Kitz ADSL Broadband Information
adsl spacer  
Support this site
Home Broadband ISPs Tech Routers Wiki Forum
 
     
   Compare ISP   Rate your ISP
   Glossary   Glossary
 
Please login or register.

Login with username, password and session length
Advanced search  

News:

Author Topic: Things that go bump in the QLN  (Read 1703 times)

toonshorty

  • Member
  • **
  • Posts: 18
Things that go bump in the QLN
« on: July 13, 2021, 09:15:21 PM »

I've recently moved into a new house (different cabinet, same exchange) and despite Openreach supposedly not activating the line until Wednesday 14th, I left the HG612 connected and the DSL light came alive on the Sunday. I don't know the exact line length but I suspect one of two routes which would give lengths of 580m or 720m; given the current speeds, I would be surprised if it was the latter though. I've just been looking over the charts and there is a noticable spike on both the QLN and Hlog charts at around tone 3200. The QLN also shows some bumps at around 1350 and 1700 but those aren't reflected in the Hlog. Is this normal for a line of this length or does it indicate an issue of some kind? The impacted estimates for the line were 59.3 - 39.3 so it's well within that, so I probably can't kick up much of a fuss. The one thing I haven't been able to is an actual quiet line test with a phone as I don't actually own one. The master socket is an NTE5C which doesn't seem to really sit properly and the faceplate comes off with the lightest of pulls - is that normal or is the socket itself broken? My old address had a Mk3 NTE5a which was screwed on so never had any issues.

I'm hoping that the DLM will start to bring the SNRM down to 3dB over the coming week or so (assuming DLM gets reset when you restart a stopped line?) but for now the stats are:

Code: [Select]
DSLAM type / SW version: BDCM:0xc01c (192.28) / v0xc01c
Modem/router firmware:  AnnexA version - A2pv6C038m.d24j
DSL mode:                VDSL2 Profile 17a
Status:                  Showtime
Uptime:                  2 days 1 hour 9 min 17 sec
Resyncs:                0 (since 13 Jul 2021 18:24:48)

Downstream Upstream
Line attenuation (dB):  21.4 0.0
Signal attenuation (dB): Not monitored
Connection speed (kbps): 54766 9793
SNR margin (dB):        5.3 6.2
Power (dBm):            13.7 7.0
Interleave depth:        8 1
INP:                    48.00 0
G.INP:                  Enabled Not enabled
Vectoring status:        1 (VECT_FULL)

RSCorr/RS (%):          0.8537 0.0173
RSUnCorr/RS (%):        0.0000 0.0000
ES/hour:                0 0

This is a BDUK cabinet that was installed for EO lines and it appears this has vectoring enabled as well which given the cabinet is currently full is probably helping an awful lot.





« Last Edit: July 13, 2021, 09:20:16 PM by toonshorty »
Logged

Chrysalis

  • Content Team
  • Addicted Kitizen
  • *
  • Posts: 7382
  • VM Gig1 - AAISP L2TP
Re: Things that go bump in the QLN
« Reply #1 on: July 14, 2021, 12:54:35 AM »

Pretty nice bit loading on your D1, I expect that's vectoring doing its job nicely. :)

Logged

kitz

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 33881
  • Trinity: Most guys do.
    • http://www.kitz.co.uk
Re: Things that go bump in the QLN
« Reply #2 on: July 14, 2021, 09:42:44 PM »

>> The QLN also shows some bumps at around 1350 and 1700 but those aren't reflected in the Hlog. Is this normal

Yes.   Because the 2 tests are intended to show different things.   

QLN measures the amount of noise on the line.  The noise source could be RFI/EMI but not exclusively.   It could be due to a faulty joint or crosstalk.

Hlog is based on line attenuation and how much the signal degrades over distance between the modem and the DSLAM.  Faults that show up in here could be faulty filters, water ingress, capacitive faults and wave reflections from a bridge tap.


The fact that both show 'something' at around tone 3200...  based on the shape of the rest of the hlog plot and QLN would make me take a stab in the dark that its either a slightly worn cable or bad joint that could be difficult to find.  It covers so few tones that it wont be making much difference to your overall speed... and outside your bit load range anyhow.   

Your line is a wee bit noisy and not as clean as it could be,  but I've seen far worse.  The thing I like the least is the noise spike at around tone 1700 affecting your bit load...  but again only covering a few tones at around 7bits each, so hardly service impacting in the grand scheme of things. 

>> Pretty nice bit loading on your D1,

Isn't it just :)
Logged
Please do not PM me with queries for broadband help as I may not be able to respond.
-----
How to get your router line stats :: ADSL Exchange Checker

tiffy

  • Kitizen
  • ****
  • Posts: 1319
Re: Things that go bump in the QLN
« Reply #3 on: July 15, 2021, 04:13:02 PM »

Further to the observations and comments from Chrysalis and kitz:

Quote
(assuming DLM gets reset when you restart a stopped line?)
Yes, would certainly assume this to be the case on a Huawei cabinet, from my own experience migrating from Plusnet to Zen on VDSL-2 service last December, DS SNRM started off at approx 6dB and then migrated back to 3dB by 1dB steps at 2 day intervals producing the associated DS synch speed increases.
During one of the DLM instigated steps my G.Inp re-tx profile also changed from low to high and I had to perform moderate modem DS speed clamping for a few days after achieving 3dB to encourage DLM to revert to DS G.Inp low profile again.

Provided you don't have excessive error rates there's no reason to think that progressive DLM action won't achieve DS 3dB SNRM.
As you are one of the very lucky few to have active Vectoring your error rates I would have thought will be minimal.
DS attenuation of 21.4db by normal standards would indicate that your line is long(ish), unless Vectoring effects this indication, however, should still be well capable of very reasonable speeds on 80/10 service especially considering the presence of active Vectoring.

From observation of the modem stats provided I would guess that your line is currently on DS G.Inp re-tx low profile, you can positively confirm this by looking at Bearer 0 INPRein, 0 = low, 1 = high.
You are likely aware of the effect of G.Inp re-tx profile but just in case:
Low re-tx profile results in BRAS IP profile of approx 97% of DS synch speed.
High re-tx profile results in BRAS IP profile of approx 93% of DS synch speed.
ie., with low re-tx profile better data speed for any given DS synch speed. 
Logged
BT FTTP 150/30, BT Smart Hub 2

toonshorty

  • Member
  • **
  • Posts: 18
Re: Things that go bump in the QLN
« Reply #4 on: July 16, 2021, 12:13:26 AM »

Thanks all for your comments, that's really useful information.

The line is now at 4 days uptime and the SNR is currently sat at 5.2 - although I think this is more a case of the line getting noiser after it first synced rather than a 5dB target.

ES/hour on the upstream varies, I've seen it at 0 a fair bit although it did creep up earlier today but only as high as 14 which I think is well within limits?

This will hopefully just be a stop gap until a local gigabit voucher scheme gets signed off and built (I believe this is currently planned) and/or Openreach get FTTP out to us (I believe this is also planned - albeit by 2026).
« Last Edit: July 16, 2021, 01:36:19 PM by toonshorty »
Logged

toonshorty

  • Member
  • **
  • Posts: 18
Re: Things that go bump in the QLN
« Reply #5 on: July 18, 2021, 11:44:10 AM »

During one of the DLM instigated steps my G.Inp re-tx profile also changed from low to high and I had to perform moderate modem DS speed clamping for a few days after achieving 3dB to encourage DLM to revert to DS G.Inp low profile again.

Is this something you can do on the HG612?

I've had another resync overnight and speeds have bumped up slightly to 55892/10163, unfortunately it appears some of the INP settings have also changed.

Interleave depth has moved from 8/1 to 4/1, INP has gone from 48/0 to 55/0 and crucially INPRein is now 1 on the DS. The current SNRM is still 5.3dB which is pretty much where it's been for the last week so it doesn't look like it's targetting any lower SNR margins just yet. I can also see that the profile on the line has reduced to 50.85 Mb/s (according to the aquiss circuit information) which confirms your comments regarding the re-tx profiles. Is it worth trying to help it along or is it best just left alone for a few weeks to sort itself out?
Logged

tiffy

  • Kitizen
  • ****
  • Posts: 1319
Re: Things that go bump in the QLN
« Reply #6 on: July 18, 2021, 02:54:17 PM »

Quote
I've had another resync overnight and speeds have bumped up slightly to 55892/10163, unfortunately it appears some of the INP settings have also changed.
DLM would appear to be performing to expectation after a DLM reset, ie, decreasing DS target SNRM by 1dB steps at approx 2 day intervals towards 3dB which hopefully will be achievable on your line.
As predicted, DS G.Inp re-tx profile has changed from low to high and from my experience will likely stay there through subsequent DS SNRM step changes towards 3 dB.

Assuming the figures quoted are synch speeds then:
50.85/55.89*100 = 90.98%
Certainly more than indicative of DS re-tx high profile.
Have found as a general rule of thumb that DS INP of 49 or less usually indicates re-tx low profile where 50 or greater indicates re-tx high profile.
Bearer 0 INPRein (0 or 1) value is still the positive re-tx profile indication.

Quote
Is this something you can do on the HG612?
Yes, procedure for HG612 is outlined here:
https://forum.kitz.co.uk/index.php/topic,16427.msg304299.html#msg304299
Remember, applied speed clamps won't survive a modem reboot.

Quote
Is it worth trying to help it along or is it best just left alone for a few weeks to sort itself out?
I would certainly leave any modem speed clamping until after DLM has completed it's DS SNRM steps to 3dB or whatever is achievable on your line and assuming that G.Inp DS re-tx low profile has not been re-instated along the way.
DLM intervention time scale can vary quite a lot and your DSLAM being Vectoring enabled makes it a bit of a grey area which very few have any experience of, be patient.
Logged
BT FTTP 150/30, BT Smart Hub 2

j0hn

  • Kitizen
  • ****
  • Posts: 4093
Re: Things that go bump in the QLN
« Reply #7 on: July 18, 2021, 07:29:57 PM »

Is this something you can do on the HG612?

I wouldn't do it yet.
If the line is stable the DLM will continue to lower the SNRM down to 3dB.
When it hits 3dB you should apply the cap to get back to retx low.

It's likely that if you cap the line now and the line goes to retx low it could just revert to retx high on the next SNRM change.

There's usually about 2 days between each step (from 6dB to 5dB to 4dB to 3db).
Logged
Talktalk FTTP 550/75 - Speedtest - BQM

toonshorty

  • Member
  • **
  • Posts: 18
Re: Things that go bump in the QLN
« Reply #8 on: July 22, 2021, 10:02:58 AM »

After four days sat at 55.8, we've got some more ambitious movement from the DLM this morning.

The line has resynced at 60120/10099 with the SNRM currently sat at 3.8dB. Interleaving is all the same as before and the IP profile is now 54.7 Mb/s.

Errors are up, with ES/hour currently showing 10.2/4.07 - although it was as high as 30/20 at one point this morning.

Interestingly, the noise around tone 1700 seems to have vanished (for now) as well.



Logged

toonshorty

  • Member
  • **
  • Posts: 18
Re: Things that go bump in the QLN
« Reply #9 on: August 08, 2021, 09:18:16 PM »

So after monitoring over the past couple of weeks, I've noticed some interesting behaviour around bitloading in the 24-48 hours following a resync. Essentially, when the line first syncs, the bitloading looks very similar to my post on 22nd July - that is to say a pretty solid block with maybe a few notches in the D3 band. This is generally maintained for around about 12 hours give or take, at which point some of the tones around that 1700 mark will start to show drops to around about 6 bps. This essentially gets progressively worse over 24-72 hours until it looks more like the first bitloading graph I posted with a whole chunk missing around 1700. It doesn't ever recover either, it's not a case of the tones coming and going; the only way to restore them is via a resync. Does anyone know of anything that might cause this behaviour?

Either way, it seems the power was knocked out last night and I've somehow resynced at 64211/10976 so I shouldn't complain too much, it's improved a good 10Mbps since the line was first provisioned - although the higher retranmission profile is eating into those gains somewhat (at least for now).
Logged

burakkucat

  • Respected
  • Senior Kitizen
  • *
  • Posts: 38300
  • Over the Rainbow Bridge
    • The ELRepo Project
Re: Things that go bump in the QLN
« Reply #10 on: August 08, 2021, 09:31:55 PM »

Does anyone know of anything that might cause this behaviour?

Bit-swapping immediately comes to mind. Are you certain that bit-swapping hasn't been disabled?  :-\
Logged
:cat:  100% Linux and, previously, Unix. Co-founder of the ELRepo Project.

Please consider making a donation to support the running of this site.

toonshorty

  • Member
  • **
  • Posts: 18
Re: Things that go bump in the QLN
« Reply #11 on: August 08, 2021, 10:02:14 PM »

Bit-swapping immediately comes to mind. Are you certain that bit-swapping hasn't been disabled?  :-\

I needed to go and enable the monitoring in DSLStats but it looks as though it is enabled.

You can see a thin blackspot in the bitloading starting to appear (15h 40m DSL update).



Bitswaps per Tone


Bitswaps per Min
Logged

Chrysalis

  • Content Team
  • Addicted Kitizen
  • *
  • Posts: 7382
  • VM Gig1 - AAISP L2TP
Re: Things that go bump in the QLN
« Reply #12 on: August 11, 2021, 10:53:22 PM »

On adsl by design if a bitloading goes down to 0 for a tone, its disabled for the life of the connection, I assume its the same on vdsl.

Disabling bit swapping would stop this, but it would also lose you stability, keep it on. :)
Logged