Kitz Forum

Broadband Related => ADSL Issues => Topic started by: konrado5 on July 28, 2016, 11:23:21 PM

Title: Shunt fault?
Post by: konrado5 on July 28, 2016, 11:23:21 PM
Some day (year 2014) I've had reconnection when there was pavement resurfacing work. Then my attenuation there was 23.5/20.2 dB instead of standard 25.0/14.9 (standard on D-Link DSL-2740B, now I have Netgear DGND 3700v2). On the next day I've had reconnection again when there was pavement resurfacing work and attenunation came back to standard 24.0/14.9 dB.

I attach two Hlogs on one graph Double_Hlog1.png. The red graph is standard Hlog. The green graph is abnormal Hlog. The undulations and lazy roll around the tone 55 is the effect of measurements errors. The proof is green graph on Double_Hlog2.png which was generated when my target SNR margin was risded from standard 6.0 dB to 15.0/17.0 dB. It is noticable that on green graph there are less undulations.

Why do I think that on my circuit there is capacitive fault? On the red graph on the Double_Hlog1.png there is significant attenuation increase around the tone 210, on the green graph there is significant attenuation increase around the tone 150.

Quote
In other words, downstream attenuation tends to increase relative to the upstream attenuation, resulting in an increase in the attenuation ratio.
http://www.google.com/patents/EP2747401A1?cl=en

I think the pavement resurfacing work caused that the shunt fault was slightly and temporary modified (the attenuation increase on the earlier downstream frequency and increasement on upstream (20.2 dB instead of standard 14.8 dB).

What do you think about my hypothesis?

Best regards
konrado5
Title: Re: Shunt fault?
Post by: burakkucat on July 28, 2016, 11:59:44 PM
It is rather difficult to either propose a cause or to analyse an observed change, for one incident, on one circuit, retrospectively.

However if the cable in question is not ducted but is buried directly in the ground, then any resurfacing work might impose an additional (physical) loading on the cable. Such crushing-like action could conceivably perturb the normal capacitive coupling between the two legs of a pair and, thus, appear as a temporary capacitive shunt across the pair.

So in answer to the question --

Quote from: konrado5
What do you think about my hypothesis?

 -- I would say that it appears to fit the observations. Unless there is some other more compelling suggestion, then I would be prepared to accept it as the most likely cause.
Title: Re: Shunt fault?
Post by: konrado5 on July 29, 2016, 12:16:08 AM
There must have been some misunderstanding. I don't think that pavement resurfacing work caused temporary shunt fault. I think that shaunt fault is permanent but pavement resurfacing work slightly and temporary modified the shunt fault effect. I think the shunt fault is permanent because on the standard Hlog graph (the red graph) there is significant attenuation increase around the tone 210 and this increase remain up to the end of graph. Pavement resurfacing work caused to the attenuation increase movement, this increase was around the tone 150 instead of tone 210.

But my observations does not reflect accurately expected result of shunt fault (higher attenuation on higher frequiences). On standard Hlog (red graph) there is also increase around the tones 33-50. On the abnormal graph (green graph) there was significantly lower attenuation on these frequiences. Unfortunately I have not any graph for upstream but the upstream attenuation was 20.2 dB instead of standard 14.8 dB. I conclude that on that day increase around the tones 33-50 was moved to the upstream.

To sum up, I see two frequency ranges with too much attenuation related to the remaining frequencies but pavement resurfacing caused movement these ranges to the lower frequiences. :)

Best regards
konrado5
Title: Re: Shunt fault?
Post by: burakkucat on July 29, 2016, 01:41:00 AM
I shall have to give your observations and hypothesis some further thought.

What I am finding a little difficult to rationalise is how you can say, with any degree of certainty, that the circuit has a capacitive shunt fault when you are only considering a sample size of one! If you had the equivalent data for, say, one thousand circuits then it would be fairly straightforward to offer up the results of an analysis of all those circuits and, thus, support the claim for a shunt fault. Obviously every circuit has a natural capacitance; a function of the product of the intrinsic capacitance per unit length and the total installed length of the cable.
Title: Re: Shunt fault?
Post by: konrado5 on July 29, 2016, 01:52:18 AM
Sudden attenuation increase around tone 210 (around tone 150 on abnormal Hlog) is reason to think about shunt fault. In this patent there is that on shunt fault there is too high downstream attenuation relative to upstream attenuation. On my circuit there is too high 210-511 tone attenuation relative to other tones attenuation.

Best regards
konrado5
Title: Re: Shunt fault?
Post by: burakkucat on July 29, 2016, 05:40:01 PM
Before the pavement resurfacing work the line exhibited a certain capacitance, Cbefore   (1)

During the pavement resurfacing work the line exhibited a different capacitance, Cduring   (2)

Upon completion of the pavement resurfacing work the line exhibited a different capacitance, Cafter   (3)

Are you proposing that the observed capacitance at event (2) is greater or lesser than at (1)? (Is Cbefore < Cduring or is Cbefore > Cduring ?)

Are you proposing that the observed capacitance at (3) is identical to that at (1)? (Is Cbefore = Cafter ?)   (4)

If the statement at (4) is invalid, are you proposing that the observed capacitance at (3) is greater or lesser than at (2)? (Is Cduring < Cafter or is Cduring > Cafter ?) Further, if the statement at (4) is invalid, are you proposing that the observed capacitance at (3) is grater or lesser that at (1)? (Is Cbefore < Cafter or is Cbefore > Cafter ?)

The major problem, as I see it, is that you are attempting to map observed results (of a set of one), whilst attempting to use that same set (of one) as a baseline reference for the observed result. A visual analogy would be the Penrose Stairs (https://en.wikipedia.org/wiki/Penrose_stairs), where one may perpetually ascend (or descend) due to the visual baseline of reference being subtly modified.

(https://upload.wikimedia.org/wikipedia/commons/thumb/3/34/Impossible_staircase.svg/500px-Impossible_staircase.svg.png)
Title: Re: Shunt fault?
Post by: konrado5 on July 29, 2016, 07:03:10 PM
Quote from: burakkucat
Are you proposing that the observed capacitance at (3) is identical to that at (1)? (Is Cbefore = Cafter ?)   (4)
Yes, the abnormal capitation lasted only one day. Next day pavement resurfacing working caused shock on circuit again and Hlog came back to usual.
Title: Re: Shunt fault?
Post by: burakkucat on July 29, 2016, 08:38:25 PM
Ah, I see. Thank you. I am now prepared to accept your explanation as for the cause which then gave rise to the observed changes.  :)

So the work to resurface the pavement reduced the physical loading on the cable and the capacitance was ever so slightly changed from its earlier value. Then when the new surface was laid the physical loading on the cable increased once again. As a result, there was another ever so slight change in the capacitance.

You are making the (perfectly reasonable) assumption that no other characteristics of the cable has changed over the time-frame that elapsed whilst the work was taking place. Hence the variation in attenuation at specific frequencies has been correlated to the physical loading, first removed from and then restored onto the cable.

(There is one other environmental factor which might have had some relevance and that is the effect of temperature. But I would regard that as a somewhat slow acting effect and probably did not play a part during the couple of days that the cable was exposed.)

In Poland, is it normal for such cables to be directly buried in the ground? Perhaps it is just the individual service feeds that are buried direct and the larger, multiple-pair, distribution cables are deployed in ducts?
Title: Re: Shunt fault?
Post by: konrado5 on July 29, 2016, 09:05:28 PM
Still there is little misunderstanding. On 1 July 2014 I've had sudden reconnect and attenuation was changed from 25.0/14.8 dB to 23.5/20.2 dB. It is probably that reconnection had at that time more people because I've got SNR margin about 4.6 dB instead of standard about 6.0 dB immediately after connection. At the same time I've heard that workers was hitting the pavement. I also did reconnects on that day. It was continously 23.5 dB/20.2 dB for downstream (upstream slightly changing 19.8-20.2 dB). Next day (2 July 2014) I've had similar sudden reconnect and attenuation came back to usual 25.0/14.8 dB. It was therefore something else than physical loading. Something caused reconnect and change of circuit parameters.

Quote from: burakkucat
In Poland, is it normal for such cables to be directly buried in the ground? Perhaps it is just the individual service feeds that are buried direct and the larger, multiple-pair, distribution cables are deployed in ducts?
I'm not sure but I think that larger distribution cables are deployed in ducts. These cables are under the pavement.

Is it possible that shunt fault causes high attenuation on two frequency ranges 33-50, 250-511 (on abnormal Hlog upstream frequiences and 150-511) relative to the other frequiences? I've read that shunt fault causes high attenuation on higher frequiences relative to lower frequiences. I see however two ranges: one low frequiences, second high frequiences.
Title: Re: Shunt fault?
Post by: burakkucat on July 29, 2016, 10:31:12 PM
Quote from: konrado5 on July 29, 2016, 09:05:28 PM
Something caused reconnect and change of circuit parameters.

Hmm . . . then I don't think we will be able to say what was the underlying cause. (A technician could have been working on a large joint-closure, for example.)

Quote

I'm not sure but I think that larger distribution cables are deployed in ducts. These cables are under the pavement.

Yes, I understand. Thank you.

Quote
Is it possible that shunt fault causes high attenuation on two frequency ranges 33-50, 250-511 (on abnormal Hlog upstream frequiences and 150-511) relative to the other frequiences? I've read that shunt fault causes high attenuation on higher frequiences relative to lower frequiences. I see however two ranges: one low frequiences, second high frequiences.

A series capacitive fault would behave as a high pass filter. Attenuating low frequencies but allowing higher frequencies to pass.

A shunt capacitive fault would show behaviour similar to a low pass filter. Attenuating high frequencies but allowing lower frequencies to pass.

You make reference to three bands of sub-carriers; 33 - 50, 250 - 511 & 150 - 511. Let's calculate the equivalent frequencies --

sub-carrier        frequency (kHz)
33                      142.3125
50                      215.625
150                    646.875
250                  1078.125
511                  2203.6875

Looking at the frequencies displayed in that table, I am having great difficulty in convincing myself that either of the two ranges (which you have mentioned) could both be simultaneously subject to increased attenuation as a result of a change in capacitive shunt effect.

I spent some time looking at the formulae on the RLC circuit (https://en.wikipedia.org/wiki/RLC_circuit) Wikipedia page which eventually convinced me that what you have observed cannot be due to just an increase (or decrease) in shunting capacitive effect. Try as I might, I cannot find any other possible explanation.

My overall feeling is that this is another case where I have to say: "Sorry, I just do not know".  :(
Title: Re: Shunt fault?
Post by: konrado5 on July 29, 2016, 11:40:03 PM
Quote from: burakkucat
Hmm . . . then I don't think we will be able to say what was the underlying cause. (A technician could have been working on a large joint-closure, for example.)
Perhaps it coincided with works of technicans. But I've not observed similar changes except for July 2014 and pavement workings.

It is important also that reconnection during pavement works caused the higher attenuation moved from 33-50 tones to upstream and from 250-511 to 150-511. I have some kind of fault that can be modified that attenuated frequiences can be changed. Moreover, this fault can't cause CRC errors, SNR margin fluctuations. My circuit is stable on SNR margin about 1 dB. Any idea?
Title: Re: Shunt fault?
Post by: burakkucat on July 30, 2016, 12:14:26 AM
Quote from: konrado5 on July 29, 2016, 11:40:03 PM
I have some kind of fault that can be modified that attenuated frequiences can be changed. Moreover, this fault can't cause CRC errors, SNR margin fluctuations. My circuit is stable on SNR margin about 1 dB.

A very peculiar circuit when compared with those in the UK. I have tried to understand it but without success.

Quote
Any idea?

No, sorry. I am completely "out of ideas". I just do not know.  :no:
Title: Re: Shunt fault?
Post by: konrado5 on July 30, 2016, 12:53:51 AM
But do you see on my usual Hlog sudden and continuing attenuation risement above 250 tone?
Title: Re: Shunt fault?
Post by: ejs on July 30, 2016, 06:11:39 AM
It's perfectly normal that higher frequencies have higher attenuation. Hlog is the channel transfer function, I think it's like the inverse of attenuation.
Title: Re: Shunt fault?
Post by: William Grimsley on July 30, 2016, 08:35:41 AM
Quote from: konrado5 on July 30, 2016, 12:53:51 AM
But do you see on my usual Hlog sudden and continuing attenuation risement above 250 tone?

burakkucat doesn't have any more ideas, as is evidently stated in his pervious post! ::)
Title: Re: Shunt fault?
Post by: konrado5 on July 30, 2016, 12:24:45 PM
I correct my observations. It is difficult to abstract from measurement errors. It seems that on usual Hlog I have too much attenuation on tones 33-100 (unknown minimum because of measurement errors) and 170-220 (210 is minimum). On abnormal Hlog I have too much attenuation on upstream, 120-170 (150 is minimum). Perhaps I have bridge tap on circuit and pavement workers changed the bridge tap parameters? Could you look at this graph and estimate if 210 dip indicate bridge tap? The 55 dip is measurement error. It is curious that on both Hlogs (first post in this thread) there is almost the same attenuation on 250-511 tones. It weights in favour of bridge tap hypothesis because HR joints and capacitive faults affect the whole ADSL band but in the different degree. Admittedly HR joint causes higher attenuation especially in low frequiences but higher frequiences are also affected. Bridge taps affect selectively.

Best regards
konrado5
Title: Re: Shunt fault?
Post by: burakkucat on July 30, 2016, 04:00:24 PM
Quote from: konrado5 on July 30, 2016, 12:24:45 PM
Could you look at this graph and estimate if 210 dip indicate bridge tap?

I have looked and it is exactly as what I have previously seen in some of your Hlog plots. It is not due to a bridging tap. Its shape is quite wrong; it is too wide and shallow.

I attach, below, two Hlog plots with data obtained from one circuit. (Unfortunately the CPE - DSLAM paring did not record the US data. (Upstream relative to the CPE.)) The first plot shows the presence of a bridging tap. The second plot is from the same circuit once the bridging tap had been removed.

The third image, attached for reference, shows the classic shape which is characteristic of a bridging tap.
Title: Re: Shunt fault?
Post by: konrado5 on July 30, 2016, 04:45:42 PM
On attached j3w9khlog you found bridge tap. But it seems to me that it has similar depth and wideness.
Title: Re: Shunt fault?
Post by: burakkucat on July 30, 2016, 05:18:22 PM
Quote from: konrado5 on July 30, 2016, 04:45:42 PM
On attached j3w9khlog you found bridge tap.

Correct. Look at the first minimum, at the low frequency end of the plot. Look at its shape. That is exactly what I would expect. Granted there is some distortion in the plot. But not everything in this world exists in an "either or" binary state. There are varying degrees in between. Experience tells me that the initial minimum, at around sub-carrier 60, is due to the presence of a bridging tap. The subsequent minima, as one views the plot in a low frequency to high frequency direction, become both less deep and wider. That is expected. They are the harmonic responses of the original response. It is by consideration of the initial and the subsequent harmonic minima that allows one to estimate the approximate length of the bridging tap. (Note the words "estimate" and "approximate" are used.) Experience tells me that the circuit, from whose data that the "j3w9khlog" plot was generated, had a bridging tap at the time the data was collected.

Quote
But it seems to me that it has similar depth and wideness.

In your opinion.

I have nothing more to add to this thread. It has deviated away from its original topic of "Shunt fault?" and has become an irritating distraction that has consumed a significant amount of my time.
Title: Re: Shunt fault?
Post by: konrado5 on August 02, 2016, 03:50:59 PM
burakkucat: Little misundestanding again. I don't think that minimum on my 210 tone is the real first minimum. I think the first minimum is somwhere at the tones 33-50 or upstream. It is difficult to estimate because on these tones there are significant measurement errors, for example minimum at tone 55 is caused by Annex M PSD mask. G.922.3 suggests that on tones 33-45 it is possible that there are inaccurate measurements.
Quote
The accuracy requirements for the downstream HLOGps (HLOGps_ds) shall apply only to the
following subcarriers (with the corresponding frequency ranges being a part of the passband), and
only if not within the downstream BLACKOUTset (see clause 8.13.2.4):
•
Annexes A and I:
Subcarriers 46 to 208.
Title: Re: Shunt fault?
Post by: William Grimsley on August 02, 2016, 03:56:59 PM
konrado5, burakkucat has said that he doesn't want to add anything more to the thread. Jeeez. ::)
Title: Re: Shunt fault?
Post by: konrado5 on August 02, 2016, 04:04:17 PM
But I had to add. Sorry.
Title: Re: Shunt fault?
Post by: konrado5 on August 02, 2016, 04:13:50 PM
Now I agree with old burakkucat statement.
Quote from: burakkucat
If you are asking me for my further thoughts on the observation I will say that, although impedance changes do occur with changes in wire gauge, it could be that the undulating roll towards the higher frequencies in your Hlog plot are the secondary (and higher) effect of the presence of a bridging tap at the lower frequency end of the plot. However it is inconclusive due to the dip that is often apparent when transitioning from the US frequencies to the DS frequencies. To be honest, a TDR (or FDR) trace may give a better idea of what is occurring.
http://forum.kitz.co.uk/index.php/topic,14437.msg270696.html#msg270696
Title: Re: Shunt fault?
Post by: William Grimsley on August 02, 2016, 04:15:15 PM
It's fine, mate. We're just chilling on here, you know? :)

Meant to ask, what's my Hlog and QLN looking like? MDWS ID is WilliamG.
Title: Re: Shunt fault?
Post by: konrado5 on August 02, 2016, 04:27:23 PM
Your Hlog is perfect. On QLN I see the radio frequency ingress on tones 211-244 (frequiences about 900-1050 kHz).
It should be helpful
" ~ Medium Wave Channels in the UK that may interfere with broadband"
http://www.kitz.co.uk/adsl/rein.htm
Title: Re: Shunt fault?
Post by: William Grimsley on August 02, 2016, 05:07:09 PM
That's great. Thanks for the information.
Title: Re: Shunt fault?
Post by: konrado5 on December 30, 2016, 04:41:25 PM
I think that burakkucat misunderstood what I've tried to convey. Look again at Double_Hlog1.png. The green graph is anomalous Hlog (when pavement works caused disconnection). It is noticable that lazy roll around 210 tone was moved to 150 tone. I think it is harmonic response on the first minima (on the red graph 33-90 tones, on the green graph somewhere on upstream, upstream don't measured unfortunately). Moreover, I don't think that 55 minima is real bridge tap minima. It is measurement error. On the Double_Hlog2.png this minima is lesser on the green graph (anomalous synchronization with 15 dB target SNR margin).
Title: Re: Shunt fault?
Post by: konrado5 on August 14, 2017, 10:36:50 PM
burakkucat: do you think it is possible that lazy roll is harmonic response to first minima?
Title: Re: Shunt fault?
Post by: j0hn on August 14, 2017, 11:51:59 PM
Quote from: burakkucat
I have nothing more to add to this thread. It has deviated away from its original topic of "Shunt fault?" and has become an irritating distraction that has consumed a significant amount of my time.
:whistle: