VDSL Performance & Tuning

Forums › New Zealand Broadband › VDSL Performance & Tuning

1 | ... | 156 | 157 | 158 | 159 | 160 | 161 | 162 | 163 | 164 | 165 | 166 | ... | 285

Webhead
3247 posts

Uber Geek

Moderator

ID Verified

Trusted

Lifetime subscriber

#907891 4-Oct-2013 14:15

I am wondering if my results are good or bad?

There is a master filter installed and I guess not much I can do - other than perhaps move the Fritzbox to where the switch box is in the apartment instead of being connected to one of the outlets.

Jarle Dahl Bergersen | Referral Links: Want $50 off when you join Octopus Energy? Use this referral code
Are you happy with what you get from Geekzone? Please consider supporting us by making a donation or subscribing.

Samsung

Shop now on Samsung (affiliate link).

hio77

'That VDSL Cat'
12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

Subscriber

#907894 4-Oct-2013 14:21

not bad, thoses spikes of errors are pretty impressive though.. little jab of RFI in the spectrum too.

you are holding DLM-1 with a pretty low error rate though, and being under 8db attenuation, in good running for a chance at a 17a profile. (where you could go up to the full 70mbit down)

if your up for a spot of tinkering, ild highly suggest downgrading your fritz to the .22 firmware and see if it doesn't make an improvement for ya.. theres some really odd stuff the 51 firmware does, alot with a few incompatibilities..

other than that, your looking pretty good there..

about moving to the switch box, what sort of cable is there from there to your jack? quality of cable? etc.. could always try it, see what your spectrum and stats are looking like then evaluate if its an improvement..

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

stevehodge

393 posts

Ultimate Geek

#907901 4-Oct-2013 14:32

hio77: not bad, thoses spikes of errors are pretty impressive though.. little jab of RFI in the spectrum too.

you are holding DLM-1 with a pretty low error rate though, and being under 8db attenuation, in good running for a chance at a 17a profile. (where you could go up to the full 70mbit down)

if your up for a spot of tinkering, ild highly suggest downgrading your fritz to the .22 firmware and see if it doesn't make an improvement for ya.. theres some really odd stuff the 51 firmware does, alot with a few incompatibilities..

I agree, I think it's worth trying the older firmware. The error rate is ok but the downstream sync seems low to me for DLM-1 with 6 db and I think the SNR on the spectrum tab is also a little low across the board. It might be worth looking at some of the earlier posts in this thread to see how the spectrum compares other connections with similar attenuation.

mikenzb

371 posts

Ultimate Geek

#907931 4-Oct-2013 15:22

So with my line running at 0 line attenuation whats a good SNR to drop it down to, 6db? 3db?

hio77

'That VDSL Cat'
12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

Subscriber

#907932 4-Oct-2013 15:24

mikenzb: So with my line running at 0 line attenuation whats a good SNR to drop it down to, 6db? 3db?

see how your errors go i susppose..

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

stevehodge

393 posts

Ultimate Geek

#908015 4-Oct-2013 17:16

mikenzb: So with my line running at 0 line attenuation whats a good SNR to drop it down to, 6db? 3db?

You can only get 4500 kbit/s extra before you hit the DSLAM's limit, and the rule of thumb I use is 1000 kbit/s per db so I'd try 5 or 6db. If there error rate looks too high drop it to 3db.

mikenzb

371 posts

Ultimate Geek

#908254 5-Oct-2013 02:03

stevehodge:
mikenzb: So with my line running at 0 line attenuation whats a good SNR to drop it down to, 6db? 3db?

You can only get 4500 kbit/s extra before you hit the DSLAM's limit, and the rule of thumb I use is 1000 kbit/s per db so I'd try 5 or 6db. If there error rate looks too high drop it to 3db.

I'll get on that when I get free time. I think u meant drop it to 9db?

also with this I can also use the same tweaking method on the first page as this is not.51 firmware? correct?

off topic: when u say hitting the DSLAM's limit you meant they cant go up to 50mbits even with tweaking (useless 17a?)

hio77

'That VDSL Cat'
12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

Subscriber

#908314 5-Oct-2013 09:13

mikenzb:
stevehodge:
mikenzb: So with my line running at 0 line attenuation whats a good SNR to drop it down to, 6db? 3db?

You can only get 4500 kbit/s extra before you hit the DSLAM's limit, and the rule of thumb I use is 1000 kbit/s per db so I'd try 5 or 6db. If there error rate looks too high drop it to 3db.

I'll get on that when I get free time. I think u meant drop it to 9db?

also with this I can also use the same tweaking method on the first page as this is not.51 firmware? correct?

off topic: when u say hitting the DSLAM's limit you meant they cant go up to 50mbits even with tweaking (useless 17a?)

50mbit is about the limit you will see on 8b due to cable loss.

when a tech connects directly to the cabinet at 8b, they see about 60~65mbit so im told with all the test that have been done on my line.

not sure what effect tweaking has on lines ability to hit 17a however.

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

hio77

'That VDSL Cat'
12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

Subscriber

#908369 5-Oct-2013 10:53

so, finally got cat5 from the ETP to our jack.

unfortunately, no improvement, so it is indeed the 049 that is having issues (i was hoping that was just an excessive bit being replaced)

apparently the team that deals with 049 cables dont work weekends, so monday, hopefully ill have a connection that i can start using. at that point ill be pressing for my DLM reset..

alas, the wait continues.

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

stevehodge

393 posts

Ultimate Geek

#908490 5-Oct-2013 15:40

mikenzb:
stevehodge:
mikenzb: So with my line running at 0 line attenuation whats a good SNR to drop it down to, 6db? 3db?

You can only get 4500 kbit/s extra before you hit the DSLAM's limit, and the rule of thumb I use is 1000 kbit/s per db so I'd try 5 or 6db. If there error rate looks too high drop it to 3db.

I'll get on that when I get free time. I think u meant drop it to 9db?

also with this I can also use the same tweaking method on the first page as this is not.51 firmware? correct?

Yeah, I was talking to terms of the drop in SNRM rather than the final SNRM (because you tune by specifying the change in SNRM target rather than the target itself). So you could try SNRM of 6 or 7db and if the error rate is too high then go to 9db. If you're on .22 then you have two options: you can either use the "Line Settings" tab (moving the "Intended SNR" to the left reduces the SNRM target by 1db per position - all the way to the left gives you a target of 8db), or you can export the settings file, edit it, then import it as described in the first post. This method gives you 0.1db precision and you can modify the target up or down as much as you want.

off topic: when u say hitting the DSLAM's limit you meant they cant go up to 50mbits even with tweaking (useless 17a?)

Actually I misspoke a little there. I'm on 17a which has a hard limit of 70000 kbit/s enforced by the DSLAM. On 8b the theoretical limit is 50 Mbit/s but it is possible to exceed that slightly. I seem to remember seeing up to about 53000 kbit/s. So 17a is definitely worthwhile - it'll give you approximately 20000 kbit/s extra speed, and at least for me, the error rate seems to be lower. I have had more success tweaking under 17a than under 8b, though back when I was on 8b we thought the CRC error limit to stay on DLM-1 was much lower than it seems to be now. I've been running a SNRM target of 8db for a few months now, which gives me the maximum downstream sync of 70000 kbit/s. When I first put that tweak on I was still only getting about 0.3 CRC/min, since the last resync the CRC has been about 2 CRC/min - that's a good example of the difference you can see from sync to sync just depending on exactly what the conditions were when the sync happened.

I have no idea if tweaking has any effect on chances of getting onto 17a.

insane

3205 posts

Uber Geek

ID Verified

Trusted

#908495 5-Oct-2013 15:46

Do some cabinets simply not do 17a?

My current session has been up for 50 days now and I'm still on DLM1 8b. I'm getting somewhere around 1.5 CRC errors / hour but no 17a love.

sidefx

3701 posts

Uber Geek

Trusted

#908499 5-Oct-2013 15:52

stevehodge: If you're on .22 then you have two options: you can either use the "Line Settings" tab (moving the "Intended SNR" to the left reduces the SNRM target by 1db per position - all the way to the left gives you a target of 8db), or you can export the settings file, edit it, then import it as described in the first post. This method gives you 0.1db precision and you can modify the target up or down as much as you want.

I think he'll need to use the file edit method. Moving intended SNR to the left actually increases Target SNRM (left is labelled "maximum stability") hence I think far left actually gives you a target of 16db. Dropping it via the file edit method actually moves it to the "right" on the Line settings tab, so when you drop below the 12db default it actually shows no radio buttons selected:

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

stevehodge

393 posts

Ultimate Geek

#908502 5-Oct-2013 16:05

sidefx:
stevehodge: If you're on .22 then you have two options: you can either use the "Line Settings" tab (moving the "Intended SNR" to the left reduces the SNRM target by 1db per position - all the way to the left gives you a target of 8db), or you can export the settings file, edit it, then import it as described in the first post. This method gives you 0.1db precision and you can modify the target up or down as much as you want.

I think he'll need to use the file edit method. Moving intended SNR to the left actually increases Target SNRM (left is labelled "maximum stability") hence I think far left actually gives you a target of 16db. Dropping it via the file edit method actually moves it to the "right" on the Line settings tab, so when you drop below the 12db default it actually shows no radio buttons selected:

Oh yeah, you're right. Dohhh!

stevehodge

393 posts

Ultimate Geek

#908504 5-Oct-2013 16:09

insane: Do some cabinets simply not do 17a?

My current session has been up for 50 days now and I'm still on DLM1 8b. I'm getting somewhere around 1.5 CRC errors / hour but no 17a love.

I can't see why some cabinets would and some wouldn't, but who knows? My cabinet actually predates that big cabinet rollout: it's not even a whisper cabinet.

We don't really know if CRC errors matter for getting on to 17a. I suspect they don't: my error rates were never outstanding, and it makes some sense that they might want to put people on 17a if they have high crosstalk in the 8b spectrum (which should cause more errors). I think attenuation is the most important factor.

hio77

'That VDSL Cat'
12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

Subscriber

#909457 7-Oct-2013 18:45

so, progress is progress..

somewhere along the lines, my dsl interface seems to have died (i suspect when the techs kept yanking the power and playing with the line voltage)

so, big thank you to dan, who leant me a fritz to test out! confirmed, dsl interface is shot and its on its courier to arrive tomorrow morning at snap hopefully.

one thing i do note, is although the connection was only up for 12 mins, so drawing conclusions are is a little far fetched, CRC rate does seem a tiny bit excessive in my opinion atlest.. but thats another kettle to take on..

dsl stats, with dans fritz

spectrum, i had to take 2 shots of this, as the SNR graph kept changing.. im unsure if its the fritz's fault or the line at this stage, but i suspect it may be the next thing to dig into after i have a working connection again..

now, i know atlest in my experience, the snr graph going like that is a bit of a loading bug it would seem, but that was every 3~4 seconds shooting down then up (should have grabbed a screen recorder to show it while i had the connection up..)

so, going back to my fritz for one last test before i condemned it..

these pictures are within the first min of the sync before it died.

so, i tried considering it maybe a miss config on my side.. took the working config from dans 3790 and put it on my 3790 as a last resort test..

unfortunately, this was also a no go..

so, now to wait on what snap has tosay with their tests..

sidefx:
I think he'll need to use the file edit method. Moving intended SNR to the left actually increases Target SNRM (left is labelled "maximum stability") hence I think far left actually gives you a target of 16db. Dropping it via the file edit method actually moves it to the "right" on the Line settings tab, so when you drop below the 12db default it actually shows no radio buttons selected:

any reason why you have your RFI dragged back one? has it made an improvement for you dragging that back one?

the RFI slider is one that its effect doesnt seem to be overly tested, so curious..

stevehodge:

I can't see why some cabinets would and some wouldn't, but who knows? My cabinet actually predates that big cabinet rollout: it's not even a whisper cabinet.

We don't really know if CRC errors matter for getting on to 17a. I suspect they don't: my error rates were never outstanding, and it makes some sense that they might want to put people on 17a if they have high crosstalk in the 8b spectrum (which should cause more errors). I think attenuation is the most important factor.

i suspect your right, but then why does the 17a profile take so long at times to get activated? why not manually do it at previsioning? line attenuation less than 8db, put it on 17a to avoid crosstalk..

would make sense to do it that way anyway!

attenuation certainly does seem to be the biggest deciding factor.

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

1 | ... | 156 | 157 | 158 | 159 | 160 | 161 | 162 | 163 | 164 | 165 | 166 | ... | 285