Broadband Related > Broadband Technology

Potential Line Length Calculation using Broadcom Diagnostics

(1/4) > >>

boozy:
I've been playing with the outputs from the Broadcom Diagnostics Tool and have come across a possible way of calculating line length.

My line was last reported by OR as 876m and from the Diagnostics output I get 890m.

I've also added a calculation from the high tones down, which should match on a good line (mine doesn't, but the value is where I think a fault lies - so might be good).

Curious if others try it or post the output to the "Run Hybrid Response Test" under the AFE menu.  This *will* cause a disconnect as it is done before sync, so bear that in mind when playing.

The range is likely limited to under 1km as I'm using a lookup rather than a calculation (as the accuracy will never be 100%), but it will try up to 7.4Km...  7.4Km is unlikely to be measurable as that would be about 2 tones per sinusoid.

The code is anything but clean, given everything I was experimenting with, but I will tart it up and post it...  Linux friendly for b*cat.

burakkucat:

--- Quote from: boozy on July 16, 2018, 03:47:33 PM ---The code is anything but clean, given everything I was experimenting with, but I will tart it up and post it...  Linux friendly for b*cat.

--- End quote ---

Thank you. I would be interested to see what you have done.  :)

As for running the diagnostics program, The Cattery is BGW-system free and I have no intention of creating a 'doze execution environment in which to run the utility.  :(

Looking at your "Lengths" plot, considering that which is more positive than -22.5 (is dBm the unit?) and if one discounts the "manic" low frequencies (sub-carrier 384 and below), is there a semi-symmetric point around sub-carrier 2304?  :-\

boozy:
I'm a bit more ecumenical in my OSes (Bill has had me over to Seattle twice, so I will always be a loss maker for him), I run many different ones.

It's badged as a HLog graph as shown by the tool but under the covers it has three outputs - hybrid response data (4096 points), Impulse Response Data (8192 points) and Phasese Response Data (4096 points between +/- 3.125...  I assume pi for now).

It was the manic bit piqued my interest.  There are clearly 2 sinusoids in the graph; one more noticeable in the low tones which attenuates quickly, but is present throughout, and one you've picked up by the symmetry.  I tried to autocorrelate, but failed in the low tones, then couldn't get a good fit for a sine wave as there were two interacting (but I feel if I reverse out the 27 and 250 using a sine function with some attenuation built in, I learn something more) - so looked at all the minima and maxima.

There were two clear periods in the data one of 27 tones and one about 250 tones, as well as extra fun caused by the interactions.  I tinkered up some code to measure across 4 avoiding noise, to give slightly more accuracy, and then looked up the distance needed for that reflection.

:) - that nearly took more typing than the code.

Edit: forgot to state the obvious.  The shortest period sinusoid should always be the line length (as that's the furthest thing measured) and if there is a second (or third) sinusoid - it'll be between the start point and the line end (hence why I calculated from each end of the frequency range, as the furthest response should attenuate quicker, leaving any other reflection visible).

boozy:
Here's code and some data - it may work immediately under mono (or chartarea may be a problem).

I've just noticed some missing code on the length lookups, it may fail for short lengths.  I'll have a play (and finish tidying the code later).

konrado5:
Perhaps hybrid response test is related to hybrid circuit. I've found some information on hybrid circuit on pages 18-19. There is also similar graph on page 19.
http://www.ti.com/lit/an/slyt091/slyt091.pdf

Navigation

[0] Message Index

[#] Next page