A disappointment I am afraid
The assumption that I have made is that HLlin returns the complex attenuation of the signal relative to a FIXED phase. As one proceeds through the tones the phase W will change as
w=2*pi*f*DT ----- equation (1)
Here the phase w is in radians and the frequency f is in Hertz and DT the time diference due to say the line length.
DT= length/(c*VOP)
where c the speed of light and VOP the progation factor of the line eg about 0.66.
This implies that the angle w given by Hlin's real and imaginary parts should vary as w=2*pi*df*n*DT where df is the tone interval and n the tone number.
If we take a Discrete Fourier transform of the series of w values as function of n and obtain a peak at e.g coefficient number m then there is phase difference of 2*pi at the "frequency" m. Frequency is in quotes as this is the frequency of a frequency change. (If there are errors in this view they will most likely arise over this interpretation of the transform of the w values).
It follows that 2*pi=2*pi*df*m*DT
i.e DT=1/m*df
then length = (c*VOP)/m*df= (3*10power8)*0.66/(m*(4.3125*10**3)) = 0.4591 *10*5/m
In short I have taken the DFT or FFT of the N value of w=atan2(a(n),b(n)) series to give N/2 values of x(m),y(m) and I then plot the length = 0.4591*10*5/m against spectral power (x(m)**2+y(m)**2). (a DFT of either of the separate a(n) or b(n) values will give essentially the same result but with some distortion of the "correct" result)
You will see from the plot above in my last post that this can look very promising. However I found that with an overnight resync the peak can move in length eg between 1200m and 300m so far!!! It does not always move though! The plot always gives a very nice looking plot with a sharp peak at the frequency at which w derived from Hlin is varying. The puzzle is that the frequency of variation changes after a resync suggesting that although the phase of HLIn varies nice and regularily to give a sharp peak it is not varying like equation 1.
I attach the best and worst? of the plots
I don't think anything more can be done without more understanding of why and how the Hlin phase changes.
Author
Topic: Can you generate a crude TDR from router Hlin(f) stats? (Read 22441 times)