Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[burakkucat]

   Hmm . . . A very good try, nevertheless.

(And b*cat thought he was being clever transmogrifying the original Fortran to C code . . .    )

[les-70]

 Whilst I should give up I am trying to understand things better.  I am intrigued by this patent which is filed in Europe and the USA by Huawei.

  https://data.epo.org/publication-server/getpdf.jsp?pn=2112810&ki=A1&cc=EP

  It is a bit devious in its presentation but Equations 20 and its derivative Eq. 21 are the key results. Fig 5  shows the total phase change in Hlin over frequency ranges and line lengths.    The Figure is an evaluation of the equations for AWG 26 and 24 lines.  If true it means that my speed of propagation calculation basis (also found in some different papers!) is largely irrelevant with the complex inductance and capacitative characteristics of the cable dominating.  The implied rate of change of phase angle is also so large (positively huge) that on my line a phase change of 2pi or 4pi would be aliased out. That would imply that the changes  I am seeing only add the odd 100m or two to the line length. 

           1.  Is anyone experienced enough to comment on this?

 Separate from this it remains a puzzle that Hlin from my modem/DSLAM gives different results every reboot.  Only the autocorrelation of the IDFT proved totally independant of reboot.

          2.  Please could others post a log files with HLin  and see if the number of tones between say crossing from +ve to -ve in either the real and imaginary parts stays the same after reboots.?

  Note that Hlin is only well defined for Hlog greater than -50 db.  It is best to look around tone 128.  I have had basically the same after immediate on/off/on reboots but never after a delay of about 10 of mins between off and on.   I attach two of my Hlin logs showing one of the largest difference that I get.  One is roughly cyclic in phase every 15 tones and the other every 4 tones.

  Thanks in advance

[asbokid]

Very interesting stuff, les-70!    Another excellent find. Thank you for sharing it.

So the novelty to Huawei's patent is that they claim to have invented a method for determining loop length that combines the phase change and the attenuation reporting by the Hlin(f) function.

If I've read it correctly, the loop length calculation from the phase change is largely invariant to changes in the wire gauge. Unlike the loop length calculation based on attenuation, which varies considerably depending on gauge.  Huawei claims that by combining the two calculations, it's possible to make accurate estimations of both the loop length and the dominant cable type in the loop.  Does that sound right?

I didn't fully understand your request (which doesn't bode well for making any useful contribution!) but am very happy to provide any line data if you explain exactly what you need.

EDIT: Just finished reading the whole patent. It seems about 80% fluff.  Perhaps that's how patent lawyers become millionaires by 30     

Linked below are the MATLAB functions for the RLCG equations and the ABCD two port matrix representation. They were copied from Walter Chen's chapter on transmission line theory.  At the time I only had an old copy of MATLAB which was locked to a specific keyserver.  And Octave, an open source alternative to MATLAB, proved not compatible enough to run the functions without modification.

Later the same functions turned up again, this time in C, coded by a Dr Messerschmitt    Not sure where his code is now though. 

cheers, a

EDIT:

[1] http://forum.kitz.co.uk/index.php/topic,10970.msg214562.html#msg214562

[les-70]

  Thanks for the thoughtful response.  I agree with your interpretation of the patent.  I am surprised at the enormous rate of phase change implied.

  Before engaging a calculation using the paper I am trying to fathom out what to make of the lack of reproducibility in my Hlin.  At its simplest and I am seeking info on whether other people get, like me, almost random rates of phase change in separate outputs of Hlin's.  Hlin is only determined on sync so the test is to see a few hlin stats after different syncs.  I seem to need a ~10 min gap with the router off between syncs to be sure that is a fresh test. 

   A post of a few hlin stats, like the ones I gave above, and after suitable resyncs would help answer my question on whether the reported Hlin is open to interpretation and whether my router/DSLAM   is reporting things in an odd way.

   Regards

[c6em]

Here you go....
from when my curosity was also aroused by the Hlin data so I stored it
Hlog, Hlin,Hlins,QLN....

[les-70]

 Many thanks c6em.  Your hlin has a nice phase periodicity that you can see in changes in sign of the real and imaginary parts. 

  Please could say what the modem/router is, and also when you can, add a couple of other hlin's separated by periods with the modem off.  I am trying to see if the nice changes that you have are same after full modem reboots.

  Regards

[burakkucat]

An offering from "TheCattery", where I am experimenting with a Huawei HG612 in non-VDSL2 modes.

[les-70]

 Thanks burakkucat, again if you have time please post another one or two at times after have had the router off for a period.   Yours shows a high frequency change -- a greater rate of phase angle.  I get ones like you sometimes and ones like c6em on other days!  I want to see if everyone gets such variations or just me!  It is hard to know how what ones like mine mean.

  Regards

[c6em]

D-link 2640B with firmware 3.06
Line 2.4km sync around 13300, target SNR mgin 3.0

Here is another set taken just now from the current session.
Note that my line had a total voice fault on it in between the earlier submission dated one and this one so everything was off/disconneted/non functional for 3 days.

[asbokid]

I am seeking info on whether other people get, like me, almost random rates of phase change in separate outputs of Hlin's.  Hlin is only determined on sync so the test is to see a few hlin stats after different syncs.  I seem to need a ~10 min gap with the router off between syncs to be sure that is a fresh test. 

   A post of a few hlin stats, like the ones I gave above, and after suitable resyncs would help answer my question on whether the reported Hlin is open to interpretation and whether my router/DSLAM   is reporting things in an odd way.

Aha.. I see (as the blind man said to the deaf man) 

Here's the Hlin data from an HG612 on my parents' ADSL2+ TalkTalk LLU line.  According to TDR, the loop length is close to 1000 metres.   I will power-down the Huawei and resync in ten minutes and post another Hlin dataset to this message.

cheers, a

EDIT: second Hlin dataset attached below

[les-70]

 Thanks 6cem it looks like your results have same character as mine.

  Your earlier set has a is periodic in about ~16 tones and the later one in about ~4 tones.  A factor of ~4 in rate of phase change.  I think the 2640b is same broadcom chip set as my DG834gv4 but the question is why the big variations?? between resyncs.

[les-70]

  Thanks asbo.  the HG612 seem to have gone from about 5 tone to about 3 tones period in its 2pi phase change.   It seems the rates of phase change are always quite definite and fairly fixed over frequency but different, unless your lucky, every resync.  You may like to try one another day but it looks like the rate of phase change in the hlin is not stable.

   This makes using it perplexing    .  I may give up or go back to thinking and reading for a bit!!!

[burakkucat]

I may give up or go back to thinking and reading for a bit!!!

I hope you will do the latter, above, for I am intrigued with what you have / are discovering. I can promise you other data sets, over the coming days. (Original ADSL, ADSL2 & ADSL2+ modes.)   

[burakkucat]

Today's offering is attached.

ADSL2 mode. Hlin and HlinS parameters collected a.s.a.p. after the HG612 was powered on and, thus, sync'd with the MSAN.

[les-70]

Thanks buakkucat.  Your phase change by 2pi in about 12 tones in the first data set and about 4 tones in the second.  We all seem to have very different values every resync. 

   I am no further on the issue of why things are different every time, that seem to be the key issue.

 I think the Huawei patent is in fact using much the same approach with Hlin as I did at first.


   Going back to my first try with w=2*pi*f*DT  and   DT= length/(c*VOP)

 Here the phase w is in radians and the frequency f is in Hertz and DT the time diference due to the line length, c the speed of light and VOP the propagation speed constant.

   This gives a rate of phase change/metre at 10MHz of 0.31/radians/metre with vop=0.66  (the value at 10Mhz is often quoted in cable specs. and about 0.3)

  The Huawei patent numbers give a value of about 19 radians/metre at the same frequency hence my "enormous" comment.

  I suspect it is meant to be 19 angluar degrees /meter   i.e. 0.33 radians/metre (corresonding to a VOP of about 0.62) 

  If true it is wonderful that patents can't even get the units right.