Page 1 of 2
DEVELOPING WORLD AUDIO DESIGN 300B VALVE AMPLIFIER
(See WAD 300B AMPLIFIER for original design concept etc)
This is an unaltered copy of the original 1994 article, re-published for the information it contains. We do not now supply kits, or parts, and cannot help with supply or repair.
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 transformer sets. Price is currently £495 per set (2 outputs, 1 mains, 2 inter-stage, 2 chokes) inc VAT and shipping in the UK.
from Hi-Fi World, September 1994 issue (not the Supplement).
We didn't know how our original DIY 300B valve amplifier design would be received. In the event we were swamped by interest. Rare and esoteric this amplifier may be, unpopular it was not, especially in the Far East and the States. The 300B valve has more of a reputation than we knew.
Our first prototype worked well, but production engineering a kit is another matter altogether. Long term reliability and safety have to be taken into account, which brings in a host of extra considerations. It took us roughly six months to overcome some of the most peculiar and intractable problems I have ever come across. Here's a brief history explaining the delays, why the design has changed and what our 300B kit, now in stock, has to offer.
I thought the original driver transformer screening cans were Russian surplus tank parts. Made from steel 3mm thick, I'd never seen such construction. "Why does 300B have to survive the next nuclear war?" I asked their designer facetiously. "Good driver transformers are very difficult to design" I was told. "Every so-and-so will try to copy mine so I want you to pot them in cans that are impossible to saw open. Put steel ball bearings in the potting compound so that if they get through, the balls will break the blade". It sounded like something from Raiders of the Lost Ark.
Andy was down on his luck. "Can you design driver transformers?" I asked. "Yes. I've done quite a few". ''Then I've got a job for you".
All designers have their own ideas and Andy Grove had his. For the production version he felt things could, with benefit, be done differently. The ECC83 front-end went out, replaced by a 6072 which is designed for low noise and microphony. "It's similar, but it has a clearer sound".
The ECC82 driver valve was replaced by a 5687 which dissipates 4watts per channel. "It's dead linear, needs less drive and has a low output impedance".
The all-important driver transformer posed some interesting and difficult problems. Getting a symmetrical square wave out of the thing was the crucial test and finally, with a particular winding arrangement, we got it, together with good audio bandwidth - without resorting to feedback. The use of driver transformers is a major difference between our 300B amplifier and most others.
We got the new prototype up and running, but with large amounts of heat being dissipated in a relatively compact chassis, running tests showed that after 12 hours or so the chassis was getting hot. The chassis was redesigned.
The new chassis was fine, but during soak tests a GZ34 rectifier started arcing internally, making some fearful cracking sounds; then another went. Whilst original Mullard GZ34s, rated at 550V, had worked perfectly, Chinese and Russian versions, we found in these tests, had a short life of just weeks. The original Mullard GZ34 was a very precisely made rectifier with small clearances, in order to make it compact, minimise losses and reduce heat production. Modem versions do not match it, so we chose instead a bridge rectifier for reliability and to cut electrical losses and heat production. A GZ37 was employed as a slow start series diode. From that moment on, our power supply problems vanished. There are plenty of original GZ37s around, beautifully made to military spec. and priced at just £5 or so.
The first smoothing capacitors needed to be increased in value to reduce ripple on the H.T. line, so we resorted to electrolytics, retaining polypropylenes for use after the choke, where they were needed most to pull the H.T. line down to ground with respect to a.c. The polyprops are manufactured to our own specification using an audio grade metallised film.
Finally, we ran long listening tests on the new design to optimise subjective component balance. Here we found, for example, that although we all liked the sound of modern paper-and-oil capacitors, Solen audio grade polyprops got close but, by not being leaky, were potentially more reliable. Reliability and safety are both very important of course. Experimenters can substitute components of their choice at a later date.
The re-design and re-prototyping delayed 300B for six months. However, the final design is now so linear that feedback has been removed and made optional - it can be switched in and out on the back panel. I prefer it switched out, finding this gives greatest stage depth and most 'freedom' in the sound. But with loudspeakers having a strongly varying impedance characteristic it is best switched in, to reduce the amplifier's output impedance from around 6 Ohms to 2 Ohms or so.
This amplifier is a fine match for electrostatic loudspeakers, delivering a good 28watts per channel, or 60watts in monoblock form. With 280mV input sensitivity (feedback in) 300B needs no preamp, just a passive volume control, except with LP of course.
Although built like the Forth Bridge, possessing no fewer than seven transformers bolted down to a rigid welded steel chassis, this sort of amplifier is simple and elegant in its circuit topology. There are just four active devices and no feedback - how much simpler can an amplifier get? It capitalises on the innate strengths of the 300B power triode to give a super smooth sound with superb sound stage dimensioning, offering a standard of reproduction not really available from today's overly complex, feedback-reliant solid state circuits.