KISS 122
KISS 122
Putting the wu back:
The relationship of The KISS Amp
to the Western Electric theatre amps
by Andre Jute
Isaac Newton said modestly that he stood on the shoulders of giants. In his library the giants were immediately to hand, just behind the vellum of their books, hugely visible to anyone of imagination. In tube amps, broadly speaking, the position today is the same. Williamson copies and direct adaptations abound; the next fashion could easily be QUAD II copies from scratch.
It is therefore ironic that in the tiny niche in which the ultrafidelista pretend to be the highest priests of high concept sound, more noble than Charlton Heston himself, the single-ended link is weak. When it comes to claims of direct descent from the Inner Temple, Western Electric itself, the link is faint. But that the link is worn and barely visible does not mean it is not valid.
These views and remarks are not uncontroversial. If you are an ultrafidelista more interested in the warm glow the myth gives you than in the truth, pass on by; there is much more in these pages to interest, entertain and flatter your preconceptions.
In the great days of the cinema as a fixed institution attended in person once a week by the rising middle classes in rich countries and elsewhere by aspirants to American ease and style, the perfect Rolls-Royce of theatre amplifiers were made by Western Electric. Unless one is well-read in the printed literature rather than the gossip retailed as fact on the internet, even the actual brand and class names of the top range might be unfamiliar: it was WECO Mirrophonic. However, contrary to claims one hears both from some ultrafidelista and their generally ignorant detractors, the key, the greatest, the most famous, the most wanted WE theatre amp was not single-ended. It was the Mirrophonic Model 86, which was a 15W all-triode push-pull 300B.
The single-ended amp in the range, the Model 91, today revered as the direct ancestor of all ultrafidelista amps, was the cheaper (nothing at WE was actually cheap, and the Mirrophonic range was WE's premium range) little sister of the more desirable WE86. The WE 91 was further removed from what we would today consider purity by not even being an all-triode amp. True, the single output was 300B, but the voltage amplifier stages in the WE91 were two 310A pentodes. The output was c7W but, as described in another chapter, that is not an audiophile output but an engineer's unsymmetrically distorted measurement under assumptions we would today find very strange indeed.
Both these amps were very high-gain devices, over 96dB from a photoelectric pickup on the film projector to huge, ultra-sensitive horns. With those horns even the 'small' WE91 with 7W was enough to make frightening sound pressure levels in a large cinema. Part of the attraction of these designs to ultrafidelista is memories of the thunderous stampedes in matinees when they were kids. Home cinema is not a new idea.
Both these amps are imagined, by those who have not taken the trouble to look up the original circuits, to be minimalist 300B amps in the modern manner. They are not. To make the gain they were by modern standards rather complicated, multitudes of components all over the place. They were not overbuilt only because WE directors and workers had a belief in giving value for money (that was a given at the time) but because reliability more than fancy features was the intended end result of the huge amounts of money WE poured into institutionalized, permanent, ongoing research. You see, WE didn't sell the WECO Mirrophonic gear. They leased it to cinema owners, forever. When it broke the WE tech had to come service it, forever. That cost real money. WE therefore designed and built their gear to last. Forever.
Thus WECO Mirrophonic gear of the prewar generation of the WE86 and the WE91 came onto the used market only very gradually and in small numbers. (As late as the early 1990s I was only minutes too late to buy a WE91 advertised in West Cork, Ireland. It came out of a cinema that was redeveloped in the 1980s and had been in continuous use until then.) At first the WE86 was the preferred audiophile choice; this was around the age when Williamson and Hafler bestrode audio electronics like giants. The price of a WE86 from the beginning was prohibitive to all but wealthy audiophiles. The WE91 was an orphan, if rather an expensive one because of the Mirrophonic label. Buying a WE91 was the audiophile equivalent of adopting Audrey Hepburn--and her dressmaker.
So how did we get from there to here, where you are under the impression my minimalist KISS amp is umbilically connected to the WE91?
The answer, my dear, is via Japan and France.
You may at this stage want to open or download these illustrations if you don't already have them:
we91Roberts.jpg
T44bis-'Populaire'-crct.jpg A new design by Andre Jute from his Modular Amplifier Series 300B, lightly worked over to be the KISS reference or 'standard good amplifier', known as the 'Populaire', technically a cross between the WE91 and the WE86
FleshBloodFig1.jpg
300Bfiprimerclassic300b.jpg
T39-KISS-300B-Ultrafi-crct.jpg
Joe Roberts, the founder of the Joenet, in its early stages the most knowledgeable tube amp mailing list, and publisher of the late lamented Sound Practices, once related how as a young high-end pusher in his spare time he would turn WE91 into hi-fi amps by simply ripping out a 310A stage to restrain the runaway gain and a good few other components not strictly kosher for audiophile use or actively hostile to audiophile use. This left one with a nasty metal box with ugly empty sockets plus a choke hanging off the back (see below). Since the WE91 was originally an industrial artifact no one except a techie ever saw, it in any event made a QUAD II look like a design prizewinner. The next step was boxing the remnants more attractively for WAF (wife appeal factor).
But I retell Joe's story simply because I know it and he gives it a human face, not because it was by any means typical. By the time Joe came to butchering or salvaging (take your pick) WE91, he was already at the tail-end of a movement that was mainly Japanese, with only isolated pinpoints of good tube light (that is, not drowned by the then fashionable sea of NFB) in an America sinking into a silicon swamp.
The Japanese have always had an intensive love affair with American cinema amplification, not only the amps but the horns as well. Since one needs to be obscenely rich in Japan just to afford the floor space for even a modest amp and speakers, never mind industrial racked gear and cinema horns, cost was not a factor to those who could afford the ne plus ultra of cinema gear, WECO Mirrophonic. The good gear emigrated.
Soon a few lateral thinkers, including some Americans but mainly Japanese, concluded that it would be cheaper and easier just to build a stripped-down audiophile WE91 from scratch. These pioneering audiophiles most often used Tango iron because it was available and it was good. On the outermost crazed fringes of this insane micro-niche, only Partridge iron was good enough but there is no doubt that Tango iron was already a huge improvement because, for audiophile as distinct from cinema use, the WE91 iron was marginal. In the mid-1990s I had a pair of Peerless TFA-204 which were claimed to be an exact copy of the replacement amps specified by WE for the WE91; I certainly was not bowled over by their quality; they were lightweights about a quarter the size of the C-core Lundahls I like. Today I would not build anything except a beginner's amp with such inadequate transformers; there is no reason to believe that transformers of the size found on the WE91 will ever be suitable for modern audiophile use, no matter how well made.
Other significant changes were made in translating the WE91 concept--or its myth, or its core practical advantage, take your pick--to a usable consumer amp on its way to becoming an ultrafi amp. The input transformer went because 30 ohm balanced input is unnecessary. The feedback resistor went. (It was usually uncoupled to let the amp run open loop even by the owners of real WE91.) A choke was added to the power filter because--wait for it--the field coil of the projectionist's monitor speaker was used as choke by the original WE91. Of course the proprietary five- and six-pin sockets were exchanged for modern phono connectors and speaker terminals.
At this point it became a domesticated WE91 with audiophile custom mods. It was known to a tiny, tiny elite of audiophiles, perhaps as many as a few handfuls in America, perhaps a few dozen or a couple of hundred in Japan. (We'll never know the real numbers and I wasn't there then. However, we can use hints intelligently. For example: Decades later, when audiophile obsession was much more widespread in Japan, many of the faithful heard ultrafidelity amps only in Sakuma's restaurant. Even when the Japanese have the money for such expensive pursuits, even if their wives do not object to the fire hazard in wooden houses, they do not have the space.)
Enter Jean Hiraga, a French journalist and audiophile working in Japan. He was intrigued by the Japanese obsession with sound reproduction. He carried the design of the modified WE91 back to France. It has been conjectured by insiders that until the coming of the internet the WE91-300B concept as the peak of audiophile bliss probably claimed more adherents in France than in the USA. From France it was reimported to the States to a much more general audience than the few handsful of the faithful who kept the torch burning through the dark decades. First hundreds and then, with the spread of the Internet, thousands of people learned about the WE91 clone.
After the initial reverence wore off, people started thinking about improvements. Remember that we are talking about DIYers, at a time before their best designs were commercially cloned. Remember too that they are now a decade or two removed from the true pioneers: they are starting from a base which has already lost one pentode and all the negative feedback from the WE91.
The first improvement commonly made was to replace the pi power filter with a choke input filter for better noise rejection.
Then people started wondering whether if one triode was good, all triodes would not be better. After all the upmarket sister-amp, the WE86, was all-triode. This line of thought was given much impetus by the topcap on the 310A, a dangerous device in a domestic setting; its six pin base was also becoming difficult to source. In fact, the 310A was often replaced by a 6SJ7 for precisely those reasons. But why not go all triode and win more silence and the generally better quality sound triodes deliver in zero negative feedback amplifiers?
Thus it came about that I published a set of a dozen 6SN7>6SN7>300B designs collectively called The Modular Amplifier 300B which could by simply connecting the expensive parts (iron and tubes) differently give the audiophile an introduction to 95% of ultrafidelista practice with only one spending jolt. (The other five percent is split between the KiloVolt Brigade of the Reckless and the Ultrafinicky MicroWatt Heroes with their own cinema horns. I'm talking here of fractional minorities even within the niche within a niche of single-ended audiophiles.) One version of the 6SN7>6SN7>300B from my Modular Amplifier Series 300B has been developed into the 'standard good 300B amp', more formally the T44bis 'Populaire' described elsewhere in these notes. J C Morrison of Fi published another of 6SN7>6SN7>300B design and Herb Reichert a third, called the Flesh and Blood, a descriptive name of what these amps will do for your musical experience with the right speakers.
Note that two out the three people who had the same idea about the same time, Reichert and me, either specified or leaves the option to the DIYer to employ a higher output impedance than the hitherto standard 3K; Reichert likes 5K (at one stage he imported Tango iron to the States) and I suggested the 5K6 tap on the Lundahl transformers I recommended for their price/performance ratio.
By comparison to the original WE91, we had by the early 1990s lost the input transformer, changed the input and driver stage tubes to triodes, lost the global negative feedback, added an onboard choke to the power supply or redesigned the power supply altogether to be choke input, and put a higher load on the 300B. To make all his work we need to run the 300B at a higher voltage and plate current than WE did, and some of us then went higher still in search of inky-black silences between notes.
At this point we have broken with WE, whose primary imperative was reliability. It is not that our amps are unreliable or that we burn out 300B tubes particularly quickly--a wellmade 300B is an industrial-strength tube--but there is no denying that our primary imperative of hedonism (achieved by Reichert and me through the silence of high voltage, high current and high load, and by Morrison through tricky direct coupled design) is fundamentally opposed to the standard practice of the makers of our starting point.
It is really very doubtful that a WE engineer, looking into the pretty box housing our pseudo-WE91, will recognize our amp's heritage without an extended explanation.
So, by now we are all a very long way from the WE91, though we all honour it explicitly as our inspiration.
***
By the time I finished publishing all the dozen or so versions of my Modular Series 300B, of which the original T44 was one, I had several other 300B amps to hand, or had recent extensive experience of them. Between my Modular Series and the other amps I built from scratch or kits or tested or reviewed, I had experience of virtually every possible 300B configuration, some of them pretty rarified. (One, my T30/33, depended on the availability of vintage chokes I used as the grid leak for the super liquidity of its sound. I should have kept back one of those chokes for a winder to copy.)
From this wide experience my major impression was that it is very difficult to build a bad 300B amp, almost impossible if you stick to conservative practice. (I say 'almost' because I have knowledge of one amp that was not only dull but also unnecessarily distorted, in a word wretched. Its designers were jerks too impressed with what they thought they knew to take the advice freely available on the internet.)
All these very different 300B amps were excellent overall and most were superb at several aspects of my desired musical experience. It is doubtful that the original WE91 would in such company sound as impressive as once it must have. Though we spout much nostalgia, and often imply carelessly that tube-made music is better because of the history, because the mantle of mythology falls on our amp directly from the hand of the WECO the goddess, the concrete truth is that tube amp design and understanding has advanced some small but significant way, that components have advanced vastly, and that our ears and experience have advanced totally out of recognition to your 1939 consumer of music watching Gone with the Wind. In short, our viewpoint has changed. It isn't that the WECO Model 91 has grown less capable, it is that our expectations have changed, and that with our expectations our opportunities (components, sources) have improved so that the amps we can build now are better matched to modern expectations.
Consideration of the various wildly contrasting schematics all claiming or implying descent from the WE91 led me to wonder if some of the designers were as willful as the pair of idiots described above--but luckier. I called several of them and found all of them immensely open and willing to talk to me at length. All of them possessed impressive electronic qualifications and/or experience. They weren't idiots who lucked out. They did exactly what I did: a proper paper design, discussion with peers, a proto tested and rigorously developed, some kind of peer review of the result, then the final version.
The WE91 is thus an inspiration, not a constraint.
The Japanese have a concept of wu which they took over from the Chinese. Wu is generally inadequately translated as 'the soul of the artist as revealed in his work' or even 'the luck of the artist'. If you speak Japanese even badly, it soon becomes clear on questioning that wu is very much akin to the Calvinist concept of predestination. I am both a Calvinist and, as a painter, novelist, designer of artifacts and goods, an artist. A Calvinist is a swan: what the uninitiated consider luck visible above water is the result of ten times more effort invisible below the surface; every Calvinist I know earns his luck the hard way. An artist is a craftsman who has mastered technique to the point where it becomes an irrelevant fraction of the aesthetic value of his work; when that happens his imagination has taken control and he produces art rather than just a quality piece of craftsmanship. (I have nothing but contempt for those who believe that an artist is 'made' by 'living an artist's life' and displaying the resulting dirty laundry as 'art'. But that is another story for another day.)
Previously I had chased AC balance in PP amps to some kind of a bitter end where I realized it wasn't all that big a deal, a few intervening holy grails with a variety of implications in the more or less successful outcomes, and then silence in 300B amps and kilovolt class transmitting tube amps to the extent that my results in both were superior by any standard. Silence, as the inverse of distortion, is, given correct design of the power supply for the amplifier topology, effectively the same as linearity.
The most 'popular' (i.e. cost-effective) of my 300B designs were in the lower end of the Modular Amplifier Series 300B, 6SN7>6SN7>300B>high impedance load, about as far as you can go from a WE91 and still claim kinship; at the wilder high end of the Modular Amplifier Series 300B the only connection to the WE91 was the input transformer and the 300B. Was it time to take a fresh look at the WE91?
Could I, by relaxing the mechanical requirements long since mastered which had given me an excess of silence, increase the wu in my amps? If you're uncomfortable with the semi-mystical phraseology, try this: Could I use my newly-confirmed technical expertise to give added value to already good amps?
The obvious next question is: What more can you possibly want than an amp with 6SN7>6SN7>300B>high impedance transformer? Anything more linear, more silent, more precise, will require topologies (PP) or devices (transistors) which bring with them the unpleasant side-effects of negative feedback, or development which brings the lesser downsides of more transformers, cost, space and weight. I had been down all of those roads with push pull tubes, transistors and lots of iron (the Zero-Capacitor Option, the Extra Constant Current With Everything Option). I had also been down the road of clean power, right up to a bigger tube to shunt every plate. What I wanted was not something more but something different and better.
So the question shifted laterally: What is there about my existing 300B designs that can be highlighted or enhanced?
Well, I read quite a bit of history, one of my favourite teachers taught economic history, some of my best essays are in economic history, I have even written two novels that revolve around known history (Reverse Negative, The Zaharoff Commission), because of my background I have a personal connection to quite a bit of history, and we have just taken a detour through Japan where ancestor-worship is actively practiced.
So, after an entirely personal consideration, for this is nothing but a personal choice, what I could possibly strive for was a 300B amp nearer in spirit to the original WE91 than the technocrat's delight of the vastly successful 6SN7>6SN7>300B, which was (in this very circumscribed context) already pretty common after only a decade.
After that it was easy and a lot of fun. First, I wasn't going back to a pentode driver, but simply to a single driver, as those early adopters and adapters of the WE 91 did. It had to be a Western Electric tube. The obvious choice was the 437A or the 417A; I already had a stash of the latter. These are impressively linear tubes though not in the 6SN7 class (but then no other small signal tube can touch the 6SN7). But that did not matter as I had a margin of silence to play with. Furthermore, I already had experience of the 417A as remarkably warmblooded and pleasing tube in single-tube 'potato' amps and as a capable input tube to 300B in booster amps to transmitting tubes with high drive requirements (where the 300B amp is used to drive an 845 or similar--think of a sort of super pre-amp with a 300B as the final gain stage). Even better, the 417A is capable of well over 20mA. This would be important because I wanted to reduce the maximum dissipation of the 300B to somewhere nearer the spirit of the WE91. I knew from experience that the lower (left and down on the Eb-Ia-Eg curves) you run a 300B on the transfer function, the more important it becomes to put quite a chunk of current on the driver to deal with Miller. If you don't, the output loses that final edge of clarity which distinguishes a 300B from almost all other tubes. (It is a loss so subtle you need lots of 300B experience to hear but it is there, and you can calculate it as a bandwidth number.)
The result was my T39 'The Tradition' Mk VI, for short the KISS 300B 'Ultrafi'. Does it measure as well as the blameless 6SN7>6SN7>300B? No, it doesn't. It can't because I deliberately sacrificed some of the precision for greater warmth. Does it sound better? Depends who you are and what you expect. Americans previously accustomed to the banksa6550/NFB paradigm may well prefer the drier 6SN7>6SN7>300B. I think the KISS Ultrafi sounds better but then I like slightly warm amps. Japanese audiophiles who like what they call 'British sound' may well consider it rather on the precise side of the tropic and wonder if I have detuned it enough.
But the main question is, Does this amp have wu? Yes! In spades, all the character you could ever wish for. Whose wu? Mine, obviously, but I also like to think I can hear the wu of the humming research laboratory at WE in the 1930s. If those guys were alive and making audiophile tube amps, this could be their amplifier for that tiny niche of the market which knows its own mind and can afford the best.
THE VOLTAGES IN THIS AMP WILL KILL YOU.
GET EXPERIENCED SUPERVISION IF IT IS YOUR FIRST TUBE AMP
All text and illustration is Copyright © Andre Jute 2005
and may not be reproduced except in the thread KISS xxx on rec.audio.tubes
