I was brought an old amplifier that, at first glance, looked like a relic from the past. It was a push-pull with 2A3 tubes, originally driven by ECC88 tubes, but over time it had lost all its glory. The amplifier had been abandoned in a cellar, where its condition had deteriorated. The owner, after finding it, decided to have it repaired. However, his story about the crackling channel didn’t prepare me for what I discovered.
The First Encounter with the Wreck
When the amplifier was switched on, the result was not exactly encouraging. One channel oscillated on its own at 64 kHz, while the other delivered barely 0.4 watts. Once I removed the cover and looked at the circuit, it was clear that the amplifier needed a complete overhaul. The photos I show testify to the state of the chassis and the internals, which were heavily compromised. At that moment, I realized the only way forward would be to completely dismantle it and rebuild it from scratch.
From Dismantling to Design
I started by emptying the shell, carefully cleaning the chassis, which had a strong smell of mold due to its time in the cellar. Once the cleaning was completed, I designed a new circuit that was compatible with the existing chassis, keeping the original output transformers. The choice was to use four 2A3 tubes and two noval tubes for the various stages of the circuit.
One detail I had to face was the use of Lundahl LL1621/P-P interstage transformers. According to the datasheet, these transformers could not be driven by a single-ended stage, but only by a push-pull stage. Despite being an integral part of the original amplifier, after reflecting for two days, I decided to eliminate them.
In their place, I opted for a cathodyne phase inverter based on an ECC82, one triode for each pair of 2A3s, while for the input tube I used an ECC83. Although I had considered the idea of using a long-tail inverter, the need for an additional noval socket led me to choose the simpler, more practical solution.
Power Supply and Biasing: Crucial Details
A fundamental aspect of the project was the power supply for the 2A3 tubes. I decided to power them with alternating current, with four dedicated secondaries for each of the four 2A3s. This approach allowed me to separate the filaments, avoiding interference between the tube biases. I chose to implement a self-balancing bias following the Blumlein method, to perfectly balance the bias of each pair of 2A3s. This not only ensured an even distribution of current but also prevented the output transformers from being saturated by possible DC currents, achieving greater efficiency and uniform tube life.
To achieve this, I had to replace the power transformer with one specifically designed to meet the project’s requirements. Using advanced techniques, going beyond the use of simple resistors for filament centering (allow me to keep my secrets), I managed to eliminate any hum from the amplifier.
All components are firmly fixed on both sides to an anchoring point. No component relies on the support of adjacent ones. Every anchor point is screwed, including the nylon square pads, to which I fastened the wires using ties. Even these pads are screwed, with no adhesives used to secure components.
The filaments of the ECC83/82 tubes are also powered by AC, a solution I’m quite proud of.
The Final Result: Performance and Sound Quality
After assembling the entire circuit, the result was immediately satisfying and hum-free. The two pairs of 2A3s work in push-pull class A, delivering 6 watts RMS per channel, with a damping factor of 4.6 and a total harmonic distortion (THD) of about 0.18% at 1 watt. The resulting sound is extremely clean, powerful, and free from unwanted noise.
Bandwidth 20Hz -0.4dB / 75kHz -1dB
THD
Square Waves at 100Hz / 1kHz / 10kHz
The Tone Stove project was not only a technical challenge but also a true reinterpretation of the classic tube amplifier, optimized for maximum audio quality, without compromises or ideologies.