[Radio Wrangler]

Screens in mains transformers are good over part of the spectrum, but above that the inductance of their ground connection forms a common element in the capacitive input loop and the capacitive output loop.... and then you have resonances as well.

Broadband RFI filtering is an impossible task. The frequencies you want to pass are well-defined and easy, but those you want to stop are limitless.

Repelling low frequency boarders requires large inductors and large capacitors, tending to infinite as the frequency tends towards DC. This isn't so bad. Transformer isolation is fine down here and a screen extends the range usefully Regulated power supplies take out the very low frequency that Ls and Cs can't touch. Once you get higher in frequency and want to repel RF, then transformers with screens run out of steam. Mains filter modules are the usual weapon. Ferrite beads come into their own.

In the late 70's/early 80's Germany started implementing new rules on RF emissions/susceptibility that were a bit fiercer than anything seen before. I was given the task of going through a number of existing HP products and bringing them up to the new standard so we didn't lose sales in an important market. I had to develop cost-effective techniques for fixing problems which also could be done without major redesign.

After the learning curve was climbed, I found it wasn't hard. If such measures were implemented at the design phase, they weren't terribly onerous or expensive. Done later, they certainly cost more.

What is missing in the audio industry seem to be people experienced in these areas, and people who view taking these precautions as simply good housekeeping.

The Agilent Noise Figure Analyser covers 10MHz up to 26GHz in basic form, with extenders to higher frequencies. It can measure the noise contributions of transistors, amplifiers, receivers etc. It measures noise figures of cryogenic amplifiers which use serious cooling to reduce their noise. It can resolve noise down to the quantum mechanical floor at a few dozen Kelvins. This is very sensitive... there isn't much more sensitive that anything can go. There are TWO switch mode power supplies inside its cabinet, and a microprocessor system. It doesn't just have to repel outside noises, it makes enough inside. I originated this box and led the hardware design. What I'd learned fixing existing products 15 years earlier came in rather useful.

Twisted pairs routed together, if the twists have the same pitch can couple quite strongly. Offset one half a pitch then the coupling nulls. Want to add a third pair? then you're in trouble. No offset works, you need a different pitch.... but then you get eventually forced into very long and very short pitches as you add more pairs. There is another way! Look up 'Walsh Functions' used to swap over telephone line pairs to avoid crosstalk.

I'm not a fan of woven speaker cables. They work for RFI avoidance, but they add stray C and that C is good to rather high frequencies. Some domestic audio amps aren't stable into such loads. Driving capacitive loads is a field in itself. The classic case was Naim amplifiers destroying themselves into capacitive cables. Someone thought trying to make cables with 8 Ohm characteristic impedance would be nice and fancy weaving was needed to get the C/length up high enough. Those amplifiers didn't have the traditional 'Zobel' network on their outputs. I think the designer didn't like them or maybe wasn't RFy enough to know what they did.

All this said, the pricer seen on some cult cables are ridiculous, and when pseudoscience gets wheeled out to justify things, it's risible.

Wisible? Always look on the bright side of life.... Fare thee well, Terry Jones.

David