Kitz Forum
Broadband Related => FTTC and FTTP Issues => Topic started by: Browni on July 15, 2018, 07:50:10 PM
-
Up and running in Rochdale.
The speed is rather disappointing though, hovering around 200-210Mb/s when the estimated speed was 260-290Mb/s :'(
-
Maybe crosstalk on your g.fast?
-
And G.Vector technology is a mandatory part of the standard, no? So it should be good at taking on the crosstalk with its inverse pre-coding.
-
@Weaver, querying the router returns this
BusyBox v1.18.5 (2018-05-28 19:53:14 CST) built-in shell (ash)
Enter 'help' for a list of built-in commands.
/ # bcm_xdslctl info --vectoring
bcm_xdslctl: ADSL driver and PHY status
Status: Showtime
Last Retrain Reason: 2000
Last initialization procedure status: 0
Max: Upstream rate = 35012 Kbps, Downstream rate = 203511 Kbps
Bearer: 0, Upstream rate = 34884 Kbps, Downstream rate = 205805 Kbps
Currently not in VDSL modulation --vectoring is only for VDSL mode
-
Perhaps there is a degree of misunderstanding occurring here . . .
The modem will be performing ITU-T G.993.5 which is vectoring. What the router does is irrelevant.
I presume that you have been provided with an Openreach badged Huawei MT992 modem, to which your router is then connected? :-\
-
I presume that you have been provided with an Openreach badged Huawei MT992 modem, to which your router is then connected? :-\
Yep, I've got one of them but my ASUS DSL-AC88U functions both as a modem & router so no need to connect both.
-
Browni, I just dug out a post of yours back from March:
For reference I am 175m straight line from the cab and a GEA test gives an estimated line length of 283m.
The speeds seem about right for the estimated line length. Not really disappointing at all. ;D Your line length would probably have to be under 200m to see speeds of 260+ Mbps.
-
Hmmm,
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fi64.tinypic.com%2Fruzghi.jpg&hash=8b16b1f1c1ac92ffde1a8804d6205eeeae0ca0cd)
-
Yeah I've seen your estimates before. Perhaps a bit overzealous considering the GEA test's estimated line length. Out of interest, what speeds did you get on FTTC?
What are your CURRENT estimates? As far as I know, the DSL checker should update estimates every now and again to reflect the line's performance.
-
On FTTC I synched at the full 80/20 and the current estimates are still the the same as those in the image.
-
Have any stats from back when you had FTTC? Huawei cabinet? It would have been somewhat possible to sync at full 80/20 with 3 dB profile + G.INP enabled at your estimated distance.
I don't think there are any applications that are able to properly graph G.Fast at the moment. Perhaps you could post the output of the following in a text file so I or someone else can graph it:
bcm_xdslctl info --Bits
bcm_xdslctl info --SNR
bcm_xdslctl info --QLN
bcm_xdslctl info --Hlog
I don't have G.Fast, so I have no idea on what the output should look like. Furthermore, G.Fast uses Time-Division Duplexing (TDD) as opposed to Frequency-Division Duplexing (FDD) so I really have no idea if that would make things more complicated.
-
I certainly didn't need a 3dB profile, the noise margin was usually around 9.5dB
Have a look at http://i65.tinypic.com/9jex03.jpg (I can't seem to embed that using img tags???)
I've attached the other results with appropriate names.
[attachment deleted by admin]
-
4th attachment
[attachment deleted by admin]
-
In reference to when you had FTTC, perhaps there were not that many subscribers on your cabinet creating crosstalk? Might be why you were receiving the full 80/20 with a nice bit of head room. Either way, I would imagine that you are probably 250m from the cabinet or thereabouts, close to the GEA test estimate.
But oh dear. I think I am out of my depth here since I know virtually nothing about G.Fast. I'll perhaps give it a crack later, but I am a little bit mythed because I though the maximum bits per tone were 12, yet I am seeing 14 in the provided file. :shrug2:
-
After a quick glance at the image and files provided, I can't see anything that "says" G.9700/G.9701 (a.k.a. G.Fast) to me. ???
What does the ISP/CP say about using the ASUS device in place of the Openreach provided Huawei MT992 modem?
-
This (http://browni.co.uk/dslstats/Webserver/fullstats.htm) clearly shows G.fast.
Openreach have been out and didn't even bat an eyelid at the ASUS device (which incidentally is one of the router/modems currently undergoing MCT accreditation.)
-
I'm sorry that I can't really help much more. Graphing it probably doesn't help so much either. But here are some graphs anyway.
-
And for SNR...
-
D'oh. :doh: I just remembered that Hlog and QLN figures are static and remain the same from the moment of syncrhonisation. Part of the reason why I was confused about the statistics. Though it seems like there are some oddities with the Hlog here that I have not seen in VDSL. I understand that VDSL uses FDD and G.Fast uses TDD, but perhaps this is the reason to why it seems a bit... weird?
Either way, if I am not mistaken then around 1,700 tones (combined DS and US) with at least 1 bit allocated. I read G.Fast will allow allocation of up to 2048 tones - but HOW? Because I do not know. I am a little confused.
Edited to clarify post.
-
No need to apologise re0, I'm delighted you've shown interest and created the graphs.
Just noticed your latest post, will the figures remain static? I know there's some form of SRA involved on my line because the line synch rate is constantly changing.
-
Hlog and QLN figures are taken at the moment of synchronisation and so after a couple of seconds these figures are no longer up to date (therefore no longer representative of the current line conditions). These won't change throughout the whole duration of the conneciton as they are not monitored after the initial connection link is established. These figures are unlike the Bit Allocation and SNR which are always being monitored by the modem and therefore are always changing as long as changes occur to the conditions (which they always are).
To answer your question more concisely, the figures which are not Hlog and QLN are always changing. And yes, SRA is mandatory on G.Fast so the line speed and SNR will always be adjusted accordingly.
-
After a quick glance at the image and files provided, I can't see anything that "says" G.9700/G.9701 (a.k.a. G.Fast) to me. ???
The image was from when he was on VDSL.
-
Yep, I've got one of them but my ASUS DSL-AC88U functions both as a modem & router so no need to connect both.
Is the sync speed with the Huawei the same?
-
Is the sync speed with the Huawei the same?
It doesn't have a UI but speedtests were within a few meg.
-
I just saw this from a post you made back in May. I was a bit silly to ignore it, because THIS highlights the bandplan of G.Fast used on the OR network.
bcm_xdslctl info --pbParams
bcm_xdslctl: ADSL driver and PHY status
Status: Showtime
Last Retrain Reason: 0
Last initialization procedure status: 0
Max: Upstream rate = 34767 Kbps, Downstream rate = 198057 Kbps
Bearer: 0, Upstream rate = 34651 Kbps, Downstream rate = 198057 Kbps
Discovery Phase (Initial) Band Plan
US: (368,1930)
DS: (368,1930)
Medley Phase (Final) Band Plan
US: (368,1339)
DS: (368,1211)
I imagine those stats are not much different from how they are now. However, this should help me interpret the data you provided (and the graphs). The bit loading and SNR for downstream and upstream occupy different parts of the graphs - it appears that the downstream is on the left and the upstream is on the right. The Medley Phase confirms how many tones your connection is using, while the Discovery Phase is the tone range that is on the Band Plan (so a better connection could technically utilise this range).
At least I am no longer confused with the bit allocation stats. ;D Perhaps I was just being a bit stupid initially. Though not sure how the Hlog should look because of how G.Fast works. ???
-
Browni ditch the asus as a modem and bridge from the openreach supplied device to it, otherwise this is probably my final comment on it, as I am sick of the problems asus modems cause on the openreach network.
What reason are you using the asus device as the modem instead of bridging to it?
-
Some more graphs... this time with labels... and Bit Loading and SNR combined.
-
What reason are you using the asus device as the modem instead of bridging to it?
Eh?
It's a Broadcom based combined modem/router.
Why wouldn't I use it as designed?
-
Browni ditch the asus as a modem and bridge from the openreach supplied device to it, otherwise this is probably my final comment on it, as I am sick of the problems asus modems cause on the openreach network.
What reason are you using the asus device as the modem instead of bridging to it?
Its the AC88U which has a BCM63138 (https://wikidevi.com/wiki/ASUS_DSL-AC88U). Asus appear to have ditched using the MediaTek chipsets which 'let down' some of their former modems. I wouldnt touch the DSL-AC68U etc.... but if I had g.fast I'd definitely give the AC88U a shot. :)
-
@Brownie. Whats your latency like.
I've briefly had a look at some of your telnet stats and can see the SRA params. I also see RS/FEC/INP and re-tx yet supposedly 0 interleave delay.
-
Pretty decent
C:\Users\ian>ping bbc.co.uk
Pinging bbc.co.uk [151.101.128.81] with 32 bytes of data:
Reply from 151.101.128.81: bytes=32 time=8ms TTL=58
Reply from 151.101.128.81: bytes=32 time=8ms TTL=58
Reply from 151.101.128.81: bytes=32 time=8ms TTL=58
Reply from 151.101.128.81: bytes=32 time=7ms TTL=58
Ping statistics for 151.101.128.81:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 7ms, Maximum = 8ms, Average = 7ms
C:\Users\ian>
-
Nice one. :thumbs:
Its usually about 10-11ms from Manchester on FTTC so that's excellent.
-
Eh?
It's a Broadcom based combined modem/router.
Why wouldn't I use it as designed?
Still I would bridge the openreach device instead. If you not even at least willing to try it and compare it then thats it from me sorry.
Its the AC88U which has a BCM63138 (https://wikidevi.com/wiki/ASUS_DSL-AC88U). Asus appear to have ditched using the MediaTek chipsets which 'let down' some of their former modems. I wouldnt touch the DSL-AC68U etc.... but if I had g.fast I'd definitely give the AC88U a shot. :)
I am guessing he wants an all in one solution to save one power cable, and ethernet cable, but I think he should at least try the officially supplied device.
Me personally I actually deliberately separate modem from router so any router firmware updates, maintenance etc doesnt cause modem downtime. So I am the opposite.
-
Still I would bridge the openreach device instead. If you not even at least willing to try it and compare it then thats it from me sorry.
I have tried it with no gain in speed and complete loss of visibility of stats.
-
ok thanks for the information.
-
The image was from when he was on VDSL.
Thank you for the simple explanation. :doh: :hat:
-
Forgive my ignorance, but what modem device do openreach supply for a g.fast connection?
-
A Huawei MT992 (https://forum.kitz.co.uk/index.php/topic,21315.msg369226.html). :)
-
I apologise for picking up an old thread.
G.fast can target 3 dB SNR, though I can see that Browni's connection is not anywhere close to that for the downstream (it likes to hover around 6-8 dB, so 6 is likely the target?). I wonder what is preventing the connection from targeting 3 dB since it would give a decent speed boost. There are a lot of parameters that I've not seen before. :silly:
Browni posted a link to his stats (http://browni.co.uk/dslstats/Webserver/fullstats.htm) on page 2 so I am just linking it here for convenience.
Any thoughts?
-
Possibly DLM? My daughter's connection started off quite good, but experienced rapid deterioration after emergency work to a gas mains (https://forum.kitz.co.uk/index.php/topic,22133.msg379751.html#msg379751) which supposedly damaged a portion of their UG cable. DLM took some very harsh action and quite rapidly took 230Mbps throughput down to a 95Mbps cap and then right down to 24Mbps.
Unfortunately she only has an MT992 so no stats other than the screenshots (https://forum.kitz.co.uk/index.php/topic,22133.msg379718.html#msg379718) taken when Openreach was there last month :(
One of the screenshots shows 3.10dB target SNRM after he did a DLM reset.
-
Well of course the DLM. ;D But I wonder which parameters are making the DLM back off from reducing the margin.
Possibly DLM? My daughter's connection started off quite good, but experienced rapid deterioration after emergency work to a gas mains (https://forum.kitz.co.uk/index.php/topic,22133.msg379751.html#msg379751) which supposedly damaged a portion of their UG cable. DLM took some very harsh action and quite rapidly took 230Mbps throughput down to a 95Mbps cap and then right down to 24Mbps.
Unfortunately she only has an MT992 so no stats other than the screenshots (https://forum.kitz.co.uk/index.php/topic,22133.msg379718.html#msg379718) taken when Openreach was there last month :(
One of the screenshots shows 3.10dB target SNRM after he did a DLM reset.
Yeah, I've seen that thread when frantically seeking information about G.fast speed/distance. :D I guess I'm just excited G.fast is coming. Shocking how much of an impact it had on the connection though!
Anyway, you said her connection was about 200m to the cabinet, though me thinks it's actually equivalent to around 250-300m of 0.4mm. :)
-
DLM has finally dropped the target margin to 3dB giving a ~45Mb/s increase in download synch rate 😃
-
That's pretty nice. Dropping the SNR margin is going to make a much larger difference on G.fast than on VDSL (in reference to the actual figure, of course :D). Hopefully it will stick around since that is a nice increase. :)
-
Hey Browni! What's happened to your downstream? :o
-
@re0, no idea mate!
Unfortunately I was in hospital when all the interesting things happened on my line including a synch rate of 267Mb/s which is within the clean estimate, ( a speed never obtained before.)
It then reverted to the pretty much standard 202Mb/s until this morning when a resynch with a retrain reason of 2000 occurred and a 9Mb/s bump in synch occurred.
Will the speed rise again?
Watch this space!
-
I hope you're feeling better now.
I can't say I have the answer. The SNRM has historically been a bit all over the place but it looks like target is 3 dB at the moment. Perhaps the DLM is business as usual ("DLM maketh, DLM taketh") since maybe there are some parameters responsible for the drop in speed, but I don't know about G.fast and probably won't until I get to play with it myself or gather enough statistics from the forum. I would have thought that the parameters would have not been wholly responsible and perhaps it's due to the fickle nature of interference with the frequencies utilised by G.fast.
By the way, it looks some historical screenshots are not available. Most have a few from September, one from October. Not much else there.
Maybe you have the bitloading and SNR logs from when the sync was at its high point and now? It would be interesting to make a comparison.
I guess I have to wait until the next episode to find out whether the speed increases. ;)
-
There's lot's of data hiding behind the scenes but it's not my software so haven't the foggiest if it can be displayed in a meaningful format, most graphs fail probably due to unknown data ranges.
Does the bitloadingText.zip file attached throw any clues?
-
Perhaps you could chuck us the SNR logs as well so I can graph them together?
-
Is this the one you mean?
-
Apparently so. I'll look at it when I get bored. :D
-
:lol:
-
Graphs based on data from 13:36:16, 2018-09-30.
Strong downstream sync 265467 Kbps. Upstream 36788 Kbps.
-
Graphs based on data from 18:00:03, 2018-10-12.
Lesser downstream sync 202667 Kbps. Upstream 36623 Kbps.
-
That downstream graph is looking very janky both SNR and bitloading. Not sure what could have changed.
Also just a short disclaimer: I cannot be held responsible for my shoddy work if anyone becomes injures or dies as a result of the bits being slightly shifted or inaccurately represented. :P
-
That downstream graph is looking very janky both SNR and bitloading.
Yuck. :yuck:
b*cat waits for a Huawei MT992 to be donated to The Cattery. (Though looking at the images of the PCB I'm not sure where a connection should be made for the serial console.)
-
I feel a bit disappointed that the next person to reply (I'm looking at you, b*cat) didn't joke about my disclaimer and my liability for injury if the data is accurately represented since it was not explicit. I'm sorry... I'm terrible. :lol:
b*cat waits for a Huawei MT992 to be donated to The Cattery. (Though looking at the images of the PCB I'm not sure where a connection should be made for the serial console.)
You're probably going to be waiting a while. :D Anyway, for the majority of the forum users, they're going to want an easy way to retrieve stats. This method will only be for the few. People who want stats will likely opt for the ASUS DSL-AC88U (like Browni) or next equivalent modem or modem-router combo that allows stats retrieval.
Good luck with getting one. I did mention in another topic somewhere that Alibaba has sellers stocking it but it's not a site I've used before.
-
The black model on Alibaba would help initial examination and give an idea as to what needs done. A cheap machine to make mistakes.
No idea what the minimum order value is on the Alibaba listings it's been a couple months. Might be a decent modem for FTTC.
Who's chipping in on a bulk buy? ;D
You would still need to source an OpenReach branded MT992 though.
-
The black model on Alibaba would help initial examination and give an idea as to what needs done. A cheap machine to make mistakes.
No idea what the minimum order value is on the Alibaba listings it's been a couple months. Might be a decent modem for FTTC.
Who's chipping in on a bulk buy? ;D
I would be interested to know what sort of scale in terms of quantity and price per unit you could negotiate. :)
You would still need to source an OpenReach branded MT992 though.
Yes, agreed.
-
No idea what the minimum order value is on the Alibaba listings it's been a couple months. Might be a decent modem for FTTC.
All that I can see have a piece minimum order of 1. It may be possible to negotiate a price as low as 50 USD (but I doubt that in small quantities). Then you'd have to worry about shipping and VAT, and even customs if you intended to buy more than a couple.
-
So roughly 70 times faster than one of my lines, downstream. Another world.