Kitz Forum
Broadband Related => FTTC and FTTP Issues => Topic started by: colin79666 on January 20, 2018, 08:45:04 PM
-
Posted this on the BT community forums but since discovered this site. What an excellent resource! :)
We have had BT infinity 2 (up to 78 Mbps) for nearly 4 years and since the day of connection until a month or two ago any speed test indicated we pretty much got the headline rate. Now seeing 58Mbps down and 14Mbps up. When it was installed the BT engineers gadget showed our line was more than capable for 80/20. Of late however I noticed some activity was a bit slower, especially my backups to Azure. Unfortunately I don't have any line stats to go back to.
From the info found on here I discovered we have a Huawei cabinet but a ECI modem so couldn't get stats. I've never used the BT Home Hub, preferring an Asus RT-AC68U for the past 2 years. I've acquired a HG612 from eBay and had that running a couple of days, installing the unlocked firmware. Since installing it I've had to disconnect the phone line once as the modem seems to run hotter than the ECI and I needed to move it for more ventilation, other than that the connection has been stable (and going back weeks the router logs don't show any disconnections).
I've got a better modem/phone cable on order but currently using standard flat cable the modem came with (don't think it will be twisted). Modem is connected direct to the NTE5 socket. Line is underground from the cabinet to the DP in the street and from the DP to the house. Distance from the house to the cabinet is about 200m as the crow flies but unsure how long a path the line takes. In the house the old master socket wasn't in a good position so the original line was extended by the engineer about 10cm into a new NTE5 master socket also into which there is a phone extension running from upstairs. I've made no changes to extensions or phone handsets since. No noise on the line to be heard when conducting a quiet line test.
With the background out the way I'm hoping someone more experienced than I might be able to shed some light on the possible reasons for the performance to have dropped?
-
probably due to cross talk if the speed has reducded over time (new subscribers). 80/20 is up to so full speed never guaranteed. your sync is actually 62mb so you have lost a bit. Looks like dlm has you on a 3db snr so not likely to gain anymore at the moment.
with attenuation of 16 you are over 200m from the cab.
-
With an attenuation of 16 I would estimate a length of around 350m.
If you are a member of MDWS you can view other user stats. I would say you are probably at the lower range of sync speeds for that db especially with ginp and 3db active. I think this points more to just high take up at your cabinet rather than a fault though but check if you are below the handback speed for your line at the checker if you want to risk reporting a fault to see if BT will pair swap you.
Signing up to mdws and recording stats for a few days may be beneficial as I am sure the more experienced stats experts will give their opinion if they believe there is a fault.
-
Thanks both for the advice. I've signed up for MyDSLWebStats (colin79666) so that might help keep a track although I won't be able to leave my PC running 24/7. Handback is 53.7Mbps so I'm above that threshold for now. Quite happy with current service but worried it might drop further, especially if cause is unknown.
The drop of around 10Mbps for DL appeared to be quite sudden rather than a gradual decline. Is that expected with cross talk caused by more FTTC activations?
-
You tend to get three kinds of crosstalkers, 1) makes little difference (1% loss), 2) makes a small amount of difference (5% loss) and 3) a major disturber (15% loss). Just plucked these three educated guess % figures from thin air just to give you an idea.
You get loads of 1, several 2 and a couple of 3. Crosstalk normally costs you 30-40% on a full cabinet (seems to be the accepted % figure, vectoring really helps offset this but only deployed on a few cabinets).
If you find mdws useful I recommend getting a RPI so you can keep stats running 24/7. I think there is a link somewhere for All User Stats and you get a table of all user data so you can sort on the data like cabinet type and attenuation - very useful.
Hopefully a friendly expert will comment on your qln, hlog and other stats.
-
Thanks for the info. I've got a Raspberry Pi Zero on order so that can be left gathering stats once I get it configured.
Was across the road at a neighbours this afternoon to help with her PC. Did a quick speed test while I was there and she was getting 72 down, 16 up with the HG612 and Home Hub 3 so I'm more convinced than every that my line has been affected by some internal or external factor.
-
I've taken a quick look, via MDWS, at your Hlog and QLN plots.
We can see from the Hlog plot that the US2 and DS3 bands show some slight, abnormal, undulations. I am unable to say what could be the cause of that effect.
The QLN plot also shows signs of abnormality in the US2 and DS3 bands. That could be the result of one (or more) cross-talkers coupling into your circuit.
Overall, the circuit appears to be quite good and I would suggest that you examine the local wiring within your premises to ensure that best principles prevail.
-
As noted by bcat definitely ensure you are testing from the master socket to ensure your local wiring is having no impact.
Also worth mentioning trying a different modem may give positive results either speed or much lesser errors wise. Can be expensive as may make no difference so unless you are looking to change to a combined unit or using your current as a spare you need to have accommodating friends.
I use the decent Zyxel 8324 can't remember if its the A or D model (and it does make a difference). I think there is a thread mentioning newer models (1312?) that has the same chipset though - again be careful you get the correct variant though. Definitly confirm first as I am not 100% sure on model numbers (someone may post details?).
-
Thanks again.
Performance dropped with the ECI modem and appears much the same with the HG612. Unfortunately I don't have another VDSL modem to try but I guess having tried two with different chipsets that probably rules out a failed modem.
I haven't made any other changes internally but will disconnect the phones (including extension) tomorrow along with the NAS that sits next to the modem and see if there is any improvement.
-
I wasn't really trying to say a dodgy modem more the original HG612 & ECI modems are very old and use old chipsets, a newer modem with the correct newer chipset can make a difference. Also can make no difference.
Not sure what else to suggest apart from let your stats gather a bit more and ask if someone else is happy to give a 2nd opinion.
I still think you are certainly a bit unlucky not to be getting a better sync speed especially on a huewai cabinet with ginp and 3db at your line attenuation.
-
Raspberry Pi Zero should be turning up today or tomorrow so I'll be able to get stats over a bigger timeframe soon. That might help establish a pattern or at least give me something to wave in-front of BT should the sync drop again.
Took the NTE5 apart this morning and tried the HG612 straight into the test socket. Also powered off both DECT phones in the house. Got a slight improvement of sync rate and slightly less of a drop in the SNR around tones 400-500 but nothing significant so that can probably be put down to time of day/weather/random chance.
I should have my hands on a Zyxel VMG1312-B10D by the weekend. I don't expect it will magically get the profile back to where it was but as you say with the more modern Broadcom chip it might improve things a little and probably won't make it worse anyway. Apparently it supports vectoring so if my issues are indeed crosstalk and BT enable the functionality at my cabinet then I will have the equipment to take advantage of it.
Thanks for your help :)
-
The main thing I value my Zyxel for is it keeps the ES very low for my line and as I am on an ECI cab that is good news as it has no ginp.
I cross fingers your new modem syncs faster. The better chipset (?) will hopefully play better with the dlm so it should be positive. Even if not immediate sync increase less errors may allow the DLM to adjust power and bit loading over a few days?
Also good you will have before and after mdws. Be careful I am not sure how long stats are kept if you do not donate though!
-
Hmm connected up the ZyXEL VMG1312-B10D yesterday around 1700. Had a few issues getting the RPi reconfigured so it wasn’t monitoring quite straight away.
Stats don’t paint a pretty pictures. Errors (of all kinds) have shot up!
Now the question this raises is the ZyXEL just bad for my particular line or was the HG612 not reporting the truth? I’ve afraid the leave things as they are as DLM may well kick in with this.
-
Have you gained any sync speed? If your FEC errors have shot up I think that is fine as these are automatic error corrections. Work out your estimated ES/hr is it reasonably within the amber range for a 24h period?
I think someone who knows the stats better than me is going to have to suggest next steps?
You can look at the forum yourself for others singing praises of the Zyxel modems (did you check the D variant was the better one?) so you must be very unlucky for it to be worse than a hg612!
-
Slightly lost sync speed :( E/S hour have gone well into the red (almost zero recorded on the HG612)
I've actually got two ZyXEL VMG1312-B10Ds and have tried both. Each has much the same stats after an hour of running so I think I'm perhaps just unlucky that they don't agree with my particular line. I've seen plenty positive comments from others about their performance although one reviewer on Amazon did comment that the B10D did give them more errors than their HG612.
-
If I was in your position I would put the hg612 back if i was getting that level of errors.
Not sure what to suggest other than seeing if anyone posts with a similar situation as you and how they solved?
I would be very surprised if hg612 was the best solution though.
Can anyone help me (as I think I am at my limit of knowledge) and obviously the the op by posting some make and models of modems that should make a difference and why so he has options if he wants to pursue further research?
-
Went back to the HG612 just after my last post - sync up and CRC errors back to near zero.
Looking back at the stats it appears something off happened yesterday afternoon (before I switched to the ZyXEL) where the attainable speed jumped back to probably where it was before all my troubles started towards the end of last year. I'm wondering if a cross talker disconnected for a period or something. If I spot it again while it is happening I'll run down to the cabinet and see if an engineer is in there tinkering!
-
Went back to the HG612 just after my last post - sync up and CRC errors back to near zero.
Looking back at the stats it appears something off happened yesterday afternoon (before I switched to the ZyXEL) where the attainable speed jumped back to probably where it was before all my troubles started towards the end of last year. I'm wondering if a cross talker disconnected for a period or something. If I spot it again while it is happening I'll run down to the cabinet and see if an engineer is in there tinkering!
Running down the road in your dressing gown and slippers! Will be a strange site indeed! ;) :D
-
FWIW I am also on an ECI cabinet, and my usual router is a Lantiq based Draytek Vigor 2760, which seems to suit my line perfectly. Over the last few months I've tried a Zyxel VMG1312-B10A, a Zyxel VMG8924-B10A and a Technicolor DGA4130, all broadcom based of course. They have all produced a much higher error level than the Draytek, with only a modest increase in sync.
-
FWIW I am also on an ECI cabinet, and my usual router is a Lantiq based Draytek Vigor 2760, which seems to suit my line perfectly. Over the last few months I've tried a Zyxel VMG1312-B10A, a Zyxel VMG8924-B10A and a Technicolor DGA4130, all broadcom based of course. They have all produced a much higher error level than the Draytek, with only a modest increase in sync.
It does seem to be a case of suck-it-and-see rather than there being hard and fast rules. I'm also on an ECI cabinet, getting a (banded) 14999 downstream sync with a Broadcom-based Billion 8900AX-2400. I tried a Draytek 2760 and it synced at 11399, which is a huge drop. I didn't leave it connected long enough to compare error rates!
-
Well this just gets more odd by the hour. Logged into see the stats this evening and it looks like not long after the HG612 went back in the attainable rate climbed up from being stuck around 61Mbps to 69Mbps DS. I risked another resync and sure enough the HG612 is now synced at that rate! Not too far off
Perhaps all the errors with the ZyXEL has unstuck the DLM on my line?
-
I tried a Draytek 2760 and it synced at 11399, which is a huge drop. I didn't leave it connected long enough to compare error rates!
11399 (11.4Mb) is 1 of the banding profiles. The Draytek might be worth another try if you still have it as it would likely sync higher now your banding has increased.
I've actually got two ZyXEL VMG1312-B10Ds and have tried both. Each has much the same stats after an hour of running so I think I'm perhaps just unlucky that they don't agree with my particular line. I've seen plenty positive comments from others about their performance although one reviewer on Amazon did comment that the B10D did give them more errors than their HG612.
Any reason you bought 2 x VMG1312-B10Ds.
The VMG1312-B10A has a better chipset and should sync higher.
The Zyxel VMG1312-B10A is the best performing modem I've used, on both sync rate and errors.
It can have IPV6 issues if being used as a router so I'd only recommend it to anyone who uses a modem in bridge mode with a separate router.
-
Any reason you bought 2 x VMG1312-B10Ds.
The VMG1312-B10A has a better chipset and should sync higher.
The Zyxel VMG1312-B10A is the best performing modem I've used, on both sync rate and errors.
It can have IPV6 issues of being used as a router so is only recommend it to anyone who uses a modem in bridge mode with a separate router.
Simply was part of a job lot so ended up with 2 for the price of about 1. I know the VMG1312-B10A has the better chipset but figured the D would at least be newer than in the HG612. AA and Zen are both using the VMG1312-B10D so figured if it was good enough for them then it would be worth a shot. The VMG1312-B10A is considerably more expensive at the moment on the most well known auction site but I'll keep an eye out as they occasionally pop up for a bargain.
-
@colin79666
From your MDWS stat's, note that your 4 re-synch's today were "LOS" as was yesterday's, looks like the last DLM instigated re-synch was on 23/01, recorded as "NEG", the usual DLM instigated re-synch indication is "RDI".
From your posts, you have changed out your router very frequently over the last few days which would account for the "LOS" re-synch's (manual power down of your router)
As you most likely know, frequent re-synch's for whatever reason will eventually annoy DLM and should be avoided if possible.
Do have a read at the excellent DLM information section of the kitz site which explains how manual re-synch impact on DLM can be minimised, very informative.
Your DS ES/CRC rates are abnormally high especially considering that DS G.INP is active, currently well into ILQ red, I would have thought that DLM action is imminent soon. (usually actioned after midnight unless very severe, early hours of the morning after DLM system has made it's assesments for the previous day's line performance)
As you have substituted a number of different routers on your line I would have thought it's now looking unlikely that a router is the underlying cause of your noise issue.
Is it time to get your ISP to run a line check, hoping that the noise issue is not an intermittant fault on your line, obviously ensuring to the best of your ability that you don't have any internal wiring/equipment faults as per your earlier actions running the modem directly from the master socket.
In the mean time, I would be inclined to leave one router in service, your choice, no more manual re-synch's and give DLM a chance to take remedial action on the line which it certainly should do in it's current condition.
-
Thanks tiffy for your input.
Yes I'm certainly going to leave well alone over the next few days to let things settle down. :fingers: that DLM doesn't kick in and clobber the line overnight as it is looking good just now. The errors only became a thing with the ZyXEL but the change of attainable rate over the last 24/48 hours is a mystery.
I'm perfectly happy with the current sync rate (about as good as possible at my distance to the cab) but should it drop off again this time I will have stats to fall back on so will be getting onto BT to complain.
-
@colin79666
I'am certainly no expert and very much still on a steep learning curve, but, in my opinion although you are happy with your line's current DS synch rate the MDWS DS ES rate is certainly abnormally high @ 5892, well into ILQ red status, with that and 4 retrains in 24 hours I would be surprised if DLM action did not take place tomorrow, will be interesting to see exactly what that action is, hopefully not line banding.
There has to be some reason for the very high (possibly intermittant) DS ES/CRC error rate which I am inclined to believe is not directly associated with your routers, could of course be wrong !
-
11399 (11.4Mb) is 1 of the banding profiles. The Draytek might be worth another try if you still have it as it would likely sync higher now your banding has increased.
When I tried the Draytek my banding was already at 14999 with the Billion - I literally connected the Draytek, got the 11399 sync, then switched straight back to the Billion again at 14999 (and returned the Draytek to Amazon!).
-
@colin79666
From your current MDWS router stats:
As predicted and totally to expectation, you had a DLM (RDI) intervention in the early hours of this morning, main observations:
DS synch rate has gone back to virtually where it was before the last re-synch.
US synch has remained at max attainable.
DS SNRM has remained virtually the same.
US SNRM has been increased considerably.
G.Inp retx profile has been changed from low to high (see B0 INP Rein parameter, changed from 0 to 1)
DS interleaving has remained the same but INP has increased.
Changes mostly to expectation considering the previous very high DS ES rate, DLM trying to stabilise the line.
Not sure on the substantial increase in US SNRM but would not worry as you are at max US synch rate anyway.
Also be aware that your DS data rate will now be lower by virtue of lower synch rate and the G.Inp retx profile change, data throughput ration approx. 91% on retx high and approx.97% on retx low.
If this was my line I would be patience, leave everything as it is, no more manual re-synch's or router changes and wait to see what DLM's next move is.
Important to keep the stat's uploading 24/7 to MDWS to trend any changes that may/will occur, if it should turn out that there is an intermittant line fault it will show up, if you keep changing routers and manually re-synch'ing it will be difficult to tell the difference.
-
Thanks tiffy, yes got the alert email this morning to notify be of the somewhat expect DLM instigated resync.
Odd that the US rate has remained higher than it was for previous week, probably related to the big SNR changes that have happened since switching back to the HG612.
All the CRC errors were when the ZyXEL was connected so I will leave the HG612 alone as you suggest and see how DLM behaves when things are more stable. I know switching them around won't have done any favours but leaving the ZyXEL in place was not an option and with all the errors would probably have cause a bigger DLM reaction.
-
Odd that the US rate has remained higher than it was for previous week, probably related to the big SNR changes that have happened since switching back to the HG612.
My line G.Inp retx profile has changed quite a few times low/high, one effect I have noticed that on retx high profile (as your line is now) my DS synch is higher but DS data throughput lower due to the lower data throughput ratio associated with the retx high profile.
Usually works out what I gain on one (DS synch) I loose on the other (DS data TP) with any low/high retx profile change.
However, I would imagine the retx high profile should be more effective at dealing with DS noise sources.
Check your present BTw BRAS IP profile from the BTw WS Performance Checker (further diagnostics) or use the mouselike.org checker.
DS data throughput ration can then be calculated:
BRAS IP profile / DS synch speed X 100 = TP ratio %
Expected approx. ratios for retx high/low profiles as per previous post.
Apologies if this is obvious or you are already aware, I certainly had to ask not that long ago.
Were both of your ZyXEL 1312-B10D's suspect "noisy" or just one unit ?
-
Thanks tiffy, useful details as always :)
Both ZyXEL VMG1312-B10Ds had the issue. First one was on through the night, the second just for the final hour before I switched back to the HG612. They both came from an eBay job lot but one was like new in box so I'd be surprised if they are both duds. Perhaps I'm just unlucky that the combination of that specific router and my line aren't a good fit.
The current Downstream BRAS rate is: 67 Mbps
The current Upstream BRAS rate is: 20 Mbps
Sync is 61Mbps DS.
-
@colin79666
Yes, I find it hard to believe that both your ZyXEL 1312-B10D's are faulty, there was so much going on with manual re-synch's and modem swapping over a short period of time that it's very easy to compound any external or internal line issue that may or may not have existed at that time.
You can always try the 1312's on your line again at a later date under controlled conditions after everything stabilises and DLM has had a reasonable time to settle down.
From MDWS, your line is still looking OK, note that US SNRM has taken quite a substantial step down towards where it originally was and your US & DS ES rates are very low / zero.
Your DS BRAS IP profile is confusing me somewhat, I would expect this to be lower that your DS synch rate !
For example my line current state: (on a 40/10 provision)
DS BRAS IP profile = 38.67
DS Synch Rate = 40.00
Data Throughput Ratio = 38.67/40.00 X 100 = 96.67% (to expectation as currently on retx low profile)
I have never seen DS BRAS IP profile greater than DS synch rate on my line ?
Strangely enough (to me anyway) apply your line DS BRAS IP / Synch figures the other way around:
DS Synch / BRAS IP X 100, 61/67 X 100 = 91.04% (to expectation for retx high profile)
I'am confused, perhaps we need input from some of the more knowledgable forum members on this ?
-
Bit of an update on this one.
Things continued along the same lines with uptime of around 2 months between resyncs until about 3 weeks ago when suddenly DLM kicked in overnight and the sync speeds jumped back up to what I had been experiencing for the first 5 or so years of VDSL. I've had one more resync overnight since then but otherwise stable. I did spot an OR engineer at the cabinet the day before this happened so perhaps a cross talker has been moved from next to my line.
I note that the downstream interleave depth is now 16 but is is staying synced despite the below target downstream 3db? SNR margin. Some CRC errors on upstream now and then but nothing flagged as a serious error.
xdslcmd info --stats
xdslcmd: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 1
Last initialization procedure status: 0
Max: Upstream rate = 20638 Kbps, Downstream rate = 90148 Kbps
Bearer: 0, Upstream rate = 20000 Kbps, Downstream rate = 79999 Kbps
Bearer: 1, Upstream rate = 0 Kbps, Downstream rate = 0 Kbps
Link Power State: L0
Mode: VDSL2 Annex B
VDSL2 Profile: Profile 17a
TPS-TC: PTM Mode(0x0)
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 2.2 1.5
Attn(dB): 16.6 0.0
Pwr(dBm): 12.4 7.0
All this has been on the Huawei OR modem which I think I'll stick to given the issues I had previously with the ZyXEL (although I don't think the ZyXEL itself is faulty, just didn't suit my particular line conditions).
-
xdslcmd info --vectoring
might explain it? or could indeed be a big crosstalker.
-
xdslcmd info --vectoring
might explain it? or could indeed be a big crosstalker.
Vectoring state 5 so I guess not configured.
QLN in DSLStats from today attached along with one from MyDSLWebStats in January for comparison.
-
You need to click on "change tone range" on DslStats and expand to show all the tones up to 4000.
At the moment you are cutting them off at tone 800.
-
Here we go :-[