sound practices reading club /33

Today, we have something unique: a sound practices ghost article. It never appeared in print and it is also not on the CD. But it turned up during my research for issue 17 articles, before I had the CD (Joe kindly sent me one, recently).

a gift from Joe Roberts, SP editor: here is the ghost article.

The article is called The fixed bias Story, by Herbert Ravenswood, is originally from 1958 and was scheduled for sound practices issue 17. Or so it says in the pdf. In it, Herbert points out that on paper, fixed bias delivers more power, but that listeners disagree. He then investigates, starting with cap coupling and the shift of bias point. Next, it is shown how fixed-and self-bias handle this, leading to the conclusion that self bias amps can play louder. Then Herbert shows that direct-or iron-coupling makes fixed bias behave and the interrelationship with feedback. He finishes by sprinkling some more question marks around contemporary engineering practice.

practical note: it looks to me that this article never went through final proof-reading after having been transcribed from the original. Here are my three edits that make the text logical:

1. page 1, top of third column, second sentence: ‘…allowing plate and screen current to rise, under the zero signal condition…’
2. page 2, bottom of first column, paragraph above Biasing effects: remove ‘very extra little input is reached,’
3. page 3, top of the first column, start of the second paragraph: ‘In the self-bias amplifier, …’

teaser quote: ‘Many amplifiers have been built, using fixed-bias conditions to get a bigger output. But, for all this evidence of the advantage of fixed bias, probably an even larger number of people have made comparative tests of amplifiers using these different circuits. Their impressions frequently contradict the test-bench figures.’

my take

Todays episode inadvertently started last week. I thought we were going to talk then about 2A3s and flea power. But after diving in, that had to do some time-sharing with basing arrangements and driver stages. More of that today.

Even if, like me, your interest is class A, single-ended triode amps, there is enough to take away here.

What can be learned from Herbert’s argumentation, is that by rolling back a part of audio engineering ‘progress’, one can avoid a handful of ills. First, let’s ditch class AB or B operation, which gets rid of the crossover distortion. This allows us to take out the feedback, which avoids worsening the clipping behaviour. Next, improve B+ regulation, stabilising amp gain. This can be done passively by reducing the DCR of the B+ supply, many builders and designers have been (religiously) on this trail. In short, pure class A, no feedback, low DCR supplies: a familiar mantra.

One could ask if we could avoid one final ill by simply using self bias instead of fixed. We have seen this being argued already, for instance by Arthur Loesch and Herb Reichert. Good question. The one reason for me to to be interested in fixed bias is also a big one: not having to use a capacitor on the kathode of the output tube. Losing that smearing is worth all the trouble.

About the kind of driver stages that make fixed bias work: first I though the common denominator was ‘no coupling caps.’ But actually, the message is no high resistances on output tube grids. It is this resistance that turns grid current into a bias shift, the coupling cap stores it and smears it out.

Certainly on the last page of the article we can see that Herbert Ravenswood had his heart in the right place. He remembers the time that high fidelity was amplified with a pair of triodes, no feedback, and it worked. Still, we can clearly see that already in 1958, it was non-negotiable that amps were engineered using class AB, push-pull pentodes with feedback around everything. It stayed like that for a long while, until the rise of ultra-fi.

Now go and read the article, see you next week.