Another my unusual project. Hybrid headphone amplifier. Powered by a USB power bank. So portable as a booster for a portable player. No volume control because this is already the case in the sound source.
The tube are Soviet-made low-voltage triodes 6S63N with a metal-ceramic structure known as a nuvistor. Power amplification is provided by LME49600 buffers (audio version BUF634A, diamond topology). There is a power-on delay in switching on the outputs. The whole is closed by a DC/DC converter that provides voltages: + 12.6V for tube heating and output buffers, -11V for DC-servo and + 36V anode voltage for tubes. DC/DC converter works with a clock speed of 1.6MHz.
The first approach is to have no voltage gain for the entire amplifier. I achieved this through the voltage divider between the tube and the output current buffer. At the same time I have a reduction of the signal level and I have a voltage reduction of about 18V at the anode to the level of about 6.3V polarizing the input of the buffer. Tube-buffer DC coupling requires a DC-servo system. A series resistor at the output provides short-circuit resistance and offers an almost constant output power for various impedances ranging from 16 to 40 ohms.
Unfortunately, there were no readable characteristics of the 6S63N tubes available. Since I have a curve-tracer for electron tubes (uTracer 6), I took the characteristics off myself and adjusted the spice model of this lamp to them.
The problem of the constructed unit turned out to be the stability of the converter. The voltage adjustment loop compensation capacitor (C32) had to be removed. The good working point of the tubes offered the dominance of the second harmonic in nonlinear distortions, a much lower level of the third harmonic and a very low and negligible level of higher harmonics. The prototipe made it confirmed with the sound of, and it was confirmed by the measurement results. The sound is a bit brighter and the bass has more energy.
However, I immediately made a decision that this amplifier should offer a slight voltage gain. This will allow the sound source to be set quieter, which will favorably affect the distortions that arise in it. I did it through a simple modification consisting in adding 1.5uF capacitors in parallel to the R13 and R14 resistors. I obtained an additional 3.3x voltage gain (~ 10dB) for the 60 ohm impedance of my Koss Porta Pro headphones. The modified amplifier was measured.
The next version, keep this voltage gain, there will be a slightly different operating point of the tubes. There will be no DC tube-buffer coupling, which also means no DC-servo system and -11V power supply. All other functional features will remain unchanged.
Finally, I will add that there is a potential in this system to drive high-impedance (400 and 600 ohms) headphones with good efficiency. This will require a higher signal level and work with a THD level of around 3%, but here the second harmonic will dominate, which is not perceived by our sense of hearing as a distortion.
Simplified internal structure of LME49600 buffers:
Matching the model to the characteristics of the 6S63N tube obtained from uTracer 6:
Schematic diagram of the first version, assembled as a prototype:
PCB project:
Measurement results for the input signal level 1Vpeak (0.707Vrms)
Intermodulation distortions as defined by SMPTE, DIN and CCIF:
Crosstalk at 1kHz, 100Hz and 10kHz for each channel:
Channel L: -67.9dB, -67.5dB, -66.2dB
Channel R: -67.7dB, -67.3dB -65.8dB
Noise floor:
THD without weight filter:
THD with A filter:
Spectrum of THD with specification of harmonics for different levels of the input signal (0dB = 1Vp):
THD as a function of frequency, detailing the first harmonics. Two measurements for each channel:
Visualization of THD spectrum differences for both channels resulting from the dispersion of lamp parameters (input lever differs from 1Vp):
