[Radio Wrangler]

Where do I begin.........

Once upon a time, amplifiers were designed by engineers, using maths and normal scientific principles. At first there were plenty of things they didn't know, but as problems came to light they worked to analyse and circumvent them.

Speakers were relatively tame beasts and people traditionally wired them up with bell wire from woolies.

Time passed. People wanted louder music, and the ohmic drop in Woolworth's finest looked like it was worth reducing with some thicker copper. Not terribly thick, just a good bit thicker than bell wire,

Meanwhile great advances were being made in marketing techniques. Some companies overdid the copper size increase thing and made claims that it would sound even better. here was an outbreak in gigantism in the connectors as well. This trend has continued to the point where cables are the size you could jump-start a tractor with and the connectors are finely machined engineering jewellery. It's almost as if people aren't aware of the differences of electrical and mechanical energy and the differences in ways of coupling them. Again there are plenty of claims of audible superiority.

Human perception is famously susceptible to immediately prior experience and other things around for contrast. Stare at a blue wall for a few minutes, and your colour perception will be skewed for several minutes. This is a characteristic of our eyesight and of how our brains handle it. No big deal. The sense of smell is the same. Go work in a cow shed for a day and you'll stop noticing it, but maybe you shouldn't trust yourself to check perfumes for a while.

For some reason, in the hifi world, the comparable psychological effects on the perception of sound must not be admitted. You mark yourself out as inadequate if you admit your perception is not infinite and absolute. Yeah, right!

No-one dare admit "Well, I can't hear any difference." So the stage is set for a showing of the Emperor's New Clothes.

So far, I'm just considering increases in the cross sectional area of copper.

But that wasn't esoteric enough for the marketeers. Something more exclusive could be more profitable. But they wanted some scientific background to give it some legitimacy.

Someone had heard of transmission lines, commonly used for radio frequency connections, say between receivers or transmitters with their antennae. These transmission lines are real and serve a necessary purpose when their length is anything more than a small fraction of a wavelength at the frequency concerned.

I've used them professionally to frequencies of tens of billions of Hertz. Things get more difficult up there, conversely, things get a lot easier as you go down in frequency, and 20 kiloHertz is a long way down from millimetre waves. Audio wavelengths in cables are about 9km at 20kHz.... now, how big is your lounge? But the audiophile world has evolved rules and the number one consequence of believing that there is no limit to your hearing's capability is that there is no imperfection which can be considered so small that it does not matter.

So your speaker cables need to be designed as transmission lines. Transmission lines have a natural 'Characteristic impedance' usually abbreviated to Zo. Oooh, so what Zo should be chosen? Well, the loudspeakers are 8 Ohms... it says that on them right by the terminals. Let's make some 8 Ohm transmission lines! the impedance Zo is set chiefly by the ratios of the spacing of the conductors to their diameters, and the dielectric (capacitive) constant of the plastic/air/whatever insulation between them.

8 Ohms is a problem. The spacing would have to be tiny, In fact you have to take a different approach of interleaving the two conductors within each other. You divide each conductor of your pair up into a lot of thin insulated strands in parallel and you plait them in and amongst those of the return conductor.

This looks GREAT! Do the strands in different colours. Have fancy braiding schemes. Dress them up with some fancy woven jackets, with elegant end pieces to stop them fraying and lastly, most importantly rev up the prices to the limit. You'll need some slick marketing words to go with them, and above all a slick, slightly clever name. Plenty of people will swear they hear a difference. You'd think, the way this tale is going, that the difference is psychosomatic....

Well, it wasn't always imaginary. Sometimes the listeners really did hear a difference.

The best valve amplifiers from the end of the valve's reign relied a lot on feedback, but the amount they could use had to be limited because stray effects in their output transformers would build up unwanted phase delays well above the audible band, but before all the amplifier's gain had rolled off. This risked instability. The designers simply had to limit how much gain/feedback they could use. The losses in the output transformer helped by buffering just what the valves could see at much higher frequencies. The dragon slumbered.

Transistor characteristics, like valve characteristics are curved. But the transistors' are more sharply curved and there is the issue of switching devices on and off round the cycle for class B. So transistors need a lot more feedback than we ever did with valves. This is OK, because with transistors we can easily do away with the output transformer. We can have lots of feedback if we want it.

We can have a wideband amplifier and tons of loop gain, and lots of feedback. Distortion figures are pushed right down, the response is flat over a very wide band, and the output impedance is very small scoring points for 'damping factor'

There is a problem.

Such feedback amplifiers don't like working into capacitive loads at high frequencies. The capacitance load creates an extra pole (mathematical term for a point at which a roll-off and a phase lag come into play) in the response, and this acts within the feedback loop. OOps, the stability margin just flew out the window. The amplifier will now oscillate at rather high frequencies, far above audibility even for bats. Things get hot, bothered and go up in smoke (or just pop internally with no external visibility)

So, weren't the designers aware of this?

Indeed they were. What's called an 8 ohm loudspeaker is a joke. At some frequencies they drop to a few ohms (nearer 2 for Isobariks) and at others they can reach over a hundred. But these instabilities we're talking about are far above audio. Your tweeter has turned to a piece of cheese and all the parts in your crossover are acting mostly like their strays. Your 8 ohm cable is not seeing an 8 ohm load, so neither does your amplifier.

If you have a woven (Litz wire) speaker cable, then they do tend to look strongly capacitive at radio frequencies. Straight-forward cables look lossier, and aren't so dangerous but still could excite oscillation.

So the designers added a little network between the amplifier proper and its output terminals. It's job is to disconnect the amp from the speaker cables in the frequency region at risk, and also to bring in some damping to help out. It has an L-R parallel network from amp to terminal, and a series R-C network from amp to ground.

People tend to call this a Zobel network. Otto Zobel was a very smart cookie at Bell Labs. He worked extensively in network theory and filter synthesis, doing original research. THere is no one Zobel network.... there is a book full of different ones.

So, all was OK, until a boutique amplifier manufacturer decided he didn't like having that network there. It offended his sensibilities. It got between his beautiful brainchild and the speakers. It must be doing no good. Most speakers and cables were OK without it, so he sold amplifiers with no output isolation networks.

Some other manufacturers picked up this trend, some scaled back the networks they still had.

The fates chose this time for the esoteric speaker cables to enter the scene.

The rest is history

Who says the fates don't have a sense of humour?

But only few people understood what had gone wrong and why. It survives as rumours and legends. A few people remember that Naim amplifiers and various esoteric cables didn't go together, but there is plenty more vulnerable stuff.

If you use moderately thick speaker cables and don't need a motorbike to get around your lounge, and you can really really hear differences in speaker cables, then one setup or the other has something going very badly wrong.

David (Both typing fingers now sore!)