Mullard Mustard capacitors

[Mark Hennessy]

Hi Al,

By coincidence, this is being discussed on UKVRRR at the moment. Allow me to quote a couple of my posts here:

There are many differences between valve and transistors amplifiers that can all be objectively measured and discussed. It's too simplistic to dismiss it as "audiofoolery".

I had typed a decent reply which got lost when I clicked on Preview, so in sum, here are the main points:

Differences in harmonic content: we like 2nd order harmonic distortion, which is why we like valves, vinyl and tape. We don't like odd harmonics, nor do we like high order harmonics. Transistor amplifiers produce these for many reasons, not least crossover distortion.

Differences in the amount of negative feedback that can be safely applied: valve amplifiers have much less - transistor amplifiers need rather more to counteract crossover distortion. This changes the amplifier-loudspeaker interface, and as loudspeaker designers assume a zero ohm voltage source, it's not surprising this affects the sound.

Differences in the clipping behaviour: in part, caused by the differences in NFB. When a solid state amplifier clips, it does so with "sharp" edges, that generate lots of high-order harmonics. We don't necessarily interpret mild clipping as obvious distortion, but we do notice the sound becomes "harder", especially in the midrange. By comparison, a valve amplifier rounds off much more gently - gently compressing the peaks in the same way an over-modulated tape does, and it sounds quite reasonable to most people. This is where the notion of "valve watts" comes from, BTW. There is no such thing; it's just that it's a nicer form of distortion compared to the solid-state equivalent.

There are numerous other factors too, including the output transformer distortion and power supply rejection/modulation to name but two. But it's certainly possible to make very nice and very poor amplifiers using either valves or transistors

There are reasons outside of the amplifier too. Because "valve watts" don't come cheap, loudspeakers had to be efficient, which means lightweight paper cones. But paper cones are known to cause colouration, and are also known to be be inconsistent from sample to sample, which is why the BBC started making its own drive units in-house back in the 1950s. In general, the most tonally accurate loudspeakers will be inefficient (I could cite countless examples), and won't respond well to a typical low-powered valve amplifier with a high output impedance. Efficient paper-coned loudspeakers are certainly "exciting" to listen to, but are not the last word in accuracy. Interestingly, paper is making a comeback as a cone material, and I can only assume that modern production methods yield more consistent results today...

And...

I'm not sure that the transfer function of a valve is the primary cause of "valve sound". The relationship between voltage in and current out is broadly exponential, and the same can be said of bipolar transistors (see the Ebers-Moll equation) and similarly JFETs (which are more square law than exponential, but still rather non-linear).

It probably has more to do with factors like the need for simplicity (in a transistor circuit, it's no hardship to throw in a few small-signal transistors to linearise parts of an amplifier - perhaps current sources or current mirrors or EF buffers). The high voltage operation of valves mean that output transformers are essential (although I did once see a direct-coupled valve amp - it was an exotic and expensive beast). The high impedances seen in a valve amplifier contribute to the operating noise, and the lack of current sources and supply regulation in typical valve design compromise the power supply rejection ratio. Going to the effort of building a regulated power supply for a typical valve amplifier is well worth it - I know a few people who have done this.

Some years back, famous designer Tim de Paravicini put forward the argument that it's possible to build a valve amplifier and a transistor amplifier that both sound the same. Hi-Fi News reviewed them - they were commercially available at the time - and concurred with this. It was an eye-opening article that showed that valves and transistors are not as different as people assume; both are voltage-in, current out devices after all. I'm fairly convinced that the differences have more to do with the typical circuits used than the actual devices, some of these difference I mentioned in my previous post.

Hope some of that is of interest,

Mark