DLM reset request (ADSL2+)

[#1]

---

@Mardler wrote:

It seems that the DLM ups the SNR as soon as a decrease to c.5.4/5.5 is detected while retaining sync. If that's the case there's no point in SNR tweaking now, I will however, pursue why the full short line potential isn't being achieved.

I expect this SNR behaviour to continue (it probably has in the past with months long stable connections, I just wasn't looking) so unless something dramatic happens I won't clutter the topic further on this.

---

Unless there is a disconnection and resync, the DLM is doing nothing here.

At the risk of telling you something which you know already, the following is a simplification of how ADSL line management works...

• Depending on  the line's performance history it has a target SNRM set by the DLM - 6dB is the standard, lower suggests an exceptionally stable line, higher suggests that the line is being managed for stability - the values are altered in 3dB steps
• At the time of line sync the router and the DLM negotiate the fastest sync speed which can be obtained whilst still delivering the target SNRM over the then present line noise
• Then subsequently if the background noise level...
  • abates - the observed SNRM will rise
  • increases - the observed SNRM will fall

Remember that this is a MARGIN figure, that is how much signal there is ABOVE the background noise.

Your reported observations of current SNRM suggest to me that around 19:05 (dusk) you have a "noise generator" waking up and by 23:00 is much worse but improves towards 00:45 and abates altogether sometime later.  The scope of the variation is 5+dB which with a line having a target SNRM of 6dB would most likely assure a sync disconnection.  The DLM setting the target SNRM to 9dB gives extra headroom to maintain stability - at worse the margin would fall to 4dB.  Gaming the target SNRM within the router removes that safety margin, resulting in a disconnection (possibly every night).  That would seem to be a close match to your documented experience.

The fact that the observed SNRM is seen to be above 9dB would suggest that the back ground noise at the time of observation is less than it was at the time of the last resync.

Here is the challenge - profile the noise pattern (routerstats is the best tool for that).  Having obtained day on day profiling, seek out some electrical appliance which changes its operation profile at the times indicated by the graphs.

I have done this several times over the years and found correlation with...

• Street lighting
• CCTV PSU - turned out that when the IR illuminators kicked in they overloaded the cameras PSU - I only located that one after thinking "I have unplugged everything ... anyway this is on all day" ... only after unplugging it did the SRNM return to what it was during the day ... and on plugging it in again, it gradually deteriorated to its dark hours worse
• Passing West Coast electric trains (but only the 11 coach variants) - having eliminated the likely, I search for the obscure and found that I could match the SNRM excursions to the train time table

If you can identify and eliminate what looks like a source of RFI you might yet get a solid 22mbps or better.

I am now slightly less inclined to thing you have a line fault, beyond possibly a leg imbalance which gives rise to an enhanced susceptibility to RFI.  Whatever that 5dB variation is huge!

[#2]

Here is an example SNRM monitoring plot from a previous investigation.  Note however that this interference here was predominantly on the US (all depends on the frequency range of the interference).  Upstream failure can equally impact the line's stability ... and the DLM will attempt to manage that by slugging the DS!!

[#3]

No problem with the explanation, TM, and yes I knew that hence saying seems above.

I expect the fall in reported SNR at dusk but not the jump. The first time I observed this it was around midnight only, yesterday it was also early evening.

I agree that RFI could be the culprit. There's no internal device switching at those times, though the TV has a measurable effect it isn't switched at those times. Finding an external RFI source isn't easy, years ago I wandered around the vicinity and along the line wearing headphones with a portable radio tuned to c.512-532metres AM & heard nothing apart from the horrendous noise from the TV (the phone line runs under the eaves right above it so back then just turning the set on killed the connection) once the rotted, wet, cabling was replaced the TV had no effect.

I used to use routerstats years ago with a Netgear DG834 but no longer have a fast PC. I will try routerstats on the old laptop.

[#4]

Hello again, folks, I hope for the last time on this topic.

In short, the connection has been stable for long enough to close the ticket.

Here's most of my response to my PN tech contact:-

"the connection has been stable.

While I would still like to know why the line isn't delivering its full potential, 18.243 d/l on a stable connection is good enough for our purposes. I tried to include a jpeg of the router stats here but pasting didn't work, if you want a copy I will try it in a dedicated response with no text.

I reckon pursuing another 2Mbps is pointless. So, until the next series of disconnects and/or a power cut or we go FTTP, I consider the matter resolved.

I will update my forum topic accordingly.

Many thanks for all of your help, your service has been excellent."

Fingers crossed!

[#5]

Good timing - I planned to enquire on progress later today.

I suggest that there’s a fairly short answer to your question - your line is very stable with an elevated SNRM at the expense of a higher speed.

Your late at night line state is apparently not stable at 6dB which would deliver a higher speed. Did you explore running routerstats to profile the later evening deviations?

[#6]

Hi TM,

Not sure about the timing: as I was posting on the ticket and here I was thinking that knowing my luck that'll guarantee a disconnect tonight. Sure enough, at 23:20 it happened.

The reconnect was at 6.2SNR with 17783 d/l (upload very slightly improved) so not much difference. SNR is now 10.1.

I could try a router reconnect (IIRC it has a "silent" reset) in the afternoon but am not too bothered.

Been busy with other stuff so haven't run routerstats yet.

R

[#7]

That is in a way great news / a profound observation: at the time of the disconnection (and re-sync) there was at least 4dB of additional background noise. A re-sync around midday might well deliver a higher sync speed, but there will not be enough headroom for the evening noise spike. Consider that a 3dB change is a doubling (or halving) of what is being measured.

Personally I would now focus in profiling this late evening noise. Router stats install and running … and possibly ly a MW radio switched on ready to listen for and possibly track down the source.

Excellent perseverance here - I think you now have clarity on what is blighting your service: other than the slight possibility of poor lag balance (which can increase susceptibility to REIN) I do not think it is likely that there is an issue with the line.

[#8]

Thanks, TM.

The AM radio has been useful over the years. It revealed nothing untoward a few days back but it's ready to go.

Shame my PC is kaput, routerstats is on it. Will try the old laptop soon.

ITM, SNR is currently 11.5 and line stable since last night's disconnect as well it should be. The attainable rate is 22440 which means that even allowing for overheads 17783 is low (I would expect sync at c.19500).

Whatever it's good enough for us at the moment.

[#9]

The attainable rate is that which might be attained at a 6dB SNRM.  If you were to re-sync now, that is what you might see; however the night time noise profile (5+dB ?) would probably be enough to cause a loss of sync and the resync at that time will be slower.

It is good to have this now well profiled: a good line, target SNRM is still 6dB ... however there is pronounced local noise around late evening.  If you can live with the speed, then I'll leave you with it.  If you want help with router stats or the plots, then give me a shout.

[#10]

Thanks, TM.

The fluctuating evening SNR has been nagging me for days. I spotted another drop tonight so immediately tuned my portable AM radio to 612Khz and was assailed by horrendous noise. To be certain I wandered round outside: the noise dropped and the Spanish station on that frequency came in clearly the further away from the house.

What could it be? The worst culprit is the TV but it's off. Eventually it was traced to the kitchen GU10 LED spotlights running from a dimmer switch. The switch is a double but the dimmer on this pot doesn't work just the on/off. I now have to confirm my opinion that it's the switch not the lamps.

This completely explains the intermittent timing of the SNR drop.

Time to fit the spare double dimmer!

I had drafted a response to your latest so here it is albeit somewhat overtaken by events:-

"I understand attainable rate hence my musings on such a short line not delivering the expected sync.

The recent disconnect and resync dropped the rate by roughly 500kbps which for our usage is immaterial. The increase in SNR from an average daytime 10.4 to 11.4 is intriguing, if the connection was rock solid for 12 days the DLM should have dropped the SNR not raised it. At 17:10 this evening it was 10.8. The previous drops around 10:00 to 11:30 seem to have ceased or more likely I haven't been looking for them due to the stable line. Btw, the lowest observed rate was 5.2.

PN report the last disconnect as follows: "The drop reason is also showing as an 'interim update', so a normal disconnection cause, and is not hitting our triggers for further investigation." "

I am still talking to PN but mainly to make it clear that the ticket can be closed.

[#11]

This thread reminds me of when I was last on ADSL2+, a couple of years ago. Why wait for FTTP? I switched to FTTC (40/10 - no longer available AFAIK) and have never looked back. 😊

I assume it is still possible to switch from ADSL to VDSL?

[#12]

Looking back, can you correlate the use of that light with disconnections?

Can you correlate dynamic changes in SNRM with that light dimmer being turned on / off?

I recognise the satisfaction of finding the “it could not possibly be that” cause of RFI. When I discovered that the PSU for the CCTV cameras was the source of the issue … though incredulous it was a joy. I found that switching it off brought an instant reduction in noise, though on turning it back on, it took several minutes for the full level of noise to reappear.

[#13]

TM:-

"Looking back, can you correlate the use of that light with disconnections?"

>>> In the past, definitely not necessarily as disconnects happened during daylight hours as well as dark. This time, quite possibly. I think I have the old logs on a backup so will check.

"Can you correlate dynamic changes in SNRM with that light dimmer being turned on/off?"

>>> Definitely and c.5db too. I will try to post a jpeg shortly but all previous attempts have failed.

The TV accounts for another 1db reduction which the time isn't a problem.

[#14]

As suspected, unable to post a jpeg.

[#15]

pvmb:-

I agree that we need to go FTTP.

However, it's far from straightforward especially hereabouts where OR subcontractors are notoriously bad.

No initial OR survey is one problem but by far the biggest is losing the fixed line phone & number which is a separate topic which need not be disclosed here.