Speed capped - out of the blue... again!

[#1]

I say again... it hasn't done it for over 2 years... but still!

Thursday 12 December 2024 around 05.30 my connection was brought down and reset remotely (ie not by me or my equipment). Prior to this reset my stats show no errors or errored seconds since around 22:30 the night before. My connection/synch was just under 10Mb/s with SNRM of around 10dB - I manage my connection to try to maintain around 10dB SNRM. The connection had been up for 4 days with quite normal looking errors (I think I was still trying to recover from a power outage that came at an inconvenient/noisy time of the day... usually the connection stays up for weeks at a time until I decide it's time to hard reset the modem).

When the connection returned after the remote reset it was sat at 6.6Mb/s with SNRM of 6dB (max line synch reported at 12+ Mb/s).

I tried a reset from my end at around 10.30 with no improvement. So I left it for the rest of the day. I got barely any errors.

This morning 13 December I powered down the modem around 10:00 and left it a few minutes. When I powered up, the connection returned with the same 6.6Mb/s (max 11.4Mb/s), SNRM 6.9dB.

Can I get a reset and removal of this cap please? The line has been fine (well, relatively... its a long, bad line!).

Thanks

[#2]

@Anywho What product and what router are you on?

[#3]

---

@Anywho wrote:

My connection/synch was just under 10Mb/s with SNRM of around 10dB - I manage my connection to try to maintain around 10dB SNRM.

... I think I was still trying to recover from a power outage that came at an inconvenient/noisy time of the day... 

When the connection returned after the remote reset it was sat at 6.6Mb/s with SNRM of 6dB (max line synch reported at 12+ Mb/s).

I tried a reset from my end at around 10.30 with no improvement. So I left it for the rest of the day. I got barely any errors.

This morning 13 December I powered down the modem around 10:00 and left it a few minutes. When I powered up, the connection returned with the same 6.6Mb/s (max 11.4Mb/s), SNRM 6.9dB.

Can I get a reset and removal of this cap please? The line has been fine (well, relatively... its a long, bad line!).

---

Sounds like you are not using a Plusnet supplier router and might be gaming the SNRM / DLM process - why?

There is zero merit in raising the SNRM (it simply reduces the sync speed) if the line can maintain a stable connection at 6dB or below.  There is limited merit in running with a 'gamed' SNRM in situations where there is significant intermittent REIN which you cannot manage out of your environment.  However, if it is that bad, the DLM will do that for you.

What noise are you referring to?  Audible noise or a known source of RFI?  If the latter, what has been done to locate it and eliminate it?

What do you mean by a "reset"?  Simply power cycling the router or something else?

What "cap" are you referring to?  A target SNRM of 6dB is the normal DLM setting; if its is higher, the sync will be slower; if the line is extremely stable, it might drop to 3dB in which case the sync will be faster.  There is an inverse relationship between TARGET SNRM (TS) and sync speed: if TS goes up, the sync speed will be lower.

It is a long bad line - how long?  Had any bad weather recently, high winds?  Is it possible that the marginal line has degraded further and what is really required is a fault report and line test?

Sight of your router stats would be helpful please.

[#4]

I appreciate the interest in my post, but I require a DLM reset to return my line to the same state I have been managing it in for years.

This used to be the best place to get Plusnet support - if this is no longer the case then please do let me know.

I am on ADSL2+ Unlimited Broadband, currently running a HH5a with OpenWRT (without any SNRM tweaks right now). But the result would be the same with any of the ADSL modems that I own, since this banding/cap has been placed on the line.

FYI I have several Plusnet supplied modems/routers that also allow me to tweak SNRM (old ones that work fine like the old Netgear DG834 for one - I have never been sent a newer one than that!! I had to buy my own HH5a!)... I just prefer to use the HH5a with Openwrt because the tweaks persist through a power cycle (and for other reasons like SQM).

I fully understand the benefits and pitfalls of tweaking the SNRM... no part of my post implied that I needed any help with any of this - I only mentioned it for completeness and perhaps to demonstrate that I know my way around a modem. DLM will happily manage my line down to an unusable level if I let it - and it has done so previously. If you would like to check my line history you will see I can manage this line quite happily at 8-9 Mbps with perfectly acceptable levels of errors and with no excess latency from BufferBloat. If your line has zero errors then you are leaving bandwidth on the table.

"There is an inverse relationship between TARGET SNRM (TS) and sync speed: if TS goes up, the sync speed will be lower"

But I explained my synch speed dropped AND my SNRM dropped (I am less interested in the target SNRM and more interested in the actual SNRM at a given time)? What does that tell you?

The reference to noise was merely part of a small explanation as to why there had been such a recent reset. It was in reference to the times of day when my line has reduced SNRM - I believe all ADSL lines experience a change in SNRM during a 24 hour period. I have learned that if I reset my line at a quieter time of day (21:00 - 22:00) and target a SNRM of 10dB my line will cope with the varying conditions just fine with an acceptable number of errors (usually not so many errors that DLM decides to increase my margins and reduce my bandwidth - or band my line).

So the noise I'm referring to is the same noise referenced by the noise denoted by "N" in "SNRM".

The resynch from 4 days earlier was related to a power outage. The power was lost at around 4am a few days before and when it returned it bounced a couple of times. When this happens I tend to wait a few days and then resynch at my usual time of day. Yes, I need to get this modem on a UPS - its awkward!

In the context of this post a reset is a complete resynch of the line. If I just say "reset" it usually means a soft resynch (usually an OpenWRT system reboot, but sometimes a stop/start of the modem service) without any power cycle of the whole device/router.

The "cap" is the DLM banding or whatever that has been applied to my line - I forget the correct terminology. The exact same bandwidth cap was imposed just over 2 years ago for no apparent reason... exactly 6.6 Mbps. And whilst 6dB might be a fine target for a nice new short line it would destroy my connection without this lowly DLM cap. Even with the cap, if I don't intervene here eventually the line will degrade to become unusable - ask me how I know.

It is entirely possible that the line has degraded further or a fault developed/worsened (I think a line test showed a fault on my line back at the end(!?) of COVID - but because my house has more than 1 vulnerable person I did not want an engineer to visit... the connection was functional enough). I never revisited this since the line performance was adequate.

Line attenuation still shows 40dB - this has not changed in 12 years.

I'm uncertain what stats you would need to prove that my line has been banded/capped? The first picture is a snapshot of the DSL line. The second is the line stats that I collected from the Thursday morning showing no errors at the time of the remote resynch (you can see the SNRM drop, the synch speed/bandwidth drop and of course the max bandwidth increase).

I think some of these errors are misnamed/mislabeled.

In order to get acceptable performance in the home (ie 2 standard definition streams and some internet browsing simultaneously) I like to synch at around 8Mbps or more with a SNRM target of around 10.5dB. This gives me some headroom to run SQM Cake at around 7Mbps and eliminate any unnecessary latency (BufferBloat).

This has been working fine for 2+ years - which is why I haven't been here whining until now!!

Again, I'm grateful for the interest, but I really need Plusnet to intervene here.

[#5]

BTW - I just started putting my lights up today (Humbug!)... so it wasn't that either!

[#6]

Something is very amiss here and I do not think this is speed banding (at least alone) - if it were you would be seeing a far higher SNRM.

A clean line having 40dB attenuation (circa 2.8kms) should deliver around 10mbps at a SNRM of 6dB.  Remember that SNRM is not an absolute metric, it is a margin over the background noise.

I would suspect that your line has deteriorated markedly, so that to attain the margin over the noise, the line needs to run much slower than normal.

Yes, ADSL lines do suffer NIGHT TIME interference from continental radio and the evening time slot you mention is the noisiest for such interference.  At worst (on a clean line) that is unlikely to deliver more than a -2dB variation. By the sounds of things, your variation is at least twice that and such would typically come to local equipment.  There was a recent case of such arising from GU10 kitchen light fittings.  That too was pursued as a DLM reset request initially.

If you wanted to reset your line for speed (at the risk of stability) your should do it at midday.

Have you checked the line for noise (dial17070 option 2)?

Have you considered running routerstats to plot the SNRM variability?

Have you invoked any of the fault reporting mechanisms?  Generally if the DLM has kicked in, to manage the line's stability, then there is little point seeking a reset of that intervention until the reason for the intervention has been identified and rectified.

@Marsh / @MatthewWheeler - any suggestions please?

[#7]

Hmmm - certainly some things there for me to think about.

FYI - I never sync lower than 9dB SNRM (sometimes DLM might kick in and move me from a 12dB target to 15dB target or vice versa messing with my offsets, but I always correct it with my tweaks that you dislike so much!) because I would get disconnects. And I always sync faster than 7Mbps usually 8-10. I have no real evidence to show though.

You may get the impression that I'm constantly tweaking, but the truth is I look at the stats every couple of weeks and just resynch if I feel like it and if it's convenient. Honest!! It can get fiddly when the electric bounces up and down, cos DLM gets upset... if I'm around then I just power down the modem when the power goes off and wait for it to return properly (same kind of problem with the pond pumps... but I digress).

The last stats that I screen dumped prior to this problem is from the week up to 13 May 2022 - the end of the last issue: Shows 8.6Mbps@10.8dB (though it does show fewer errors).

And before this latest "intervention" I was running at almost 10mbps@10dB - I had noticed my line stats had been improving lately, I just figured it was because everyone was on fibre and my cabinet was empty (no FTTC in my cab)!! (not the 40dB Line attenuation... that never changes).

The line was in use until midnight at least, with no errors after 22:30.

I've been mad busy the last few days, and I did not collect all of the stats before power cycling the modem... but I did take a copy of the last 24 hours (it doesnt have the errored seconds though).

The 7 day data for this event in my previous post makes it look like the  the datarates and SNR drops were gradual (prolly something to do with averaging). But the 24 hour data (below) shows it to be a single event - I'm sure Plusnets stats will show it as a single event as well?

This shows the 4 days up time, then a drop and a Last Uptime of 0.2 days (or 4.8 hours). Sadly it does not record the number of resynchs but I'm certain I checked the live stats and saw only 1 loss of signal for the whole 4+ days.

The synch speed and SNR changes look more precise/singular here (ie 1 event).

It does show that I was synching right on the limit of my lines capacity at 10dB and it shows a lot of errors (relative to what I usually see) over the busy evening period... I'm now thinking this was my fault for not leaving enough headroom. I probably intended to keep a closer eye on it - but life happens!

The resynch (that I initiated) from 4 days before this graph was done with a -2.0 Target SNRM. This implies DLM was targeting 12dB... why would it jump straight to banding without targeting 15dB first?

I haven't yet done any of the diagnostics you suggest.

I will definitely check the Quiet Line or whatever... I did introduce an extension to a telephone socket in the Kitchen a couple weeks back. But that socket doesn't carry digital/broadband (none of them do except the modem one!) so should not matter.

I may use routerstats again (although I think I had issues last time I tried on this HH5a, not sure. It worked a treat on the old Netgear) but OpenWRT is already recording this information so it is not a priority.

The GU10 bulb is a new one on me. I have a few of them, and I think a couple of them are real cheap and nasty too. They are rarely on, but I will keep it in mind.

If DLM was as conscientious about removing banding as it is about applying it then I would just wait it out... but I seem to remember that DLM will never remove banding once it is applied.

In the past Plusnet had no problem with requesting a DLM reset... has this all changed now?

Thanks again btw.

[#8]

Ah telephone extensions … that’s a whole challenging discipline in its own right. They might not “carry” the digital signal as you put it, except if they are connected to the filtered side of a faceplate. Even then if they are star wired with other extensions … there can be unwanted complications. Is the ring circuit connected?

[#9]

The DSL is filtered at the point of entry (modern NTE5 thingy?) so none of the extensions can be used for broadband and any ring/bell wire signal is filtered here also so should not have any effect.

As it happens, I had an old novelty dancing Tigger (from Winnie-the-Pooh) 'phone plugged in at that kitchen extension a while ago - this required me to use an ADSL splitter (or anything else with the necessary ring capacitor) in order to get it to ring. So I know there is no bell/ring wire signal at this extension.

You said "They might not “carry” the digital signal as you put it...". Did I get something wrong or is this just pedantry? How would you describe an extension that has had the carrier frequencies for broadband signals pre-filtered? I'm always keen to learn to be more precise - especially on specialist subjects that I rarely look at.

[#10]

---

@Anywho wrote:

Data Rate 6.648/1.12 Mb/s

SNR 7.3/6.1dB

With these figures there is no way that a DLM reset to 6dB would improve your speed, especially if you then move the SNR to 10dB. So either:

This data was captured at a time of high noise on the line, or, your tinkering with SNR has got your router/DLM confused and is incorrectly reporting the true situation. Maybe if a reset solves the problem it is the latter?

Have you tried a factory reset on the router?

[#11]

---

@Anywho wrote:

You said "They might not “carry” the digital signal as you put it...". Did I get something wrong or is this just pedantry? How would you describe an extension that has had the carrier frequencies for broadband signals pre-filtered? I'm always keen to learn to be more precise - especially on specialist subjects that I rarely look at.

---

Nah!!  Just a turn of phrase, off the words which jumped out at me ... skim reading the post on a mobile device.  Your description was fine and indeed technically accurate, but I think most would have said that the extension wire was filtered at the master socket.

Extensions wired off the back of the NTE5c (or which ever later version) is very much the way to do things, however no everyone does these things right and bitter experience points to the wisdom of checking.  It is important though that if there is more that one extension, that they should not be star wired - that is M->X1 and M-X2 - they should be wired M->X1->X2.  There should be no long tails on the wires pushed into the connectors.

Pond pumps ... sounds like another potential source of interference.  Take a look at this - ::. Kitz - REIN, SHINE & RFI .:: - and have a wander around with an AM radio tuned to 612kHz.

DLM resets are orders on BTOR which are probably chargeable to the business; there is certainly less inclination to do speculative DLM resets before checking out the circuit for faults.  That said, if the DLM is aiming for a 6dB SNRM, it is not capped.  The reduced speed is that attainable whilst delivering a 6dB MARGIN over the present (apparently chronic) interference.

Setting a 10dB SNRM will...

1. Establish a lower sync speed
2. Make the line more stable and less susceptible to REIN noise spikes

The best advice here is to look at what is generating noise / what is making your line more susceptible** to it. the best steps are...

1. Do a QLT - report a VOICE fault if there is noise
2. If the line is quiet - report a BROADBAND fault (slow speeds)
3. Hunt for REIN / PEIN / SHINE sources

** Faulty lines are more susceptible to picking up noise.

Router stats - those provided by the router do not show sufficient granularity to be informative.  Looks as though something grim happened at 17:30.  Prior to that there was a lot of signal headroom (10dB), then there was a resync with (at whatever speed) a 6dB margin.

Your challenge is to find out what happens at 17:30 and runs thereafter.  The error stats look significant from after 17:00 until late evening.  I have seen multiple scenarios like this, arising from CCTV IR illuminators (only an issue when it gets dark ... but nothing obvious switches on at dusk), passing mainline trains ... all made worse by yards of UNTWISTED PAIR star telephone extension wires.

[#12]

"a DLM reset to 6dB" - Hmmm... I think one of us is misunderstanding. It is probably my terminology.

What I want is for the banding (the cap, the newly applied DLM restriction... whatever) to be removed from my line. In the past, I rather got the impression that this was something Plusnet could easily do/order. If that situation has changed over time then I guess I need a new strategy.

I most certainly do not want to be reset with a 6dB target SNRM - my line has never been able to sustain a connection at that margin (unless capped like now). My optimum target was 10.5dB. But anything more than that works too.

After my problem previously (over 2 years ago) I thought my line had been configured to stop DLM intervening (using some parameters Plusnet could set for how many errors/resets etc are acceptable or something) - though my recollection is poor and it's possible this got reverted... we had quite a protracted exchange of ideas back then! And many "cooks" by the end.

As then, I would love it if you could reset my line to be completely unmanaged/unbanded/uncapped and remain completely ignored by DLM - allowing me to control my errors/resets/etc. A situation where I pay for the line and I'll manage the line sort of thing.

Those live stats you quoted @Baldrick1 are from over 30+ hours after the mystery reset/banding.

Recap

There was a mystery reset at 05.30 12 Dec 24, that just took all of my bandwidth and apparently capped my synch speed at the 6.648Mbps shown (just like it did 2+ years ago) despite being able to support more. I have since performed a software reset of my modem/router and a full power cycle and each time the speed is capped at the exact 6.648Mbps shown. I do/did not have any SNR tweaks in place for the last power cycle so I am running as DLM intends.

If you look at the graphs from before the mystery reset you will see the line was running (well, erroring) at nearly 10mbps with 10dB SNR margin.

All of the errors for that 4-day period prior to the mystery were my fault; I think I got a bit cocky (following a power cut) and started to push my line (needlessly since SQM cake restricts bandwidth to around 7Mbps anyway - I was just looking for more headroom for SQM and shot myself in the foot!!). I hadn't monitored my changes, and the errors weren't causing huge headaches (although with hindsight it might explain the few streaming buffering problems I was seeing that I just put down to internet/network/provider issues).

Prior to my misplaced optimism my line would run for weeks at a time at 8-9Mbps with 10-11dB SNR margin - with some errors, but not enough to upset DLM. You must have stats to confirm this? Only resetting because I like to power cycle that router occasionally.

Perhaps interesting to someone is that my margin seems to be very gradually INCREASING (yes, increasing?!) since my last reset. This is since my power cycle:

15Dec24

You can see I get a good few errored seconds with such low (8dB!!) SNRM even with the speed cap in place.

And if you wanted the latest live stats

15Dec24

Maybe I can circumvent this DLM intrusion by restricting my modem to G.992.3 (ADSL2) only? I know I used to use ADSL2 before I had access to persistent SNR tweaks.

I can perform a factory reset - can you explain how me resetting could remove the DLM banding?

As for confusion, BT's own speedtest reported my download speed at around 4.8Mbps yesterday and just gave me 5.39Mbps just now. I don't think my modem is confused I think it is reporting the line state pretty accurately; I'm banded and capped! (Incidentally, the additional diagnostic tests at BTWholesale still fail for me!)

15Dec24

So, can I get the banding removed? Or is this something that has become difficult to achieve?

[#13]

Fair enough! I agree, it is fair to assume lesser knowledge. And I agree my wording was not explicit and your phrasing is better! I don't use forums often; that's my excuse!

I will check my pumps, though they all appear in good order and nothing has changed (to my knowledge). But worth a check. I bought a tiny radio last time I had issues just for this purpose!!

I have absolutely no idea about the phone extensions. Stars? Looks like I need to do some research - but not just yet... I can just unplug the 'phone in the kitchen for now!!

In the past I did not get the impression that the DLM resets (or perhaps I should be saying removal of banding or profiling or something?) cost Plusnet - they always seemed happy to help? But I take your point... if this is now chargeable I agree free avenues should be thoroughly explored first.

You can see on the graphs that the max line speed is higher than the current synch speed... the max line speed reflects the speed for the current SNRM... the very fact that the line has synched lower than the max speed indicates to me that a cap is in place (or the line was very noisy during negotiation; but since this has reset/renegotiated 3 times at the exact same speed I'm inclined to think that noise was not the limiting factor).

Doesn't routerstats require a PC to be running and regularly polling the modem? I don't currently leave PC's on over night so I wouldn't have caught this. Maybe routerstats can be run in OpenWRT?

What granularity is required? It's too late for this event, but the stats are all recorded at much lower granularity using rrdcollect or something like that I think. But the data files are all recorded in memory and I have not had the need for a process to collect and backup this data - but that is possible.

I did not collect the data prior to my last power cycle; I honestly thought (hoped) the power cycle would correct everything!!

Woulda shoulda coulda!

[#14]

The DSL is filtered at the point of entry (modern NTE5 thingy?) so none of the extensions can be used for broadband and any ring/bell wire signal is filtered here also so should not have any effect.

@Anywho just checking that you are aware that the filtered connector on the NTE5c Mk 4 is not the one on the faceplate . Unlike previous faceplates the connectors on the back of the faceplate are unfiltered. The filtered connector is the one on the backplate. 

[#15]

---

@Anywho wrote:

"a DLM reset to 6dB" - Hmmm... I think one of us is misunderstanding. It is probably my terminology.

What I want is for the banding (the cap, the newly applied DLM restriction... whatever) to be removed from my line..

---

I am certainly not misunderstanding.  A DLM reset removes any speed capping (if applied) and resets the TARGET SNRM to 6dB.

If this line was speed capped, you might see an elevated dynamic SNRM, as there would be a higher margin at the slower speed.  All the evidence offered points to a line fault.  Why not simply invoke the faults reporting options?  If no fault is found, THEN you have grounds for exploring other options.

If there is a fault, doing a DLM reset will most likely simply spin the merry-go-round yet again.

As for current stats, snap shots are not helpful, what is needed here is 10 second interval plotting of the current SNRM (and before and after sync speed in the event of a disconnection).  Views such as the below provide a valuable insight as to what is really happening on the line.  As you note there are spates of chronic errors ... a DLM reset will not fix that.

This 10 second** interval SNRM plot shows a line suffering chronic short burst** noise from 20:28 (what got switched on?) until 22:50 when it stopped abruptly (what got switched off).  Until you can see what (if anything similar) is happening on your line you are not going to FIX your issue.  The approach you have been taking appears to be little more than managing the symptoms at the margin.  It is probable that what ever is the issue here has degraded further and is no longer manageable by the same methods.

** A longer sample interval would be unlikely to discern the short burse noise issue.

Do the QLT test and then raise the appropriate fault report - VOICE if there is noise on the line - BROADBAND if there is not.