Page 1 of 4
WORLD AUDIO DESIGN 300B VALVE AMPLIFIER
Part 1 and Part 2 are reproduced here, from August 1993 DIY Supplement No4
and October 1993 DIY Supplement No5.
See 'Developing 300B' for the updated final design.
This is an unaltered copy of the original 1993 article, re-published for the information it contains, but with better pictures We do not now supply kits, or parts, and cannot help with supply or repair. You can however still buy parts to build one and they are in build today. We sold hundreds of these amplifiers around the world and they are still difficult to better as a platform for the 300B and a pure triode sound transistor amplifiers are unable to match.
See SP Wound Components (http://www.spwoundcomponents.co.uk/) for transformers and the World Design forum for information. SP Wound tell us they still (2011) supply – and are supplying – transformer sets. Price is currently £495 per set (2 outputs, 1 mains, 2 inter-stage, 2 chokes) inc VAT and shipping in the UK.
300B Valve Amplifier, Part 1, DIY Supplement No4, August 1993 issue.
The 300B is one of audio's most linear valves.
Tim de Paravicini designs; Noel Keywood provides an introduction.
This amplifier is our first step into original and specialised design with valves. It uses the highly regarded 300B valve, a low distortion triode designed for high power audio work, now being re-manufactured by Golden Dragon in China. But the circuit is original, designed by Tim de Paravicini to meet various modem criteria. So this is no re-run of an old design. It is a brand new valve amplifier in all respects and has been specified to satisfy current thinking on the subject of audio reproduction. And before I go any further, let me tell you the most important thing of all - it sounds absolutely gorgeous!
What were our criteria? Not to produce a replica of any sort, or even an acknowledgement of the past (although to some extent it is that), but to produce an amplifier that can offer the very best sound quality by modem standards. The notions of simplicity and purity were paramount. It is a minimalist design in some respects, with as few amplifying devices as possible, of basic but well engineered circuits, with superb component quality and possessing the best measured performance possible within these criteria. For those not 'au fait' with valve amps and audio arguments (!) the triode valve is the most linear of all and this is why we chose to use it, come hell or high voltage - just look at that mains' transformer!).
Yes, we do fret about distortion, frequency response and all those things, but we also measure and listen in-house on the magazine, in order to be able to assess their impact. Our view is that not all measured distortions are so bad, or even audible, but that there are insidious, but significant problems that blight modem solid state amplifiers. Trouble is, at present there are few really good valve amplifiers around; it's are-emergent technology and too many circuits in use today are old, and often poor.
When engineers were struggling to design valve amplifiers in the thirties, they looked towards achieving certain basic measured goals, the ones we use today, like low distortion. The 'perfect' amplifier did not then exist, as it does today, so the validity of these goals couldn't be questioned. Naturally enough, engineers tacitly assumed that perfect measured specs gave a perfect amplifier with, of course, perfect sound quality.
I first measured and heard a 'perfect' amplifier somewhere around 1974 and I've been measuring and listening to them ever since. A 'perfect' solid state amplifier sounds little different from an imperfect one; sometimes it can sound worse. For example, I pitied Hitachi with their HMA7600 MOSFET amplifier; it had the most astonishing measured performance, yet nobody liked it much. The Japanese design engineers must have cried!
These days, we have the benefit of such experience. We can compare a 'perfect' solid state amplifier - in fact a range of them - against a modem valve amplifier that is, for all intents and purposes, inconsequentially imperfect and listen to differences. Many people now acknowledge that the latter is audibly superior, even if the reasons remain obscure and open to speculation. The coarseness, apparent modulation noise and distortion (harshness) of solid state may well be related to the fundamental non-linearity of a transistors p-n junction; if we can 'hear' cables then we can also hear this underlying phenomenon I suspect.
Whatever, our 300B amplifier offers a converse design philosophy to that of transistor amps. It uses thermionic amplification; it uses very few active devices; it uses innately simple yet linear amplifying devices - triodes - and it uses just a little feedback. What feedback it does use can be varied or removed altogether - it makes little difference. I'm not anti-feedback, nor is Tim de Paravicini who designed this amplifier, but we both feel that ideally it shouldn't be needed and that an amplifier for music reproduction must be acceptably linear without it.
If say, an amplifier produced 10% distortion at ordinary music levels, we wouldn't add an amplifying stage to get more gain then apply feedback around the whole caboodle to get the distortion down - we'd junk the design and start again! It is because the 300B is one of the few high power triodes designed for low distortion audio amplification that it is so favoured for purist amplifiers like this one - it can be used without heavy feedback. There is a small amount of feedback used in this design, mainly to keep the high voltage drive stage very linear, but the 300B output stage has none.
There is one other factor I should mention that distinguishes modem valve amplifiers from old ones: the use of better grade components. In particular, the transformers we have used are of a very high standard, designed by Tim of course. They have the best thin-lamination, grain orientated silicon steel cores, without comer fixing holes which produce flux concentrations. The windings are specially laid down for optimum audio performance. Even the mains transformer (our design) is efficient and runs cool because of its generous core.
There are some limitations to a purist design like this, ones common to triode valve amps without feedback. Firstly, this is a self-bias pure Class A amplifier, so power output is limited, in this case to 28watts; reasonably efficient eight ohm speakers need to be used. All the same, we found the amplifier went loud without difficulty; in use it didn't seem limited in the way that 7watts (say) designs do.
What we did suffer initially though was a dull sound caused by loudspeaker mismatching (silly us - we were too eager to listen to the prototype!). Unlike solid state amplifiers, a triode amp. without feedback has a high-ish output impedance of many ohms. Connect it up to a loudspeaker that possesses strongly falling impedance at high frequencies and it may sound dull. Our own three-way loudspeaker, also featured in this supplement, worked wonderfully with it, and so did Heybrook Quartets. In tnuth, triodes are less sensitive to load matching than pentodes, but the latter generally have feedback applied, which alleviates the difficulty. Pondering on this, we decided that overall feedback should be applicable to reduce output impedance in order to overcome this problem, should it arise. You'll see an optional overall-feedback connection in the circuit diagram, from output back to the top of R3 in the cathode circuit. More about this in Part II.
There was, long ago, a debate on the sound of triodes versus pentodes where such differences were at the time noted. They still exist today, but the solution zisn't to raise gain and apply feedback by using pentodes, it is to build a better loudspeaker! Triodes are intrinsically more linear than pentodes (they possess more degeneration) and our choice for this amplifier was firmly for the best triode available, the 3008. It is a directly heated valve (no cathode) designed around 1932. Although difficult to make, it has a great reputation. Western Electric of the U.S.A. are most commonly associated with it and undoubtedly made some of the finest versions, although it seems that S.T.C. (Standard Telephones and Cables) of the U.K. may have designed the original around 1932, naming it the 4300A.