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

Login with username, password and session length
Advanced search  

News:

Author Topic: Can't still be Xmas trees!!  (Read 3319 times)

les-70

  • Kitizen
  • ****
  • Posts: 1255
Can't still be Xmas trees!!
« on: February 26, 2015, 01:28:18 PM »

    I had a short period of high up stream errors just before Xmas.  This went away for about a month then at lighting up time short duration upstream ES peaks started to intermittently appear and more recently steady overnight upstream ES rates.  The current upstream ES rate tends to be around 100-200 ES/hour between about 18:00 and 07:00 which means that the 24 hour total is potentially going to cause interleaving (on TTB with an expected 1440ES/day limit).  I am assuming that this is not a line fault as it otherwise seems quite normal in the day and also that finding the cause and getting anything done about it may be impossible to achieve.  Given the errors are only upstream I am assuming the error noise source is near the CAB and not near me. I can't spot any flickering street lights but there are various bits of flashing signage near the CAB. The nearest street lights to the CAB are gas with a gas powered clockwork timers - no doubt best for broadband!!

  All that I have found to do to minimize the errors is an upstream speed cap of 16Mb/s .  That seems to maximizes the upstream power and SNRM, capping to a lower speed values leads to a drop power and less SNRM.  The DSLAM seems to adjust the upstream power and seems reluctant to allow too high a value of SNRM.

   At the moment I am powering down at between 18:00 and 07:00 so the DLM does not notice.   This is not much bother to me and if needed I have a short evening session with the all powered up. 

    Unless anyone has any other ideas I am going to hope it eventually goes away or G.INP is enabled to make interleaving more bearable..  The upstream errors are all single CRC errors and should be easy fix with interleaving.
Logged

WWWombat

  • Kitizen
  • ****
  • Posts: 1674
Re: Can't still be Xmas trees!!
« Reply #1 on: February 26, 2015, 06:15:26 PM »

The current upstream ES rate tends to be around 100-200 ES/hour between about 18:00 and 07:00 which means that the 24 hour total is potentially going to cause interleaving (on TTB with an expected 1440ES/day limit).
I don't know if we should assume that upstream has the same thresholds. I guess we get to see when the day comes that you leave it powered up.
Quote
  All that I have found to do to minimize the errors is an upstream speed cap of 16Mb/s .  That seems to maximizes the upstream power and SNRM, capping to a lower speed values leads to a drop power and less SNRM.  The DSLAM seems to adjust the upstream power and seems reluctant to allow too high a value of SNRM.
There's a thing known as "upstream power backoff" which is generally used to reduce power of those close to cab, so their signal doesn't swamp that of the distant users, which would cause too much crosstalk upstream, and needlessly lose their bandwidth.

I haven't looked into the algorithms used by UPBO, but looking for high SNRM might well be the way it works.
Logged

les-70

  • Kitizen
  • ****
  • Posts: 1255
Re: Can't still be Xmas trees!!
« Reply #2 on: February 26, 2015, 07:12:31 PM »

I don't know if we should assume that upstream has the same thresholds. I guess we get to see when the day comes that you leave it powered up.

   I did have it on overnight -just once- the total upstream ES reached 1200 - based on the average over the up-time.  The second night I switched off as had I left it on for the full 24 hour the 24 hour up-time average would have been about 1700.   This suggests the ES limit is probably at least 1440 on TTB, but as you say unless you push it until it goes over a limit you can't be sure of the actual value. 
Logged

NewtronStar

  • Kitizen
  • ****
  • Posts: 4850
Re: Can't still be Xmas trees!!
« Reply #3 on: February 26, 2015, 10:14:53 PM »

At the moment I am powering down at between 18:00 and 07:00 so the DLM does not notice.   This is not much bother to me and if needed I have a short evening session with the all powered up. 

Les that's a bit drastic you should be able to get away with 2000 errored seconds per day on standard and don't take the TTB calc as the holy grail you would probably get away with 2500 es/per day.
« Last Edit: February 26, 2015, 10:19:23 PM by NewtronStar »
Logged

les-70

  • Kitizen
  • ****
  • Posts: 1255
Re: Can't still be Xmas trees!!
« Reply #4 on: February 27, 2015, 07:11:45 AM »

  I believe "standard" is 1440 per day and others on TT have had a DLM reaction not much over that figure.  As WWWombat noted we are less sure about upstream figures, I am assuming it is also 1440.  Your correct that I should be able to be less drastic though, maybe simply turn it off overnight when I am in bed.  I was thinking on doing that as now evenings and morning are getting lighter the time span of the errors is reducing anyway. 
Logged

Chrysalis

  • Content Team
  • Addicted Kitizen
  • *
  • Posts: 6406
Re: Can't still be Xmas trees!!
« Reply #5 on: February 27, 2015, 08:23:03 AM »

The current upstream ES rate tends to be around 100-200 ES/hour between about 18:00 and 07:00 which means that the 24 hour total is potentially going to cause interleaving (on TTB with an expected 1440ES/day limit).
I don't know if we should assume that upstream has the same thresholds. I guess we get to see when the day comes that you leave it powered up.
Quote
  All that I have found to do to minimize the errors is an upstream speed cap of 16Mb/s .  That seems to maximizes the upstream power and SNRM, capping to a lower speed values leads to a drop power and less SNRM.  The DSLAM seems to adjust the upstream power and seems reluctant to allow too high a value of SNRM.
There's a thing known as "upstream power backoff" which is generally used to reduce power of those close to cab, so their signal doesn't swamp that of the distant users, which would cause too much crosstalk upstream, and needlessly lose their bandwidth.

I haven't looked into the algorithms used by UPBO, but looking for high SNRM might well be the way it works.


UPBO seems to only affect U1, so if les disabled U2, he may escape UPBO but with the loss of U2 tones.

The asus modems seem to allow all upstream tones and without UPBO (breaching openreach RFC).
Logged
AAISP - Billion 8800NL bridge & PFSense BOX running PFSense 2.4 - ECI Cab - LINE STATISTICS CLICK HERE

les-70

  • Kitizen
  • ****
  • Posts: 1255
Re: Can't still be Xmas trees!!
« Reply #6 on: February 27, 2015, 09:08:40 AM »

  The ASUS has some appeal due to its many options including disabling UPBO but that apart it looks to have upset quite a few lines with its error behavior and the dodgy firmware. On top of that the lack of monitoring software would annoy me a lot.

   A while ago I started a thread on my efforts to try to introduce removable attenuation or noise. I was aiming to use it with modems that don't allow speed capping. I tried a variety of things that gave the attenuation but found it almost impossible to remove the attenuation smoothly enough and it often resynced during the removal stage.  I tried variable resistors and capacitors in various ways but although it worked once in a while it was very very unreliable, and with the resyncs it often gave, it was not something to attempt more than once a day. On the very few times when it worked it capped the upstream and downstream speed but without invoking the UPBO on the upstream.   

 
Logged

Chrysalis

  • Content Team
  • Addicted Kitizen
  • *
  • Posts: 6406
Re: Can't still be Xmas trees!!
« Reply #7 on: February 27, 2015, 11:49:07 AM »

yeah I would avoid the asus.
Logged
AAISP - Billion 8800NL bridge & PFSense BOX running PFSense 2.4 - ECI Cab - LINE STATISTICS CLICK HERE

kitz

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 32659
  • Trinity: Most guys do.
    • http://www.kitz.co.uk
Re: Can't still be Xmas trees!!
« Reply #8 on: February 28, 2015, 12:56:10 AM »

  I believe "standard" is 1440 per day and others on TT have had a DLM reaction not much over that figure.  As WWWombat noted we are less sure about upstream figures, I am assuming it is also 1440.

I believe its the same figures for both upstream and down. 

Yes 1440 would appear to be correct for Standard and 2880 for Speed.   These are the latest figures I could find specifically for FTTC  (see below) and dated June 2012.  There were some new parameters issued last spring but they seem to have only been applied to WBC adsl1/2+ and not FTTC which from all observations on MDWS still uses the 30/300 based profiles.   
Logged
Please do not PM me with queries for broadband help as I may not be able to respond.
-----
How to get your router line stats :: ADSL Exchange Checker