Slow broadband and occasional drops

[#1]

The sync speed was way above the estimates, hopefully it doesn't go down by much as a result of the Super Stable setting. 

[#2]

Thanks for keeping on top of this @Gandalf.
My speeds have been very low, this last couple of days, down from 14/15 down and 1 up. I'm now getting about 2 Mbps down and 0.3 up. Ping is also quite high. Hopefully this is temporary while they try a new fix.

[#3]

Maybe not temporary - what’s the SNRM?

Is that sync speed as reported by the router or data speed reported by a speed tester? Only sync speed reports are needed here.

In the presence of a problem (possibly REIN) connection stability is only obtained at the sacrifice of line speed.

[#4]

The speeds reported were from a speed tester @Townman. I'm pasting router stats I got just now if it helps. It also looks like there was a disconnection around 30 ish mins ago.

I'm not sure how to obtain speed sync info.

Your Plusnet adviser may ask you for the following information:
1 - Product name: Plusnet Router
2 - Serial number: N7192141B020066
3 - Firmware version: 7.275.11_F2704N_Plusnet
4 - Board version: F@ST2704N
5 - Mode ADSL
6 - Uptime: 0 days 00:31:26
7 - Data rate: 888/3071
8 - Maximum data rate: 888/21136
9 - Noise margin: 16.3/31.1
10 - Line attenuation: 8.0/17.0
12 - Data sent/received: 11.9M/183.4M

[#5]

@Mjan 

Thank you, this is highly useful, for it starts to indicate what your line should do against what it is doing.  I will take the numbered items in tun and try to explain them, is it very technical, but it might help to understand the detail...

Many of the metrics lines have A / B values - A is upstream (sending) and B is downstream (receiving) and it is the B figures which are of most interest.

10 Line attenuation - this is a factor of the length of the line; it directly informs of the maximum possible downstream sync speed (raw ones and zeros) which can be transmitted over the line, all other things being ideal.  At 17dB the line should deliver 21Mbps at a SNRM of 6dB in perfect conditions - such is reflected on line 11 Maximun data rate (only it is not actually data rate, it is line sync speed).

9 Noise margin - this is the current Signal to Noise Ratio Margin (SNRM) - Signal to noise is a measure of how much stronger is the signal over the background noise, the better the margin the easier it is to send and receive.  Consider the difference of trying to have a conversation with someone in a quiet pub compared to a noisy one.  In a quiet pub two people can stand at the opposite ends of the bar and have an understandable conversation at normal speach levels without difficulty.  However in a noisy pub, you can be stood next to each other, shouting at the top of your voices in each other's faces ... and still have no chance of understanding the message!!  When the music stops, you can then hear each other easier.

Broadband is much the same.  The Noise Margin reported on line 9 is the CURRENT value and it is dynamic.  Often it is measuring this figure at regualr intervals (by plotting of this figure) that can illuminate what is happening on a line.  Point measurements are rarely illuminating, unless you know what you are looking at.

The figure on its own without knowledge of the target SNRM is not overly informative.  The TARGET SNRM is set for the line by whatever profile the supplier sets: @Gandalf reported that the line has beens et for stability.  Targets are set in 3dB increments, normally 6dB on a standard line, possibly 3dB on a really good line, then 9dB upwards on less good lines.

When the router establishes a link with the exchange, the two ends negociate a sync speed which will deliver the target SNRM above the THEN PRESENT line noise - the more the noise, the lower the speed.  Once the sync speed has been negociated, that speed remains fixed for that xDSL session.  If the noise abates the observed (current) SNRM will be seen to increase.

It is therefore possible that on your line, there was a blast of electrical noise which caused the xDSL session to be lost.  The router then seeks to resync (in the presence of that noise) at say 15dB, to do so it has to push the sync speed down to 3kbps.  When the noise abates the margin then jumps up to something significantly higher.

It would be intersting to see what the xDSL session negociation metrics were - the router log usually records that information in the xDSL establishment log entry.

31.1dB is a massive SNR MARGIN, further it is not a nice (near enough) multiple of 3dB which in itself suggests the presence of noise.

I would wager that if you were to now bounce your router and immediately look at the stats they would be markedly different.

I do not know how to move this forward for you with access to a router which is sympathtic to being closley monitored.  The Hub Zero blocks automated access to such information.  The Plusnet Hub One is a badged BT HH5a which will work with router stats.

That said I am advised elsewhere that detailed information is available somewhere deep in BT's logging systems but that might be only for FTTC services? @Gandalf Wholesale's RRT tool?

[#6]

Attached is a RS plot which illustartes the link between SNRM and sync speed.  It shows a noisy SNRM plot (blue line), the SNRM crashes to zero thereby dropping the SNRM session.  A new xDSL session is created at the target SNRM ... but at a lower speed (red line).

[#7]

Thanks for getting back to us @Mjan 

I can see the order to activate interleaving and move your line on to the Super Stable profile completed in the early hours of the morning and since then your router is getting a speed of 18mbps and your line seems to be stable so far:

There may be a few drops over the course of the next week or so as the DLM (Dynamic Line Management) training period will have been restarted but hopefully you now won't get the level of drops you had previously.

I'd recommend monitoring this and let me know if you notice any further issues, I'll personally keep an eye on this too. I'll check back on Wednesday to see how your line is doing after a couple of days with interleaving and on Super Stable.

[#8]

@Gandalf what is now the target SNRM for this line please?

@Mjan can you please post some new router stats.

[#9]

@Townman

Your Plusnet adviser may ask you for the following information:
1 - Product name: Plusnet Router
2 - Serial number: N7192141B020066
3 - Firmware version: 7.275.11_F2704N_Plusnet
4 - Board version: F@ST2704N
5 - Mode ADSL
6 - Uptime: 0 days 00:13:59
7 - Data rate: 888/2093
8 - Maximum data rate: 888/20676
9 - Noise margin: 13.6/34.6
10 - Line attenuation: 8.1/17.0
12 - Data sent/received: 4.3M/38.8M

@Gandalf

Thanks for the updates. So far it seems stable although the speeds are variable and, sometimes, so slow it stops access to the internet momentarily. But it hasn't disconnected. I've got speeds of 15 Mbps last evening but this morning I can't get over 2 Mbps, for example.
But let's give it a few days and see how it goes. I'll keep you updated.

[#10]

The drops seem to be still happening, but not as frequent as before:

Your speeds have been pushed right down to 2mbps though by the DLM at the exchange, restricting the speed to a max. of 3mbps by applying a banding of 1M to 3M to your line and increasing the SNRM (noise margin) to 34.6dB

"WBC 1M - 3M High delay (INP 8 )6dB Downstream, UC Medium delay (INP 2) 6dB Upstream (ADSL2+)"

It looks like that's to be expected as the DLM is doing what it's been told by the Super Stable profile, it's prioritising stability and reducing the error count in favour of speed. 

I personally don't like it as this would likely be masking an underlying fault with your service, but given the engineers we've arranged previously, I'd agree in monitoring this for the next few days to see if your speeds increase to where we'd expect to between 10.5mbps and 14.5mbps without causing the level of drops you've had before.

If your speeds don't increase to the target maintaining stability, we'd need to arrange another engineer visit and I'll get an escalation raised to Openreach for a thorough investigation.

[#11]

@Gandalf, thanks again for the update.
If you think an engineer will be needed again, could we book for Friday 24th January, please?
That's the only day someone will be home during the times they attend.

[#12]

No problem @Mjan We should hopefully know before the end of the week whether we'd need to arrange another engineer. It'll also give us a good amount of notice for us to escalate this sufficiently with the engineers on the ground.

[#13]

@Gandalf 

You are quite right, thos stats are the pits!!  However ...

"WBC 1M - 3M High delay (INP 8 )6dB Downstream, UC Medium delay (INP 2) 6dB Upstream (ADSL2+)"

Does that imply the target SNRM is 6dB?

Is the observed 34dB arising from the DLM banding or possibly the presence of a high level of noise at the time the router rescyn'd?

If the DLM is banded, can it loose its bands of its own accord - or does that need wizard grade magic?

I fear there is a risk that line inspection will find not fault (because there is not line fault) and the recommendation will be to release the banding, which will be fine until a point event (or series thereof) drops the link again repeatedly.

[#14]

@Townman Yes the SNRM target is at 6dB and DLM can change the banding as it sees fit based on the parameters we've set (in this case Super Stable) and in order for a line to be banded, the SNRM is increased.

Realistically if the speeds are below expectations and/or the line is still unstable then there isn't really anything more we can do bar arrange another engineer and escalate this.

[#15]

Good afternoon @Gandalf,

I'm not noticing any improvements whatsoever. The internet is unusable apart from browsing and maybe very low quality video. Last night I was struggling to reach 1mbps on speed tests. I haven't noticed many disconnections but, to be honest, I nearly gave up using the home broadband due to frustration so I don't even pay attention that much.

It's extremely disappointing that this is still an issue and that nothing seems to make a difference.

P.S. - I just noticed that this thread has 9 pages, 135 posts and has been going since March last year.