sound practices reading club /32

In contrast to last time, today is about an ultra-fi icon: the 2A3 amp. And it’s a triple-whammy. First up is I Never Met a 2A3 I Didn’t Like by Joe Roberts, from sound practices issue 15. In it, Joe goes through his personal saga of how his own discovery of the world of ultra-fi was through 2A3 amps. There is the Shishido 2A3 and how a Joe-built Morrison Micro went to Korea. Building a 2A3 amp has never been easier. The next pages about the Fi, Bottlehead and Welborne 2A3 amps are optional reading. Picking it up from page 22 (Speakers), Joe deals with the fear of flea power. Plug it in and then judge, Joe says.

teaser quote: ‘Glory, this was a mighty fine sounding amp and I’d build another one in a second. The “triode sound” is still really there with PP topologies and…Man, you can really get some low end out of a pair of these tubes.’

my take

Whereas the 7 Watts of a 300B amp are playing it safe, living with the 3 Watts of a 2A3 amp makes a statement. This is not general purpose anymore, something really needs to cooperate: speakers (efficient), room (small), programme material (not bombastic) or listener needs (no pants a-flapping). If one of them applies you are in business. Or when all of them apply a little bit. Maybe the latter is the strongest statement of all: normal speakers, normal room, wide selection of non-plinky-plonky (i.e. non-audiophile) music, just listening and a flea-power amp playing beautifully.

hmmm, all this talk really makes me want to try a 2A3 amp.

next article: Classic Designs: The BROOK Amplifiers also by Joe (I guess), from sound practices issue 2. It presents the schematics of two famous Brook amps plus their ‘grandfather.’

teaser quote: ‘A Brook and a Klipschhorn or JBL horn system would put you in the Editor’s Choice circle back in ’51. […] This was High End before there was High End.’

bonus tracks: here is the Audio Engineering article mentioned, describing the auto-bias.

my take

What I notice is that all 3 amps feature interstage iron. With this and all the iron on page 16 of issue 1, maybe the interstage revival did not actually wait for sound practices issue 10?

The interstages get used in a variety of ways. For the Brook 12A, a tapped choke is used on a push-pull gain stage. The result is full-µ gain and because of the autoformer action of the two coils the AC voltages of the two phases are (nearly) perfectly balanced. It also achieves that although only one of the 2A3 grids at any time can demand for some current, both driver tubes see this demand and deliver it. Years ago Magnequest re-issued the tapped choke.

For the Brook 10C, the primary of the interstage transformer works exactly like the tapped choke of the 12A. Only now the coupling to next stage is transformer (with some low-frequency help of a cap, as Joe notes), the secondary windings functioning as grid chokes. Direct-coupled kathode followers means that the output tubes can be pushed much further into A2/AB2.

For the grandfather amp, the interstage transformer performs pure SE–PP phase splitting duty. The interesting thing is that it is driven parafeed-style: cap-coupled from a choke loaded stage. Knowing that the 6A5 is a 2A3 equivalent, one could call this a Sakuma-style amp. But that would be anachronistic.

pop quiz

How is the (negative) bias voltage of the Brook 12A generated? All the DC current of the input, driver and output stages (from the top of C2 A/B/C) returns through ground and the only way it can get back to rectifier V6 is through R17. Thus the end of R17 that is connected to the rectifier and R16 will be at negative voltage, i.e. the bias voltage. It is self-biassing, but the ground of the whole amp has been lifted to the level of the 2A3 kathodes. The DC current through the 2A3s will vary with the envelope of the music—yes it is push-pull, yes it is class A, but it will—and this makes the bias wobble, as for normal self-biassing output stages. What is different is that the R16+C3 filter of the bias voltage is much tighter (0.08Hz) than what you usually see on a decoupled kathode resistor. But this long time constant also means that the variations get smeared out over long time periods.

For this biassing scheme to work, the 2A3 filaments (H + H) really need to be grounded. The voltage readings table suggests that this was done at a centre tap of the filament winding. Also C3 needs to be turned around (positive side grounded, negative side to R16), or else it will explode.

practical note: bothered by strange component values? For the 12A, MF = µF and MMF = pF. For the 10C and grandfather, the M-Ohm resistors are actually kilo-Ohm values.

last article: Notes on the 2A3, by Alan Douglas, from sound practices issue 6. It describes the circumstances in the audio market which resulted in the development of the 2A3. How it was not a hit in home hi-fi but in commercial audio, and the development of the bi-plate 2A3 we know today. Then Alan tries to debunk the ‘mono-plate myth.’ To round off the article there are two pages of implementation notes by RCA from 1933.

teaser quote: ‘If the old 45 was slightly underpowered for the job, then a new triode was evidently needed. […] If the 45 could run on 250 volts, why couldn’t a tube of twice its output?’

my take

In less than a page of text, there is a lot of between-the-lines insight to inhale. How the (macho-) power race was already on in the 30s; how engineers then respond to quantitative goals by delivering stuff like pentodes and class B; how economics are always the issue; how tube relabelling (few production lines building for many brands) worked; how—as jc says—a tube type is a set of specs, no guarantee on its construction—including the quality details that depend on that, e.g. electric field uniformity (= micro-linearity) or microphonics.

Yes, I think it is very engineering to try to debunk the ‘mono-plate myth.’ Since there is physics (field geometry) to back it up, I do believe Jon Baier in the same issue (Readers’ Forum, issue 6) when he states he hears a difference between mono-and bi-plate 2A3s. Especially since he has lived with both and prefers the ‘pedestrian’ bi-plate.

It may look a bit dry, but there is some juicy stuff in the RCA notes too. Ironically, since the first article makes this the ‘flea power’ episode of the reading club, RCA talks about ‘very high power output power.’ OK, as you can see from the top graph in figure 3, they were not going home without obtaining at least 10 Watts of a pair of 2A3s, which is significantly more (5.2dB) than 3W. Another then vs. now contrast is that today a 26 is a beautiful, delicate pre-amp tube and in figure 2 it is used for the blunt task of rectifying the bias voltage,

The whole operating considerations section is a lot about exploiting secondary effects for getting more power out. First, for fixed bias, the output tubes are biased less linear, knowing that for push-pull even harmonics cancel and morph into usable power output. Then, for self bias, the wobbling of the bias point (see above, 12A bias) is taken into account to pick the best plate load. Speaking of: the more I look at figure 3, he more it looks like that for the given conditions (300V B+, -60V bias) 5K plate-to-plate looks to be the OPT value to use, regardless of biassing method. That is, if you are with the RCA program.

It is remarkable how the differences between the schematics in figures 1 + 2 mirror a lot those between the Brook 12A and 10C, respectively. As described, the bias method of the 12A can be found in figure 1, the DC resistance of L2 (speaker field coil) playing the role of dropping resistor (R17 in the 12A).

pop quiz

A quick scan of sound practices issue 1–17 gives the following list of featured amps that sport the 2A3 (or equivalent, 6A3, 6B4, 6A5) in the output stage: SP2: 4, SP3: 1, SP4: 1, SP5: 2, SP6: 5, SP7: 1, SP8: 4, SP9: 2, SP10: 2, SP11: 1, SP14: 1, SP15: 7, SP16: 2, SP17: 2. Can you spot them all (yes, there are some tricky ones)?

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