Ian Jackson wrote:

I think that the principle of this circuit is similar to the
constant-impedance equaliser - such as used to compensate for the loss
of a length of coaxial cable over a wide range of frequencies (very
common in the cable TV world). This is frequency-selective in that it
has essentially zero loss at a pre-determined 'top' frequency (say
870MHz), with progressively increasing loss at lower frequencies (the
inverse of the cable loss). As it has a constant (75 ohm) input/output
impedance, it is therefore resonant at all frequencies from 0 to 870MHz.

I've designed a couple of coax loss compensators, for very high speed
digital oscilloscope delay lines. They had to preserve the fidelity of a
high speed step to within a very few percent, which amounted to very
precise compensation of both the frequency and phase response.
Bandwidths were about 2 and 9 GHz. The dominant loss mechanism in high
quality coax over those frequency ranges is due to conductor skin effect
which is proportional to the square root of frequency, so no single
network will do the compensation. I used a number of bridged tee
networks to do the job, each correcting a different part of the time
response (equivalent to different frequency ranges), in some cases
transforming them to other topologies to accommodate unavoidable stray
impedances due to components and layout. The circuits were used in the
Tektronix 11802 and TDS820 oscilloscopes.

Roy Lewallen, W7EL