Skip to content
Nekolab

Madama Butterfly 2A3 SE Power Amplifier

Let’s get one thing out of the way: I love the 2A3 power tube. With appropriately sensitive speakers, I’ve long found that this tube just gets me a bit closer to the music than anything else. I’ve felt this way about every 2A3 amp that ever crossed my path – and fortunately, I’m not alone in this regard. Ask a group of tube amp lovers which is their favourite of all the directly heated triodes, and you will almost always get a majority in favour of the 2A3. Sweet and airy but still punchy, very musical, most of the midrange excellence of 300b with none of the treble limitations, so the consensus goes. So why don’t we see it everywhere? The problem is lack of power: 3-4 watts or so in single ended mode, maybe 8 or 10 typically in push-pull mode. Not enough, most of the time, for real world speakers. Therefore this amplifier is recommended for those who own high efficiency speakers like my K2 Klipshorn clone or single alnico high-efficiency horn-loaded speakers ( Lowther, Supravox… ).
The first warning I must make is this: this amplifier is very very heavy. Why? The answer is that it is uncompromising device. Please note 90% of tube amplifier in the market today needed double digits of negative feedback to hold the circuit together. Yes with a lot of negative feedback you obtain low distortion, wide bandwith and so on, but …. the sound ? …we spread a merciful veil !!! With a lot of negative feedback you can use cheap light transformers cheap capacitors and so on… anyway the feedback fixes everything. The best audio components unfortunately are also heavy and massive ( the two Tamura F2007 output transformers weigh 10 kg ).I want a circuit with a minimal signal path length and only one tube input/driver transformer coupled with 2A3 grid. The simpler the signal path, the better the result. In my opinion the best valve for this purpose is the russian tetrode 6E5P connect in triode mode: rp around 900-1KΩ, gain 33-35, transconductance 30 mA/V. I used this tube in my AIDA LCR riaa preamp with excellent results.
However in that simple scheme the components the signal encounters better be of the premium-most quality. The interstage transformer is a 1:1 5KΩ custom made type from AEE Automatic Electric Europe ( Netherlands ):https://www.aeetransformers.com/products/audio-transformer.html. This transformer have a double C core, a primary inductance 50 H @ 30 mA current the perfect match with 6E5P 1KΩ anode rp for well extended low end response. Remenber that low frequency in this approach is due to transformer primary inductance and the impedance of source driving. Higher primary inductance and lower source resistance give better LF response. This solution also allows me to use the fixed bias for 2A3 with the advantage of being able to use valves even if they are not perfectly matched. Another important topic is: how to polarize the 6E5P driver tube? Resistor ( Rk/Ck), LED, Diode,….the choice of the right solution greatly influences the final result. I spent many days for audio test…in the end I opted for diode bias : five C3D02060F low dynamic resistance silicon-carbide diodes give me 4,25 V on the cathode, the correct bias for 22 mA through the tube and primary transformer. The output transformer is one of the best audio component currently available in the hi-fi market : Tamura F2007. 10 Watt Max Power Output, 3 KΩ primary impedance @ 100mA max current, 25 H primary inductance, 10-60 Khz bandwith one of the best choice for a SE 2A3 Amp. A clarification about the choice of the 2A3 operating point: adopting the Sovtek 2A3 ( or better still the JJ 2A3-40 ) I think we can easily overcome the canonical 15 W of plate dissipation. Look at the photo below: the 2A3 Sovtek anode is close at the JJ300B plate. This is a super 2A3 as also highlighted by the prestigious Japanese MJ Audio Technology ( 29 cm² anode surface for Sovtek 2A3 vs 34.3 cm² for WE 300B ).When you are using very high efficiency speakers as Klipshorn every little hum is highlighted. Making a non-feedback SET as quiet as a (dead) mouse is no easy task but possible. When I supply in AC the 2A3 filaments the hum was audible and every attempt I made to eliminate it ( even with a hum null pot to the filament terminals ) was in vain so I decided to feed them in DC.The filament power transformer has 3 coils: one for 10V /2A for the 6E5P filaments regulator and two at 3,2V@ 3A for 2A3 heaters. A discrete 10 A Schottky bridge rectifiers ( MBR10100 ) followed by a 33.000 µF capacitor give the 2,5 VDC needed for the 2A3 filament. In the 6E5P filament regulated power supply I used a 5V 3A TO3 LM323k lifted up ground by 2 diode to obtan the required 6,3 V but also LM350 3A variable linear regulator can fit.
The main HT anode power supply use the 330V CT @ 0,2 A secondary winding. The CLCLC ripple filter is followed by a double ( one per channel ) Mosfet ( STW7NK90Z ) capacitance multiplier to give around 300V. The 210 V required by 6E5P driver stage is obtained from a further Mosfet ( STP6NK90Z ) regulator: two 1N5379B Zener diode on the Mosfet Gate reduce the voltage to the required value ( 210 V ). An 80 V @ 100mA winding is provided for the negative Bias power supply.
The calibration of the 2A3 bias current ( after an adeguate warm up time ) is done by placing a millivoltmeter across 1 Ω resistor and turn the bias pot to read 60mV. Brief notes about components: this amplifier is an uncompromising device and each component affects the sound. 10µF first capacitor in the main power supply must be paper ( or polypropylene ) in oil type ( 600 V ), 12µF ( bias ) and 33µF ( 6E5P regulator ) capacitors must be high quality MKP. For gate and grid stopper resistors use carbon composite type (  Allen-Bradley or similar ). All connections must be kept as short as possible and made with solid copper (  teflon-insulated ). For the signal and power groung connections, please refer to the schematics on my post GROUND!!! My opinion about listening test:With female voice, the ability to play ‘you are there’ recordings like Anne Bisson’s Blue Mind Fidelio Music, or Lea Desandre – Vivaldi Jupiter Alpha Records this SE power amplifier produced a startling lifelike quality that bordered on the holographic. In the Serà una Noche by di Santiago Vasquez e Pedro Aznar MA Records the sound is absolutely spectacular, the space produced enormous, the percussion thundering, the guitar textured and the bandoneón liquid and breathy. The shear physicality and visual solidity of the production will keep you enthralled from the start. You are in the room. The incredibly low sound floor allows it to dig out immense amounts of detail and precision in the music. Everyone who enjoys listening to chamber music can relate to this; when the Cello starts a solo the sound fills the room; one can feel the air moving in the room. New Butterfly excels on jazz and chamber music where its attention to detail and its SET roots bring out the best in the music. 
Published inAmplifier

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *