Hey kitz users! Long time lurker first time poster.
Just thought I'd give a summary of how my new FTTC line has gone from install to now 3 months later as I've found myself searching for other peoples experience when wondering when features like G.INP or vectoring might be enabled, deciding on new modems etc and what to expect out of a long-ish noisy line.
Originally had ADSL2+ on a long (~2.2km) line but capped at 8mbps as from not changing packages over the course of the last 10 years as ISP was bought again and again, eventually with talktalk business. Stats suggested a max rate of 12ish mbps. Have had a FTTC enabled cab for a few years now, but BT checker always gave "This PCP has a waiting list" so I guess it was at capacity, they stood a new larger cab last year and after checking semi-regularly it became available with estimated 'clean' range of 16.1 to 22.4 mbps down and 0.9 to 1.4mpbs up. By road we are only around 800m from the cab and the line largely follows them, so this estimate seemed conservative. Worth a ago, so ordered from talktalk business with a minimum guaranteed speed of 14mbps.
Line installed on the 2nd of August with a disappointing 13 down 1.2 up with the TT supplied HG633. Over the course of a couple of weeks it became apparent that noise spikes were causing resyncs to lower speeds, some times to as bad as 6mbps and causing various amounts of interleaving to be applied. After a while using an AM radio and walking around the street during the noise bursts found a SMPS in a neighbours house was the culprit. Even when just connected without a load it put out serious RF, and when the pc monitor it was powering was switched on enough to cause a connection drop. Removing this combined with a shorter twisted RJ11 cable and the HG633 got up to 16mbps with interleaving enabled only on the downstream with 8ms delay.
A better modem with proper stats reporting seemed logical now so an old openreach HG612 was purchased, and the new ability to graph SNR/FEC errors led to finding another (far less) noisy PSU in a computer in the house. The broadcom chipset modem combined with less noise in the house got us to 18mpbs, nearly 19 if the perfect time of day to resync was picked.
The cab turned out to be a Huawei (is this the case for all new cabs now?) so cue expectant waiting for G.INP to be enabled. After waiting over a month G.INP was enabled on the 22nd of September. So ~ 1.5 months since install date. I will add that a day or so before it was enabled I was messing around in the HG612s settings and found that at some point previously I had deselected all but VDSL2 from the supported protocals options section, since G.998.4 wasnt another protocol option and PhyRe was enabled I'm not sure if its just coincidence that I changed this the day before G.INP was enabled, but nevertheless I changed this to the 'all' checkbox, make of it what you will.
G.INP yeilded a decent increase to around 20mbps and of course no interleaving delay, so pings to google were now 9ms. Pretty good!
After noticing the graph of SNR per tone looked different and clicking around in DSLStats for a while it turned out vectoring had also been enabled.
https://imgur.com/a/bcilqThese are from my own scritps so the x-axis label is frequency rather than tone number. Ignore the ripple centred around 3mhz in the before graph, its an intermittent noise that happens some nights that I have yet to find the source of.
It seemed that vectoring had little to no effect on sync speed as the increase that occurred at the time it was enabled seemed to be solely the result of G.INP, as the max attainable had been around 21mbps all the time it wasnt enabled and so lined up with G.INP removing the interleaving overhead as I understand it, please correct me if I'm wrong. More so it seemed if anything to decrease the upstream rate from around 1200kbps to 1000kbps so I began wondering how well the HG612 supported vectoring. Much searching of forums yielded no firm answer to whether the hg612 supported vectoring properly other than one post by someone saying that they tried both one and a draytek modem on a vectored line and sync was 10mbps or so lower with the hg612 (cant find the post right now). So I began looking for a modem that 100% supported it, more on that later.
To complicate matters now, the enabling of G.INP & vectoring was accompanied by the start of the gradual change to 3db target SNR margin. So over the course of the next 10 days it dropped by 1db every ~3 days:
https://imgur.com/a/ro2taThis got the line to around 23mbps sync speed.
So to come back to how well vectoring was working on the HG612. Although I guess there are probably lots of reasons vectoring might have little effect on my line (little crosstalk to begin with, the length and noise etc) I still wondered if maybe the HG612 was to blame. I'm not sure of other people experience (and I could find little information from google) but since vectoring was enabled the HG612 reported SNR per tone became odd (both in DSLStats and own scripts). My line uses the U1 & U2 upstream bands and only the D1 downstream. As you can see from the previous graphs it was now showing no SNR per tone from the U2 band at all and showing a weirdly thin U1 band. The pbparams result from CLI showed after the 'discovery' phase the modem had selected the full U1 & D1 bands as before (7,32 & 33,859) but now a larger U2 band, around (871, 980) before it had been (871, 910). Despite this increase the bitloading graph and sync rate showed if anything a decrease in upstream rate as mentioned before. This made me question the modems vectoring support, so I looked for a modem with a newer chipset that might support it better, in the end choosing a Zyxel VMG1312-B10A.
First sync with the new modem offered a pleasing bump to 25mbps from 23 down, and a return to the original (but still disappointing) 1.2mbps up. I'm not sure if this initial increase is just down to the newer BCM63168 over BCM6368 with newer firmware and maybe better input circuitry on the zyxel or due to vectoring. The second sync 2 days later to my shock the modem started using the D2 band with a sync of 28mbps and attainable of 29, this felt like confirmation that vectoring was at least better supported by the zyxel.
However vectoring appears to be a cruel mistress. Some time the next morning after the 28mpbs sync the modem dropped the connection by itself and resynced at 13mbps... A manual resync later in the day returned it to 25mbps, but since this occurred every manual resyc since, even after several ~5-7 day uptimes has resulted in this strangely lower (seems like exactly half?) speed, with a second one later in the day getting back to 25-26mbps. Worth noting that there is no increase in errors or low SNR before or during these manual resyncs. Examining the pbparams output on the CLI shows when this weird sync happens, the discovery phase has resulted in all 3 downstream bands being selected, but no bits being put into any of the 2 higher frequency ones and a much reduced loading on the D1 band. The same odd SNR per tone graphs with missing upstream sections persist across both broadcom chipset modems so I have to assume its a firmware quirk with vectoring enabled.
So thats where I am at now, apologies for the rambling essay of a post but I felt since I've read so much information from kitz it would be worth detailing my experience with a new line in case anyone found it useful.
So has anyone had similar experiences with vectoring? Anyone with a HG612 or VGM1312 with proper SNR per tone graphs?
Has vectoring started to roll across most of the network now? I have to say I was surprised when I found it enabled.
Any clues as to what characteristics of a line result in my ~26mbps downstream with only 1.2mpbs up? From looking through MDWS users it seems common for 25 or so down to have at least 3-4 up. I have thought maybe I was wrong not to ask for an engineer back when the initial speed was so low, but I'm not sure what good it would of done unless there is some fault causing a lower speed other than the poor modem and local noise sources I had then. Does 26/1.2 for ~800-900m line seem low, given 3db + G.INP + vectoring?
Thank you to anyone still reading!
Edit: TL;DR
Approx 800m from cab, 2nd of August live date 13mpbs/1.2mpbs. Fixed noise sources and changed to HG612 modem, 18mpbs. 22nd of September G.INP, 3db target & vectoring all enabled. Vectoring seems to have no effect with HG612 so change to a zyxel VMG1312-B10A. After getting to 3db target + new modem = 25mpbs/1.2mbps. Vectoring on zyxel results in one sync as high as 28mpbs but since then manual resyncs result in temporary 14mpbs. Line stable for weeks at a time at 26mbps now, but still unsure of some fault on the line or problem with vectoring.
Author
Topic: New FTTC install - G.INP, 3db margin & vectoring (Read 5282 times)