Communications Cable for IceCube and IceTop... and maybe RICE too...

Overview

Twisted quad cable carries power from the 'counting-house' to pairs of DOMs deployed deep in the antarctic ice. The copper wire the cable is built from has resistance, which has consequences for us in two ways. The DC current through the cable times the resistance of the wire gives us a power loss (loss because it performs no useful work for us, it only heats the ice). The communications signal flowing through the wire is also attenuated by the wire, but in a slightly more complicated way. The 'skin effect' losses vary with the square root of the frequency. In our particular case, skin-effect losses dominate over the dielectric losses in our range of frequencies... below about 2 MHz.

A_/km = (143.7)(5.08/d)ν^1/2/Z₀ + 91.2νε^1/2F_{p is a high frequency model for open
wire line}

is the attenuation A, in dB per km of cable of impedance Z₀, for frequency ν in MHz, wire diameter in mm, dielectric constant ε, and dissipation factor Fp.
It's worth while to note that attenuation decreases with increasing wire diameter and increasing impedance. The root frequency component enters due to the skin effect. At very low frequencies the attenuation is simply expressed as the division of the signal between the DC resistance of the transmission line and the termination impedance. Dissipation factors vary depending on the choice of insulation.

Performance Measurements in the Lab (field measurements below)

Twisted Pair Cable measurements
    Brand G: Preliminary measurements (good, but done better with practice)... Attenuation only, with a hint of cross-talk thrown in...
    Brand E Measurement: Spreadsheet 1 of workbook... This cable is made with 0.7mm dia. wire. (not measured partly unreeled)
    Brand G Measurement: Spreadsheet 2 of workbook... This cable is made with 0.9mm dia. wire. (measured partly unreeled)

9mm to 7mm cross-talk comparison

Hub End and Tail End 0, Tail End 1 and Tail End 2 cross-talk in response to pulse input.
The above is an oranges to grapefruits comparison. The 0.7mm cable is only 1.7km long, the 0.9mm cable is 3.5km long.

Ericsson cross-talk graph

Measuring methodology and definitions:

Attenuation (dB) = 20 log10(V_out/V_in) when driving one end of the A-C pair and measuring the other end of the A-C pair.
Near End Cross-talk = 20 log10(V_B-D/V_A-C) when monitoring the B-D pair at the same end the A-B pair is driven
Far End Cross-talk = 20 log10(V_B-D/V_A-C) when monitoring the opposite end of the B-D pair with respect to the driven end of the A-C pair.

Other considerations

When using an using the math function of an oscilloscope on two of its channels, any mismatch in the gains of the two channels will contribute to an error in the measurement. To avoid this problem, it was important to make two differential measurements, one with reversed inputs, and average the two results. In a few cases this resulted in real corrections.

Performance Measurements in the Field

Quad cables installed in the field are bundled with many more cables, in a linear configuration demanded by the application. AMANDA string 18 provides an opportunity to measure the cable in-situ using digital optical module hardware. See In-Situ Tests...