It started out very simply. I'd always felt that the EL84/6BQ5/7189/6P14P-EV was the under-utilized and under-appreciated bastard child of the famous direct-heated triodes. Having visited Sasa Cokic of Trafomatic in Serbia and listened to and reviewed a few of his amps, I knew that he was the right man to prove the point. I'd commission him for an all-out statement amplifier to run EL84s. Commercial amps using this tube already exist of course. But given the valve's plebeian perception, nobody I knew had treated a project involving it as a bona fide SOTA proposition - one that you'd throw everything at just as you would for yet another $20,000 300B or 62B amp.


Putting "a contract on your head" in the e-mail's subject header to get his attention, this went to Sasa: "I am commissioning an amp from you. I'm serious. Figure out an approximate price, let me know what it is and we'll do it. Here's the rough sketch. EL84 monos. In pentode, not triode. Ca. 20 watts per side. Single-ended or push/pull I'm not sure. Perhaps ECC99 driver. Direct-coupled (IT). Perhaps input transformer. You're the boss, you know what sounds best. For the look, I like what you do now for Trafomatic. Except I've got a few changes.
"a/ no power LED, just one of your nice knobs for power in the front, nothing else.
b/ a single pane of glass that sits in two short slots cut into the top aluminum plate. The glass plate is as tall as the EL84s, no taller. Two (or four) yellow LEDs fire up into the glass from below the surface to light it up. If you want to get fancy, we can design something to get etched or sand-blasted into the glass.


"The glass will have a little space to float so it only sits inside short slots for a few centimeters on either side and clears the deck in the middle by say a centimeter. It'll look very nice. The glass itself is straight, just sanded across the edges to not cut. Very simple, very elegant.


"Transformer covers in your usual textured black, same wood chassis as you do now. Perhaps a single tranny cover all across, perhaps individual. It might depend on how many transformers/chokes you put on deck. Negative feedback is okay if it does the job. I don't care about the circuit concept, "audiophile purity" and all that, just the final sound. You're the maestro, you decide. The only thing I'm reasonably sure of is that I like the EL84 in pentode better. Perhaps parallel SEP could be the ticket. Or push/pull. I want really good bass control which perhaps favors p/p. And I like that bluesy 'bite' of pentode which separates better than triode and is more crystalline."


Sasa's reply took only a few hours. "Very interesting proposition. You actually thought of something I already had in mind, with four EL84s per channel. Would it be a power amp or integrated amp?"


"Power amp. I love my Esoteric C-03 preamp for tube amps and it's got huge gain (or none at all) to accommodate all scenarios. How much power would the amp make in p/p vs. paralleled SE with 4 x EL84?"


"Yes, power amp is always better for sound. In p/p UL mode, four EL84 can make 25-30 watts, PSE with four EL84 is not a good idea, maximally two are okay but more than two create other problems. Four in PSE triode give us 10 watts, perhaps a bit more. UL PSE is bad, I tried and didn't like it but we could get about 18-20 watts. P/P UL mode is very nice. With separate windings and some transformer tricks, very little negative feedback directly in the output transformer, you can get a real winner of an amp. ECC99 or 6N30-type input tube (in single-ended mode) coupled by transformer to the four EL84 in p/p mode is the best way for sound and of course a tube rectifier - but not the usual 5U4G. About the look, the best way is if you make me some drawing when we meet in Norway at the Bergen show but except for the glass, the rest I will design how I think it best for the amp. Is that okay?"
Of course it was okay. If one wants a masterpiece and has identified the chosen maker, it's vital to give them as much freedom as possible. Nobody creatively brilliant enjoys working in a straight jacket. I already was fully convinced of Sasa's brilliance. All I wanted was to stimulate him to throw overboard the calculator in his head which he—like everyone else— must keep ticking when conceptualizing commercial amplifiers.


I wanted him to design the very best EL84-based circuit he knew how and forget about 'reasonable' when it came to parts. Sasa is no wide-eyed mystic who spends money on silly audiophile designer parts just because customers expect them. His focus is on transformer craft. Premium iron—be it input, output, interstage or power—costs money. And, NOS glass can get costly. That's where I wanted him to feel unrestrained and go wild wherever performance would warrant it. After all, the EL84 is no massive 833 transmitter beam tube that might see 1800 volts on the rail. Neither were we talking high power nor pushing a given set of valves to their short-lived limits. Relatively speaking, extreme in this context would be quite modest when compared to a single-ended circuit with exotic triodes of similar final output power.


For a name, I quickly settled on the Sanskrit term Kaivalya which connotes ultimate freedom or liberation. It'd have a nice double ring - freedom for the designer, liberation for the music. I wanted that name engraved on the top cover right underneath where the glass cover would float, i.e. in line with and between the two cut-outs where the glass would be anchored. I saw a yellow fill for the engraving, in the color of the ambient LEDs. I'm a slut for valve glow and anything to enhance and support it should be done.



Without knowing how many rectifier tubes Sasa would use—his big 300B reference monos used two—nor how many transformers or chokes would have to go on deck, I mocked up the above drawing for just a rough notion. Dimensions, layout and specifics would remain his decision. Importantly, I wanted the industrial design to remain closely wedded to the existing Trafomatic Audio aesthetic. I did not want to complicate this project by going after some outré cosmetics. I wanted Sasa as comfortable as possible to focus on the sound-producing decisions. He had to feel assured that except for the simple protective glass cover, he'd not be challenged on the physical design beyond what he'd already done before. I also told him that I wanted to document this project for 6moons with quotes from our e-mail correspondence, sketches, revisions and such to report candidly on the process. It might encourage others to approach their designers of choice with similar requests and become small "patrons to the arts" to stimulate advances.



One Dinos Theodorakopoulos had earlier contracted with Sasa for a pair of 2A3 monos using EML glass. He was clearly thrilled with how they performed over his Avantgarde horns: "The REF-1 monos are indeed the most natural and sentimentally human intensifiers of emotion which I have heard regardless of cost. Of course they need really sensitive hornspeakers to go to heaven. But after 27 years of daily avocation, this is my final system and this is how it shall remain."

Where Dinos had wanted direct-heated triodes and just 4 watts for his 107dB Avantgarde Trios, I wanted 25 watts for my 91dB Franck Tchang speakers. Already owning first-class 45 and 300B SETs and 130-watt KT88/6550 push-pull amps, it was time for the lowly EL84 to shine as brightly as she could.
Sasa's next email: "I have two ideas but am still thinking. I am making some 'harmonic analysis' between ECC99 and ECC81. The ratio between 2nd and 3rd harmonic favors the ECC81. This ratio directly influences the sound and is critical. It's about how the choice of driver tube for the output tubes influences the sound not just with its ability to drive via current, impedance and headroom but harmonic interactions." A week later, "the concept is coming along nicely but is challenging.


"Single-ended drive of ECC81 coupled by IT to four EL84s. Rectifier tube will probably be an indirect 5AR4 with 15-second delay of anode voltage, an EZ80 for the input tube and some perversions in the output transformer (I never before built something like this) but I hope that it will become the best-sounding  EL84 amp in the world, no joke."


During our visit to Renaissance Audio in Bergen/Norway, I had opportunity to speak with Sasa in person. He explained a few of the planned transformer perversions which would require more than the customary number of windings on both the output and interstage iron. What those windings accomplish, exactly, shall remain a secret. Suffice to say that Sasa believes this has never been done before - and he isn't prone to exaggeration. (And yes, some questions as the one below which I simply had to insert as a photo caption are just too silly to be dignified by anything but silence.)



Sasa also presented me with a very attractive idea for a finish - white piano lacquer with a thick nickel plating for the metal parts. He was very excited about this. Remembering how well nickel wears versus chrome from my clarinet keys, I could see how this proposition would lend itself to an entirely new line of upscale Trafomatic models of which the Kaivalya monos would eventually end up simply having been the first. It only made sense to turn this project into formal production beyond a single commission pair. I loved the whole idea. Sasa also confirmed the basic layout of my earlier sketch. He simply would relocate the taller 5AR4 rectifier behind the three transformer cans close to the IEC power inlet. This would leave the ECC81 and EZ80 front row center and the two Russian military 6P14P-EV behind each other flanking the driver and small rectifier on either side. With all frontal valves being of similar height and girth, they'd also look good together.


The day after we returned from Bergen, a quick e-mail note from Mladenovac announced that Sasa and team were already working on the first prototype of the interstage transformer, with the equivalent output transformer proto due the next day. Besides impedance conversion and perfectly matched phase splitting to turn the single-ended driver stage into a push-pull feed for the output bottles, the IT would also be responsible for the harmonic matching between the ECC81 and 6P14Ps. How could transformer windings tweak THD parameters? Sasa explained it and how he'd do many iterative listening sessions while unwinding specific interstage sections in single turns. That's how critical he expected the tolerances would have to be to hit upon that magic spot.