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

Login with username, password and session length
Advanced search  

News:

Pages: [1] 2

Author Topic: FTTC Down and Up speed mismatch  (Read 1392 times)

kwillers

  • Member
  • **
  • Posts: 68
FTTC Down and Up speed mismatch
« on: June 19, 2017, 11:12:52 AM »

So the BT checker for my line suggest speeds of approx
Down 44-60
Up 11-17

Now I'm actually seeing Down of 62 and Up of 10

Is that usual to see the two at separate ends of the estimates or does it suggest a line issue somewhere ?

Logged
Life is what happens while you're making plans

pfSense 2.4 on APU2 with Zen FTTC 80/20

Dray

  • Kitizen
  • ****
  • Posts: 2335
Re: FTTC Down and Up speed mismatch
« Reply #1 on: June 19, 2017, 11:28:19 AM »

Can you post a lot more line stats?
Logged

skyeci

  • Reg Member
  • ***
  • Posts: 984
Re: FTTC Down and Up speed mismatch
« Reply #2 on: June 19, 2017, 11:37:57 AM »

Kev uploads as kwillers on mydslwebstats if that helps.
Logged
Sky Fibre Pro -  billion 8800nl v1 (bridge mode) + PFSENSE (APU2C4) 2.4.0 with ipv6 , AC-88U WAP- ECI cab, G.INP disabled as of 8th April 2016

http://www.mydslwebstats.co.uk user upload ID skyECI (using a pi3)

Chrysalis

  • Content Team
  • Addicted Kitizen
  • *
  • Posts: 5031
Re: FTTC Down and Up speed mismatch
« Reply #3 on: June 19, 2017, 12:03:47 PM »

may have the issue that ronski has, one other guy had it also, ill check mdws later
Logged
Sky Fiber Pro - Billion 8800NL bridge & PFSense BOX running PFSense 2.4 - ECI Cab

Dray

  • Kitizen
  • ****
  • Posts: 2335
Re: FTTC Down and Up speed mismatch
« Reply #4 on: June 19, 2017, 12:12:39 PM »

No upload tones after 2396 on the Bits/tone graph
Logged

WWWombat

  • Kitizen
  • ****
  • Posts: 1669
Re: FTTC Down and Up speed mismatch
« Reply #5 on: June 19, 2017, 12:56:55 PM »

So the BT checker for my line suggest speeds of approx
Down 44-60
Up 11-17

Now I'm actually seeing Down of 62 and Up of 10

Is that usual to see the two at separate ends of the estimates or does it suggest a line issue somewhere ?

That's a good question, but the answer might be being overtaken by circumstances.

Six months ago+, your speeds looked like 60/15 and 55/13. They look like they were in proportion.

Recently, before SNRM changes, you were getting 52/11. Slightly favouring the downstream at this stage, but not hugely out of balance.

Now you are around 62/11, where the balance has shifted a lot. However ... that balance only shifted because XdB has come into operation, and given you an extra 10Mbps in one direction only.

I would suggest that the estimator hasn't been changed to take account of XdB, and may never be. In those circumstances, most lines will look to favour the downstream much more than the upstream.

As for why your line, at 52/11, was slightly out of balance? It looks to be noise in the upstream direction. If you look at the "Quiet Line", "SNR per tone" and "bits per tone" graphs, you can see that U2 has relatively few bits, seemingly because it is noisier - especially in the upper part of U2. Compare the noise with other similar lines, like "happyhorse". It also looks like power is playing a part too, as you have slightly lower upstream power than similar lines; this may be restricted as part of the "upstream power backoff" mechanism, trying to mediate upstream capability between lines.
Logged

kwillers

  • Member
  • **
  • Posts: 68
Re: FTTC Down and Up speed mismatch
« Reply #6 on: June 19, 2017, 01:11:41 PM »

No upload tones after 2396 on the Bits/tone graph

hmmm hadn't spotted that

I've actually got a new drop cable being installed tomorrow, so will be interesting to see if that has any effect. Long story as to why its being moved.

Logged
Life is what happens while you're making plans

pfSense 2.4 on APU2 with Zen FTTC 80/20

kwillers

  • Member
  • **
  • Posts: 68
Re: FTTC Down and Up speed mismatch
« Reply #7 on: June 19, 2017, 01:13:14 PM »

As for why your line, at 52/11, was slightly out of balance? It looks to be noise in the upstream direction. If you look at the "Quiet Line", "SNR per tone" and "bits per tone" graphs, you can see that U2 has relatively few bits, seemingly because it is noisier - especially in the upper part of U2. Compare the noise with other similar lines, like "happyhorse". It also looks like power is playing a part too, as you have slightly lower upstream power than similar lines; this may be restricted as part of the "upstream power backoff" mechanism, trying to mediate upstream capability between lines.

Not sure I fully follow that but I'm trying to :-)
Logged
Life is what happens while you're making plans

pfSense 2.4 on APU2 with Zen FTTC 80/20

Chrysalis

  • Content Team
  • Addicted Kitizen
  • *
  • Posts: 5031
Re: FTTC Down and Up speed mismatch
« Reply #8 on: June 19, 2017, 01:15:43 PM »

That's a good question, but the answer might be being overtaken by circumstances.

Six months ago+, your speeds looked like 60/15 and 55/13. They look like they were in proportion.

Recently, before SNRM changes, you were getting 52/11. Slightly favouring the downstream at this stage, but not hugely out of balance.

Now you are around 62/11, where the balance has shifted a lot. However ... that balance only shifted because XdB has come into operation, and given you an extra 10Mbps in one direction only.

I would suggest that the estimator hasn't been changed to take account of XdB, and may never be. In those circumstances, most lines will look to favour the downstream much more than the upstream.

As for why your line, at 52/11, was slightly out of balance? It looks to be noise in the upstream direction. If you look at the "Quiet Line", "SNR per tone" and "bits per tone" graphs, you can see that U2 has relatively few bits, seemingly because it is noisier - especially in the upper part of U2. Compare the noise with other similar lines, like "happyhorse". It also looks like power is playing a part too, as you have slightly lower upstream power than similar lines; this may be restricted as part of the "upstream power backoff" mechanism, trying to mediate upstream capability between lines.

wombat check my line, I upgraded the billion dsl firmware again last week, and observed the UPBO is more aggressive, a circa 3db loss on the upstream margin.  I did multiple resyncs to see if it was the UPBO lottery, which I think its not.  I will downgrade the firmware again sometime this week to see if the upstream SNRM recovers, so dsl firmware could be playing a part here.
Logged
Sky Fiber Pro - Billion 8800NL bridge & PFSense BOX running PFSense 2.4 - ECI Cab

kwillers

  • Member
  • **
  • Posts: 68
Re: FTTC Down and Up speed mismatch
« Reply #9 on: June 20, 2017, 02:39:13 PM »

Well I had the new drop cable installed today

Here is the outcome for the Upload speeds

OLD CABLE
UP 10,531 with 5.3 of SNR

NEW CABLE
UP 10,646 but with 6.4 SNR

So a good 1 dB of improvement in the SNR but negligible speed and I'm still missing tones (right expression ??) in upper U2

So now time to raise with my ISP
Can anyone help me out how here explain how I frame the problem to the ISP will "missing tone in upper U2 make sense to them  ?"

« Last Edit: June 20, 2017, 02:54:29 PM by kwillers »
Logged
Life is what happens while you're making plans

pfSense 2.4 on APU2 with Zen FTTC 80/20

Chrysalis

  • Content Team
  • Addicted Kitizen
  • *
  • Posts: 5031
Re: FTTC Down and Up speed mismatch
« Reply #10 on: June 20, 2017, 04:17:53 PM »

kevin having now looked at your bits/tone data I think your issue is a UPBO problem.

Looks like you are on a long line profile where the power is focused on U1, but looking at your DS tones I expect if you had a short line power mask profile then the U2 tones would give you more bits than U1.  So its a dslam configuration issue.

Your signal in the DS areas surrounding the U2 range is actually stronger than my line, the difference is that I have a different power mask profile and as a result can comfortably max out the US at 20000kbits.
« Last Edit: June 20, 2017, 04:20:55 PM by Chrysalis »
Logged
Sky Fiber Pro - Billion 8800NL bridge & PFSense BOX running PFSense 2.4 - ECI Cab

kwillers

  • Member
  • **
  • Posts: 68
Re: FTTC Down and Up speed mismatch
« Reply #11 on: June 20, 2017, 05:04:25 PM »

I may have a first in computer science but you lost me there :-)

Sounds like if its a DSLAM issue I'm screwed right ?
Logged
Life is what happens while you're making plans

pfSense 2.4 on APU2 with Zen FTTC 80/20

Chrysalis

  • Content Team
  • Addicted Kitizen
  • *
  • Posts: 5031
Re: FTTC Down and Up speed mismatch
« Reply #12 on: June 20, 2017, 09:00:41 PM »

Well your line on the DS seems to be performing well given the attenuation.  So I make the assumption lines at comparable distance might usually have less useable bits in the U2 range but your line looks like without any UPBO would populate the U2 bits comparable to what you get in D2.

Openreach configure UPBO so either U1 or U2 has power back off applied, for long lines its U2 and for short lines its U1.  But when you have lines hovering in between they can be stuck on configurations that are not optimal.
Logged
Sky Fiber Pro - Billion 8800NL bridge & PFSense BOX running PFSense 2.4 - ECI Cab

WWWombat

  • Kitizen
  • ****
  • Posts: 1669
Re: FTTC Down and Up speed mismatch
« Reply #13 on: June 20, 2017, 10:00:54 PM »

Not sure I fully follow that but I'm trying to :-)

I'll try to explain better... A bit long, though.

If you look at the "Quiet Line", "SNR per tone" and "bits per tone" graphs, you can see that U2 has relatively few bits, seemingly because it is noisier - especially in the upper part of U2.

First, check the "bits per tone" graph for your line, and for "happyhorse". You'll note that your U2 spectrum is carrying fewer bits, and none in the higher region. Around 3 bits per tone, perhaps.

In DSL, the reason that the modem synchronises with fewer bits on any one frequency is because it has a lower SNR. That means the spectrum is receiving lower signal levels or higher noise levels; the lower number of bits is the modem's way to cope with that situation. Comparing the two "SNR per tone" graphs confirms this.

Are you getting more noise from other lines? Yes. This can be seen in the "quiet line" graph (although that only measures the noise at the time of the last sync; it isn't "live").

In U2, "happy horse" is seeing noise from other subscribers of between -122 and -124dB; your line is between -116dB and -120dB - which is perhaps 4-6dB noisier; for comparison, -140dB is very, very quiet.

In DSL, there needs to be an extra 3dB of SNR in order for a tone to carry one extra bit. Your extra 4-6dB of noise means (roughly) that each tone in U2 can carry 1-2 bits less than "happyhorse"'s.

Onto the "power" graph... Unfortunately, this only shows the aggregate transmission power, rather than the power used on each tone. So we have to guess a little.

Your aggregate upstream power is running at 1.5dB lower than "happyhorse"'s. If this means the power of individual tones is lower by that amount, then your line also has a slightly lower signal level too. A lower signal that could be worth another bit here and there.

In total, we can see reasons why your U2 tones can be carrying 2-3 bits less. When aggregated over time, that could be worth 10Mbps on your sync speed.

The question is ... is any of this speed recoverable through a fault. The answer is that it is unlikely.

a) Crosstalk noise varies, and depends on the number of your neighbours that have subscribed.

Complaining to your ISP, or calling out an Openreach engineer, isn't going to get rid of those subscribers.

b) Upstream power varies too; it is forced to reduce by the DSLAM to allow distant subscribers to get reasonable upstream speeds.

Again, complaining to your ISP, or calling out an Openreach engineer, isn't going to get rid of those subscribers.

That is, I should say, all just my opinion. There could be some form of fault on your line that happens to be augmenting the noise or suppressing the signal ... which an engineer visit could fix. But there's little that stands out as obvious.

Onto the positive news...

Going by the attenuation values, "hungryhorse" has a shorter line, and ought to get higher speeds anyway. Yet the downstream speed is actually comparable.

Go back to that "quiet line" graph, and look at the noise in the downstream spectrum. You'll see that your line is quieter by 3-4dB. That helps your line carry more downstream speed, relatively.

DSL is a matter of swings and roundabouts. You can see how much your speed depends on the noise/interference you are receiving from other lines. And that interference behaves differently for upstream and downstream.

So now time to raise with my ISP
Can anyone help me out how here explain how I frame the problem to the ISP will "missing tone in upper U2 make sense to them  ?"

"missing tones in upper U2" is, in the end, a consequence of DSL trying to make best use of your line's capabilities, in your line's environment, while also trying to make best use of other lines. It is "designed-for", not a fault.

ISPs won't care about details like this. In the end, all they care about is whether the total throughput speed is within spec.
Logged

WWWombat

  • Kitizen
  • ****
  • Posts: 1669
Re: FTTC Down and Up speed mismatch
« Reply #14 on: June 20, 2017, 10:12:30 PM »

observed the UPBO is more aggressive, a circa 3db loss on the upstream margin.  I did multiple resyncs to see if it was the UPBO lottery, which I think its not.

I note you saw power changes in those resyncs, by almost 3dB too.

Strange that upstream and downstream power changes seem to match each other at almost all times. I wonder if the stats programme is misreading one of them. I wonder too what happens to the power level in the different bands.

Logged
Pages: [1] 2