the my Hi-End super linear Headphone amplifier without vaccum tube

by Andrea Ciuffoli - 1992 - 2004

contact me for any problem or question: webmaster@audiodesignguide.com

Introduction

I am publishing this project after many year from its idea and realization.

The project is born before to start to play with vacuum tube.

At this time I think that this project could be a very interesting solution if you don't want use vacuum tubes.

I think in any case that my SESS vacuum tube amplifier is the best solution to drive any headphone with an impedance from 250ohm to 600ohm.

The new simulation, the measurement and the listen test have demostrated that this transistor only project is a very hi-end amplifier.

It could drive any kind of headphone, from the 30ohm of a Grado to the 300ohm of the Beyerdynamic.

The idea

The schematic has been develop on the idea coming studing the Pioneer input stage of the Exclusive series amplifiers A-09 and M6.

PIONEER A-09	PIONEER M-6

The configuration of these amplifers use current mirror and a pnp-npn follower to compensate the distortion of the transistors.

If we analize a single section of the M-9 input stage excluding the dynamic cascode, the input fet, and we replace the current genarator with a single resistor:

we will get this following configuration.

from the original Pioneer specification

The circuit

The targets are:

Very low distortion with any load in the range from 22ohm to 600ohm
Class A operation in any stage
NO feedback (only dc conpensation)
All discrete components on the signal path (the operational is only for the dc compensation)

soon update !!!

The simulation

The performances of this design has been optimized with electronic simulation using the WinSpice .3.1 software for Windows.

The WinSpice 3.1 has been develop by Mike Smith and it is a beatiful port of original spice3f4 develop by Berkeley University of California.
It can generate waveform plots to individual floating windows and contains a powerful scripting language

download the circuit

The result is the following on 200 ohm load with 70mA bias

Fourier analysis for v(out):
No. Harmonics: 10, THD: 0.0402076 %, Gridsize: 200, Interpolation Degree: 1

Harmonic Frequency    Magnitude    Phase        Norm. Mag    Norm. Phase
-------- ---------    --------- p;   -----      & ---------    -----------
0       0.000000e+00 -2.15765e-01 0.000000e+00 0.000000e+00 0.000000e+00
1       1.000000e+04 1.065667e+01 -2.23091e-01 1.000000e+00 0.000000e+00
2       2.000000e+04 2.648180e-03 1.002893e+02 2.484997e-04 1.005124e+02
3       3.000000e+04 3.363799e-03 2.139252e+01 3.156518e-04 2.161562e+01
4       4.000000e+04 9.505072e-05 -5.31800e+01 8.919361e-06 -5.29569e+01
5       5.000000e+04 1.483406e-04 -1.20027e+02 1.391998e-05 -1.19804e+02
6       6.000000e+04 9.038702e-06 1.606560e+02 8.481729e-07 1.608791e+02
7       7.000000e+04 1.723391e-05 7.881297e+01 1.617194e-06 7.903606e+01
8       8.000000e+04 1.866338e-06 -3.13638e+01 1.751332e-07 -3.11407e+01
9       9.000000e+04 2.287575e-06 -1.01812e+02 2.146613e-07 -1.01589e+02

on 22ohm load with 70mA bias the output stage go in class AB operation

Fourier analysis for v(out):
No. Harmonics: 10, THD: 0.295698 %, Gridsize: 200, Interpolation Degree: 1

Harmonic Frequency    Magnitude    Phase        Norm. Mag    Norm. Phase
-------- ---------    --------- p;   -----      & ---------    -----------
0       0.000000e+00 -2.00760e-01 0.000000e+00 0.000000e+00 0.000000e+00
1       1.000000e+04 5.090969e+00 -2.22745e-01 1.000000e+00 0.000000e+00
2       2.000000e+04 4.032223e-03 -9.19993e+01 7.920344e-04 -9.17766e+01
3       3.000000e+04 1.428727e-02 -6.86804e-02 2.806394e-03 1.540646e-01
4       4.000000e+04 1.228811e-03 -9.04889e+01 2.413708e-04 -9.02662e+01
5       5.000000e+04 2.086553e-03 -1.14260e+00 4.098538e-04 -9.19858e-01
6       6.000000e+04 2.145174e-04 8.655970e+01 4.213686e-05 8.678244e+01
7       7.000000e+04 3.653638e-04 1.789355e+02 7.176705e-05 1.791583e+02
8       8.000000e+04 3.178136e-04 8.800366e+01 6.242693e-05 8.822640e+01
9       9.000000e+04 3.001016e-04 1.782272e+02 5.894785e-05 1.784499e+02

That is a very good result!!

The components

This project does not use esoteric components but common good transistors easy to find in any electronic shop.

All the NPN are BC547, all the PNP are BC557 and in the output stage the final are BD139 (npn) and BD140 (pnp).

datasheet of the uA741

These are some capacitors suggested for the ultra fast by-pass 0.01uF - 0.1uF and for the 1uF 50V-63V.

datasheet of the Spectrol 64W multi-turn trimmer

All these components could be find in any normal electronic shop or from http://www.rs-components.com

About the fast by-pass and as input I suggest to use the very good Auricap capacitors by Audience.
Each channel will use 4 x 1uF 200V as by-pass onf power supply and 1 x 4.7uF 200V in input.
The high frequency cut-off is composed by the 2200ohm input resistor and the 100pF Silvered Mica capacitor.

The layout

and will have this look if we insert the by-pass capacitors to increase hi-frequency performances.

If you are interested to another layer without pcb see these implementation by Consuelo Borghesi.

The Startup phase and setting

Before to power up the amplifier you need to be sure that the bias control is set for the minimun current otherwise you could destroy the final stage BD139/BD140.

I know that could be not easy understand which is the rigth way to turn the trimmer but if you use the suggested Spectrol multi-turn trimmer you just need to follows this image.

Now you can power up the amplifier with any 12V-0-12V to 20V-0-20V dc power supply.
Usa a multimeter to measure the voltage accross one of the two output resistor and start to search your target curent.
Ofcourse the current will be egual to the voltage read because the value of these resistors is 1ohm.
I suggest a bias current about 50mA and never more than 100mA because in this circuit there is not a termal compensation (it is easy to add if you need).
After set the bias current wait 15 minuts to be sure of the value.