Hmm . . . I am having a mental block. I
think the channel characteristics function, H, is only determined when the xTU-C and xTU-R, at either end of the circuit, train-up. Hence an examination of plots of the logarithmic presentation of the channel characteristics function, Hlog, at different times
all post the same one discrete circuit train-up will be identical.
The two QLN plots you have shown are identical (I viewed them in full screen mode in a continuous cycle). My mental block continues . . .
The SNR plots show a definite hit in both the lower frequencies of the DS1 band and the lower frequencies of the DS2 band. Which is very peculiar.
In the case of the DS1 band, it is almost as if the DSPBO is being increased (the "bad" plot) and then restored to normality (the "good" plot).