dip in the mains supply to the router caused by the switch on surge current of the PC PSU.

[#15]

The connection drops seem to be happening far less frequently now that the PC upstairs is on a surge protected bar. Powerline is definitely implicated too as as soon as I plug the adapters back in, the disconnects recur.

There's still the odd disconnection - but frequency is manageable and usually it's overnight.

I'm putting it down to the house wiring and interference in the wiring causing noise in the phone line.

The logistics of re-siting sockets and master sockets is time and expense I'm not prepared to pay - as we've just finished renovating the living room where the master socket is and there's no guarantee it'll rectify things.

Are there any other methods of reducing noise in telephone lines? Would a ferrite sleeve be of use? Shorter phone cabling?