Print

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

[Weaver]

I’m going to try and collect together all the facts we know about hollow curve disease (HCD), whether it be in ADSL or VDSL2. I’m also going to try and list important open questions.

1. Has anyone other than me ever seen HCD?
2. HCD as seen so far occurs between (i) [downstream] tones 40 and 80, or more rarely (ii) between tones 45-50 and 55-60 (the ‘notch’ type)
3. HCD only affects downstream
4. It usually grows gradually over a period of two weeks. However last Friday there was an exception: a sudden blast of HCD going from nothing to more than 6 dB depth.
5. How it is cured is unknown. An OR engineer visit cures it even if the engineer does nothing seemingly significant. On occasion the fix has not been long-lasting, and the fault has developed again after a few days.
6. It is not confined to one of my four lines; at least two of those lines have been known to have it.
7. For me it started in early 2020.

I could do with looking back through kitz forum postings to see what other generalisations can be found, and what ideas may be scuppered.

Appeal to anyone who has ever seen this kind of problem.

[Weaver]

As an aside, I’m going to follow the development of the current line 2 fault which is now leaving the early stage. Pictures below. Note that they are all timestamped; look in the top left hand corner of the image where it also shows which modem is pictured by the domain name label m1/m2/m3/m4, eg m2.csw.me.uk, or if no domain name, it is a literal IPv4 address where the nth modem is shown in the address as 192.168.n.1 ie the third decimal number.

Earliest picture first:

[Weaver]

This spontaneously recovered, to perfection, and without an engineer visit. So we now know that:

8. Spontaneous recovery without an intervention in the form of an engineer’s visit is possible, but very rare. It is only known to have happened twice thus far.

[Weaver]

9. Situations have been known where two lines are ill with HCD simultaneously. At different stages though.

10. Some lines seem to be much less prone to HCD, or even HCD-free possibly. Need to confirm that latter tentative assertion. Duration: at  least for a good long while, if not permanently.

11. Shapes of the HCD phenomenon vary somewhat.
(i) The classic shape is a concave-upwards parabola, whose axis of symmetry is vertical, and which has two local maxima at the endpoints. (ii) Another common early-stage-only shape is on the rhs, past the overall maximum, and is right-downwards sloping. The depth is very shallow and the rh endpoint may not be a exactly local maximum, but more like a point of inflection, or close to one. The distinguishing feature of this type is that at the point where the curve is furthest away from the most appropriate test chord, the value of the derivative dy/dx is not very close to zero, but is always at least slightly negative.
(iii) Recently seen for the first time is a subtype related to ii. Would welcome suggestions as to what it should be called. Here is a picture taken just a few hours after its first appearance:
   
In the section to the right of x=58, the value of day/dx is fairly close to zero and in the first section, 58-63 at least, the derivative is not negative. (Noise is ignored here and these comments everywhere are after an imagined smoothing post-processing to not only get rid of noise, but to remove all stalactites too. So comments involving the derivative should take this imagined filtering/post-processing into account.)

I could do with names for these subtypes’ shapes’ names. Any volunteer contributors?