This review page is supported in part by the sponsors whose ad banners are displayed below

Funny we already knew to be Sasa code for challenging. For the T2+ he had another. Bullshit. That's Serbian for when things that should work don't work. This project had a lot of it. Had he known in Norway's Bergen what he'd learnt since, he'd not have volunteered. Apparently the wicked TakeT load made ultra-specific demands on his circuit and was rather diva-esque about obeisance. The H2+ did not like Sasa's toroidal power transformer, his 20-Henry choke-filtered power supply, his 6S45 driver tubes, his triode output stage, his purist zero feedback or convenient auto bias. The power trafo had to go EI, the chokes became Fet stabilizers, the double C-core output transformers insisted that their silicon-steel cores be ground isolated, the driver tubes became ECC88 and a final <3dB global feedback loop between triode input and single-ended now ultra-linear output stage was necessary to tweak the headphone's frequency response and boomy bass.

The Trafomatic logo introduced by the T2+ has become the official new company logo shown here on a 300B Reference mono.

The 1:1 ratio output transformers grew in size over the prototype units and the final output power into 6KΩ was fixed at 4.1 watts. (Direct-heated triodes with usual 3KΩ Z-out would have needed a 1:2 step-up ratio whose associated sonic compromises rendered their lot unfit for the TakeT.) The fixed bias on the EL34 had to lock at precisely 3mA as anything even slightly over or under reflected poorly on the H2+'s treble response. The coupling caps between valve stages wanted to be Auricap, the power supply asked for Mundorf. The enclosure's original footprint stayed put but grew in height just a tad to accommodate the beefier output iron. After wading through his personal brand of bullshit, Sasa was beginning to feel happier. "Much better than with the Japanese step-up transformers".

Photo from Morten Teinum's PhotoStream


To fully appreciate this gestation, know that in the frustrating absence of hard data from Japan, Sasa was finally forced to take one apart and conduct his own measurements on the TakeT. Only this showed why its performance was so different than his measurements into a nicely stable 7KΩ load resistor. At 100Hz the TakeT indeed measured 7KΩ. At 1kHz it was down to 3.5KΩ but by 20kHz he looked at a shocking 200Ω value to turn regular speaker behavior on its head. Combine that with a high 40nf capacitance; and the inverted resistance curve which instead of lowering in the bass rises and instead of rising or staying put in the treble drops dramatically. We're right back to where we started: a major aberration.

Transformer winding and raw stacked Ply chassis.

Output transformers.

With such drastic impedance behavior it takes no crystal ball to predict a very nonlinear response from a standard voltage-source amp. In this context it's fair to mention that in the 1970s Pioneer already had their piezo-electric SE-700 headphone with a Kynar film diaphragm whose impedance was rated at a far more conventional 4 to 16Ω with a standard sensitivity of 105dB.

T2+ boards.

Being of course fully aware of the ramifications now that he knew the extent of this bullshit—all bass, no treble—Sasa used anger to dig in his heels further. "In the meantime I've made an impedance-correction network. That goes into the small epoxy-potted non-inductive Plywood box to protect as my trade secret. I've also designed a frequency-dependent feedback loop between the ECC88 and EL34 stages. By now all of it looks more like a mad science project than by-the-book amplifier design. But I am as stubborn as a mule."


Sasa had even considered going full pentode mode—which just required a resistor change—but gain proved excessive. And he didn't like what happened to the sound when he increased negative feedback to pad down gain.

The metal casings are for the output transformers, the Ply casing is for the impedance-correction network.

Sasa also admitted that whilst he enjoys sporting competition like everybody else, he'd seriously recommend against other valve gear makers attempting their own take on the TakeT headphones. To do a proper job of it with a voltage-source circuit is like an Egyptian beetle rolling a growing ball of steaming dung up a steep hill.


Regardless, the T2+ could be a game changer. It's impossible to think that the H2+ could be anywhere near the top of its game without such a dedicated amp that takes into full account its seriously nonlinear impedance behavior.

Output transformers in their can.

One feels reminded of the Apogee Scintilla which enjoyed—if that's the word—a reputation as amp murderer. Here it's not an amp's life one is worried about. One simply sees how easy it should be to murder performance of these headphones. One also remembers Takei-San's 80dB inefficient piezo speaker for Sony which never saw production. For such a brutal load any big keiretsu would have enough common sense to predict failure in the market place. With the H2+ Takei-San pursues something just as unreasonable but now under his own banner. No corporate approval necessary.

Primer coat over the amp's Plywood carcass to prep it for the finish lacquer.

Just when I thought our tech facts were a wrap, Sasa had this: "Development finally finished a few days ago. The drawing below shows how I had to leave the impedance correction settings at the back separate for both channels. On the H2+ each side measures almost 10% different. I must suppose that's standard and account for it. With my dual-mono settings both channels can be perfectly balanced. For the NFB settings I couldn't decide whether 4dB or 6dB of feedback were better. Now I let the end user make that determination. I'll start production next week and after some 50 hours of burn-in here at the factory, I'll ship the review loaners to you."


To answer a final question some readers might have about why output transformers when the EL34's native impedance so nicely matches the 'nominal' impedance of this load, "we could have used a simple anode choke. That'd be very similar to an OPT but without its galvanically separated secondary plus coupling capacitors. Or instead of the choke we could have used a very big—almost 30-watt—resistor with again capacitors as the coupling element to a very high 500VDC from the power supply. But in any configuration without OPT we'd need coupling caps. Those I don't like at all. And of course that option is bad for the tone too. An output transformer serves double duty as anode load plus gives galvanic separation of the headphones from the amp. After all, who wants 400V on the head when something goes bad? With our T2+ the H2+ response is now 5Hz-25kHz -2dB without phase rotation. I think very few headphones can equal that. And I learnt how with the super tweeters open 60°, the soundstage gets biggest. 30° is noticeably smaller and beyond 60° the focus loosens up."