Xjyuk

As far as I know, in telephony STP or twisted pair cables are used. This creates balanced line impedances which is useful to mitigate for common mode related interference.

So using balanced cables in telephony and in audio is very crucial to get rid of any EM or RF interference.

On the other hand, in TV broadcasting or many RF systems coaxial cables are used. And most of the coaxial cables I have seen are not balanced. I can see that 50 Ohm concept is good to get rid of reflections in transmission line theory. But how come unbalancedness of coaxial cables have no problem in impedance balancing issues?

But how come unbalancedness of coaxial cables have no problem in
impedance balancing issues?

The beautiful thing about coax is that the shield shunts mostly all electric field interference to ground and the inner wire is largely unaffected. For magnetic field interference a subtle thing happens; the current in the shield due to the presence of an external magnetic field creates a volt drop along the shield and, due to near 1:1 coupling between shield and inner, the same volt drop is present on the inner core. So, providing you use a differential receiver and the sending end has somewhat reasonably the same impedance to ground on shield and inner, the differential receiver can reject the common mode interference.

If you do the math on external fields produced by a regular signal sent down coax, and analysed the external magnetic field from send and return currents individually, you find that at all points outside the shield, the opposing magnetic fields exactly cancel to zero.

The impact of this is that a proper signal’s magnetic field is only produced in the gap between inner and outer shield. A repercussion of this is that the shield therefore has to have zero inductance. This is because the outer mag field is zero and the signal’s internal magnetic field has zero effect on a tubular conductor (aka shield) hence, the shield behaves like an infinitely thick ground casing surrounding the inner.

The upshot of all my rambling is that it works despite having a significantly unbalanced impedance regime between inner and outer shield. It’s not all that easy to immediately see I grant you so hopefully I’ve done it some justice.

Andy talks about how coax works in general, but another point is that video generally doesn't have the same SNR requirements as audio to begin with. Data with 8 to 10 bits per color channel provides very good pictures, and this represents an SNR of only 50 to 60 dB.

On the other hand, to be considered "CD quality", audio must have at least 16 bits of resolution, equivalent to an SNR of almost 100 dB.

Telephony is a special case. While it doesn't require a lot of bandwidth, it does require a dynamic range equivalent to 13-14 bits. (But the coding used reduces the SNR to about 7 bits). UTP (unshielded twisted pair) is used only because it is so cheap to make and so much of it is required.