back to

To get more information contact me at:

Extreme Hi-End Hybrid Amplifier End
started on January 1 st , 2015


After checking the enormous potential of this amplifier , see the
Extreme Hi-End Hybrid, I have decided to raise the standard of quality at the highest level by using the best components on the market .

The previous version used an anodic choke with normal core that I had found in the my house with a gap not perfect for the tube used (50mA).

Here I have changed the anodic choke to 25mA gap to increase the low frequency band and an amorphous core to have a more transparent sound.

This project collects all the experiences made ​​in these years then it born with all these characteristics:

All the pcb are available at low cost on Ebay shop or you can download the Eagle files to produce these in your local area

See Ebay shop to buy pcbs I don't get money from  this Ebay shop.

The output stage pcbs has been design to optimize the signal path with large wire for the output stage.
I have increased the thickness of the high current tracks with tin (see photos).


The voltage amplifier
use a single ended vacuum tube stage with an Amorph c-core anode choke to have a very low distortion and best frequency band. I am using here the Lundahl LL1668AM/25mA anode chokes but you could use the normal and cheaper LL1668/25mA

If we use an anode choke or an inter stage transformer is necessary consider the effects of the inductance because you create with the cathode capacitor a RLC circuit with a specific resonance.
Follows the formula to calculate the resonance frequency:
so we need to keep this value very low to eliminate strange peak on frequency band.
The value of the internal resistance of the tube can modify the peak width generated in the resonance.

Many DIY ask me if they can change the vaccum tube with another but I think it is not easy to find  another type with these caracteristics:
  • voltage gain greater than 40
  • internal resistance near to 2Kohm
  • low distortion until 40Vrms
A good article about the use of alternative to D3a is on the good Bartola Valves article.

Follows a table to compare the distortion of some tubes in the same configuration with LL1668/25mA, Ck = 270uF or 560uF for 6C45 and 6S3PEV.
For this test has been used the power supply of this project without any change for all the tubes.

Tube model Pd max (W) Ri (ohm) Va (V) Ia (mA) Rk (ohm) Voltage gain (dB) THD 30Vrms load 30K THD 40Vrms load 30K THD 30Vrms load 15K THD 40Vrms load 15K
D3a triode 4.5 1900 150 24 60 37.0 0.4 0.6 0.6 0.7
EC86 2,2 5600 165 12 110 34.5 0.7 0.9 1.2 1.7
E180F triode 3.0 2700 165 15 110 33.2 0.9 1.2 1.5 1.9
6C45 7.8 1100 148 25 43 32.5 0.3 0.4 0.4 0.6
5842 4.5 1700 150 25 60 34.2 0.8 0.9 1.0 1.2
6S3PEV 3.0 5000 160 15 110 31.4 0.6 0.8 1.0 1.3

How many voltage gain you need on vacuum tube stage ? it is easy to calculate.
For 100W on 8ohm you need this output voltage:  P = V * V / R so V = sqrt(P * R) = sqrt(100 * 8 ) = 28.3Vrms.
The current amplifier loss about -0.65dB correspond to 0.93x so on output stage input you need 28.3 / 0.93 = 30.5Vrms.
I suggest to drive this amplifier directly with your source like DAC, CD and phono pre-ampl so the sensibility should be about 400mV like normal integrated amplifiers.
Now you know the minimun voltage gain necessary on tube stage  Vout / Vin = 30.5V / 400mV = 76x  =>  37dB.
If you think to use only CD or DAC probably you need less voltage gain and also 32dB are enought.
The anode choke give the best sound and the best performances but if you search a cheaper solution see my previous version where I have used normal resistance as load.


The current amplifier is a perfect Diamond Buffer used in some chip like the LH0002 and the BUF634.
This configuration has been analyzed very well in the article Bulding better buffer on Electronics World november 1992.
The same design as been used in many Audio Research amplifiers like the Audio Research D100 and the Audio Research D-400.
Also the Pioneer A-09 and the Pioneer M-6 use a Diamond Buffer on output stage but these implement dynamic current generators with some stability problems.

Advantages of this type of configuration compared to a conventional:
  • compensation of Vbe non-linearities between driver and final so very low distortion without feedback
  • high slew-rate and large frequency band
  • temperature run-away compensation without sensors if driver and final transistors share the same heatsink
  • high input impedance if driver stage use current generator (I1 & I2)
  • auto protected by short circuit on output
Disadvantages of this type of configuration compared to a conventional:
  • the bias current in the driver limits the current supplied to the load
It is evident that there are many benefits, but to overcome the problem of the current supplied to the load you need to study well the working points of the driver and final.

There is no trimmer for adjusting the bias so you must use exactly the transistors described in the project.
To use other types you need to make many test and in some case set R9-R12 to 1ohm or more.

This current amplifier have an input impedance near to 20Kohm and it absorbs a max peak current of 2mA also when it drive 2 ohm load at the max output level (simulated with 40Vrms on 2ohm load so 20A).

I have add a power supply regulator to eliminate any ripple coming from main capacitors after the diode bridge and it allow a soft start with these advantages:
This regulator integrate a DC loudspeaker protection copied from an old Bartolomeo Aloia design to shunt the regulator capacitors so switch-off the output stage if there are fault on output transistor.

The op-amp in the circuit is a DC servo to eliminate DC offset and it work in very low frequency band.

R1,R25                            56Kohm  1/4W  1%    or better use only R1  27K MK132 Caddock
R2,R14                            10ohm   1/2W  1%
R3,R36                            280ohm  1/4W  1%   
R26,R27                           2700ohm   2W
R34,R35                           15ohm     2W
R23,R33                           270ohm    1W
R24,R28                           5600ohm   2W
R5,R6,R7,R8,R17,R18,R19,R20       0.47ohm   5W
R15,R16                           220K    1/4W  1%
R55,R56                           2200ohm 1/4W  1%     or better use only R55 1K MK132 Caddock
R51,R52,R53,R54                   470ohm  1/4W  1%     or better use only R51 and R53 220 MK132 Caddock
R4,R13                            6800ohm   2W
R49,R50,R21,R22                   10Kohm  1/4W  1%

R11,R10,R12                       0ohm  (jump)

C6,C8,C9,C14                      0,47uF  63V  Wima
C1,C2,C10,C11                     100uF  25V
C3                                empty
C5                                0.1uF   100V   Vishay MKP1837
C7,C17,C4,C24,C18,C20,C25,C19     4700uF 63V Nichicon  KG  type 25x50mm
C15,C16                           1000uF 63V Panasonic or Nichicon 100

D1,D2                             Zener 15V  1W
D3,D4                             UF4007 or 1N4007
D5,D14                            1N4007
D9,D13                            0ohm  (jump)
D8,D12,D11,D10,D6,D7              1N4148

F1,F2                             FUSE 5A FAST (with fuse holder)

U1                                2N5550 o 2N5551
U2                                2N5401

Q1,Q3,Q16,Q18                     MJE15033 (PNP)
Q2,Q4,Q14,Q17                     MJE15032 (NPN)
Q9,Q5,Q6,Q7,Q13                   NJW0281G (NPN)
Q12,Q11,Q8,Q10,Q15                NJW0302G (PNP)

IC1                               OPA134 (you can use many others single op-amp)


As mention in the characteristic list at the begin of this page in this project the passive components are without compromise on any section.

On vacuum tube cathode there is an Sanyo OS-CON capacitor, my choice is .... available in the Parts Connexion web store.

The SANYO OS-CON is an aluminum solid capacitor with high conductive polymer or organic semiconductor electrolyte material.

Due to the solid electrolyte used, OS-CON achieves low Equivalent Series Resistance (ESR), excellent noise reduction capability and frequency characteristics.

Normally the capacitors with lower voltage sound better so I prefer to use a 4V   than a 16V if there is only 2V on cathode.

The interstage capacitors must be of the highest possible quality because most of the sound quality depends on these.

My choice is the Audyn Cap Reference classified with 11+ by Humble Homemade test but in my opinion these are much more transparent than others 12 and 13 result.

As tested in the
first version the anode choke give the most low distortion and the best frequency band, also the interstage transformer it more invasive so it should be used only if it is essential in the circuit as a phase shifter, see the Amplifier End.

The LL1668AM/25mA with an inductance of 100H, an air gap to support a current of 25mA and a static resistance of 680ohm give a sure result.

All the resistors in the current amplifier has been duplicated to use for any resistor a parallel of two components mounted with opposite direction to reduce parasitic inductance and electromagnetic flux.

I learned this method many years ago by the owner and designer of a famous American factory of amplifiers and it use the same also with Solen interstage capacitors.

The alternative to this method is to use Caddock MK132 with advantages.

About the power supply capacitors near the output transistors there are no alternatives to the
NIchicon KG capacitors because the my last experience made ​​the
Amplifier End has demonstrated that these components without bypass have unsurpassed musical performance.

I have used 8 x 4700uF 63V NIchicon KG capacitors for each channel so we have a great reserve of power available in a very short time.

The 0.47ohm output resistances present on the emitter of final stage transistors are necessary to compensate some differences on output transistors, stabilize operating point and to merge all the outputs.

In order to have no bad effects on the sound the my choice has been the MILLS MRA-5  non-Inductive wirewound resistors with 5 watt and with 1% tolerance.

Using the transistors specified I have on any 0.47ohm a voltage about 0.180V so the bias current is 0.38A so 4 x 0.38A = 1.5A.



Here all the files to simulate this circuit with LTSpice software.

Add these lines in the file C:\Program Files\LTC\lib\cmp\standard.bjt to have the transistors models:

.MODEL Qnjw0302g pnp IS=5.16751e-16 BF=114.657 NF=0.895716 VAF=50.2189 IKF=6.409 ISE=3.9641e-15 NE=4 BR=1.47167 NR=0.923324 VAR=255.567 IKR=6.34299 ISC=3.96408e-15 NC=2.82194 RB=2.66347 IRB=0.1 RBM=2.0828 RE=0.0001 RC=0.0652395 XTB=1.45322 XTI=1.08126 EG=1.05 CJE=2.14504e-09 VJE=0.4 MJE=0.376227 TF=2.16864e-09 XTF=1000 VTF=843.737 ITF=501.348 CJC=5e-10 VJC=0.95 MJC=0.251547 XCJC=1 FC=0.8 CJS=0 VJS=0.75 MJS=0.5 TR=1e-07 PTF=0 KF=0 AF=1

.MODEL Qnjw0281g npn IS=4.36849e-12 BF=98.1488 NF=1.01332 VAF=37.9046 IKF=9.71849 ISE=1e-16 NE=1.8326 BR=0.79921 NR=1.09994 VAR=339.743 IKR=5.77305 ISC=1e-16 NC=2.71592 RB=2.74892 IRB=0.33289 RBM=2.74892 RE=0.000344671 RC=0.03203 XTB=1.7742 XTI=1.12262 EG=1.20598 CJE=3.66793e-09 VJE=0.74806 MJE=0.85 TF=2.27115e-09 XTF=1000 VTF=912.955 ITF=296.602 CJC=5e-10 VJC=0.95 MJC=0.270858 XCJC=0.98254 FC=0.8 CJS=0 VJS=0.75 MJS=0.5 TR=1e-07 PTF=0 KF=0 AF=1

.MODEL Qmje15032 npn IS=3.7344e-10 BF=86.8313 NF=1.23974 VAF=31.5491 IKF=9.1678 ISE=9.2499e-12 NE=3.28127 BR=5.59346 NR=1.33161 VAR=2.1791 IKR=5.15023 ISC=4e-13 NC=4 RB=9.54492 IRB=0.1 RBM=0.1 RE=0.000568481 RC=0.0931741 XTB=0.737036 XTI=1.04983 EG=1.206 CJE=3.05969e-09 VJE=0.648491 MJE=0.352663 TF=4.94819e-09 XTF=1.50001 VTF=1.0001 ITF=0.999982 CJC=3.00108e-10 VJC=0.600021 MJC=0.40991 XCJC=0.8 FC=0.534651 CJS=0 VJS=0.75 MJS=0.5 TR=1e-07 PTF=0 KF=0 AF=1

.MODEL Qmje15033 pnp IS=7.51228e-10 BF=134.35 NF=1.25737 VAF=12.5778 IKF=1.88497 ISE=7.74267e-12 NE=3.34528 BR=5.14173 NR=1.47488 VAR=1.4505 IKR=7.47186 ISC=3.25e-13 NC=4 RB=4.37743 IRB=0.1 RBM=0.1 RE=0.000332989 RC=0.381218 XTB=0.223027 XTI=1 EG=1.05 CJE=3.06005e-09 VJE=0.64838 MJE=0.352991 TF=4.78203e-09 XTF=1.50001 VTF=1.00006 ITF=0.999988 CJC=3.00101e-10 VJC=0.600019 MJC=0.409916 XCJC=0.8 FC=0.534975 CJS=0 VJS=0.75 MJS=0.5 TR=1e-07 PTF=0 KF=0 AF=1



The power supply for the current amplifier use two 500VA toroidal transformers.

The good power supply transformers for audio should have an electromagnetic flux 20% lower than normal (14000 gauss instead of 17000).

I have used a rectifier module based on four ON Semi MBR40250 Schottky diode to create a full wave bridge rectifier.

Here a good description of the benefits derived from the use of Schottky diodes in audio amplifier.

The same diodes as been used in my Amplifier End but you can start to assemble this project also with a common diode bridge like the IRF 36MB60.  

After the diode bridge there are a pair of 10000uF 63V (size: 50x80mm) capacitors to create a first soft rectification.

Here you can use any normal capacitor like Kendeil, BHC, Sprague because these are not on the signal path.

As explained in the simulation before using larger capacitors does not bring us benefits.

The my choice is 36V-0-36V transformers to get about 100W on 8ohm.

You could increase the voltage of the transformers to have more output power but keep attention to the capacitor voltage and to the hot to dissipate on heatsinks. 

About the capacitors near the output transistors (on pcb) use only Nichicon KG as specified in the list  8 x 4700uF 63V KG (size 25x50mm) per each channel.



To get the max sonic performances the my choice is a slow turn on power supply for the filaments using a common LM317 or LM350 and a virtual battery operation power supply for the anodic with a power mosfet.

For this power supply I have used the R-CORE model R26-90 got on Ebay shop with primaries 2 x 115v and the secondaries 0-165v(0.2A) and 2 x 0-9v (1.1A).

R6,R8                 1ohm     2W
R74                   47Kohm   1/4W 
R7                    1Kohm    1/4W
R9                    220ohm   1/4W
R12                   100ohm multi-turn
R4,R5                 10Kohm   1/4W  1%
R3                    470Kohm  1/4W  1%
R1                    39K      2W
R2                    100K     3W

CY1,CY2,CY3,CY4       10nF   400V
C2                    150uF  400V
C1                    33uF   400V
US10,US11             56uF   400V Solen MKP
C3,C7                 2200uF 25V
C4,C5,C6              220uF  16V


D1,D2,D3              1N4007
D5,D9                 zener 10V 1W
B1                    diode bridge  4A 100V
LM350 or LM317
Q1,Q3                 IRF840
Q2                    2N2904



Follows the distortion analysis of the complete amplifier on 8ohm load,  obviously until 30V this is the distortion of vacuum tube voltage stage because the current amplifier have a distortion value 10 or 20 times less and only after 27 volts rms it goes into clipping zone for the supply voltage limitated by project.   On 4 ohm load the plot is very similar.


If you will use a 2 x 36VAC transformers the power to dissipate is near to 150W for each channel.

Any heatsink used have a Rt C/W = 0,48 so 0.24 for each channel so the temperature will be  0.24 * 150W + Tenv = 36C + 30C = 66C.

The cabinet is a Hi-Fi 2000 model  04-400B Silver 10 (1NPD04400B) buy on the
HiFi2000 webshop.

     Formato: 4 Units
     1x Frontal Panel in aluminium 10mm oxidised silver 450 mm
     1x Back in metal painted black anti-scratch
     4 Lateral Heatsinks 200mm oxidised black  (each with Rt C/W = 0,48)
     2x Covers in metal painted black anti-scratch
width: 450mm internal 360mm
depth: 400mm Internal
height: 165 mm

weight:: 11,5 KG

To add a 1BASEPD400 Inner pierced Base for Pesante Dissipante 400mm.

To fix the transistors on heatsink I will ask to HiFi2000 for a curstom work on heatsinks.



1) set the variac to about + 2v and -2V dc to check that the heatsink is not connected to these tensions
2) control the voltage after the regulators these should rise slowly in a few seconds
3) set the variac to + 15V and -15V to verify that the output voltage goes to zero volts dc
4) set the variac and -20V to + 20c to verify that the voltagee across the resistance from 0.47ohm is almost equal among all the transistors (0.05V - 0,06V)
5) set the variac at the final voltage



SOFT START - PUSH BUTTON - PROTECTION  (see the relative page)


Vandal Resistant  Push Button Switches
RS Stock No. 174-6381
Manufacturer Apem Manufacturers
Part No. AV021003C900

The panel cut out  is 22mm, the same of noval socket, so this is the ideal choice for the 10mm front panel because you can use the same cutter.



It is possible to find some components at lower cost, in red the most important for the final result

description unit price quantity total (euro)
D3a 25 2 50
Solen 56uF 250V 20 2 40
HI-Fi2000 chassie with machining (about) 300 1 300
500VA transformers 100 2 200
30VA transfomers  40 1 40
10000uF 63V 15 4 60
4700uF 63V KG 6 16 96
transistor 36 2 72
LL1668/AM 230 2 460
Audyn Cap Reference 2.2uF 40 2 80
other 100 1 100
DACT 150 1 150