back to

To get more information contact me at:

the my Inpol / Mofo SE class A amplifier - version 2   
started in the Apr 2020



This is a 40-45W Class A Single ended no feedback amplifier.

Since in 1999 I saw the first review of the T.T. of Pathos this intrigued me, I had already made my Power Follower 99 and the sound performances were surprising but with the Inpol it was possible to double the efficiency.  
After 20 year I have created a new verson of the my amplifier the Power Follower 2019 because this has incredible sonic performance but has a big efficiency problem even if this is normal in a true class A single ended amplifier.
However, the idea of ​​using an inductance to double the efficiency was exciting, even if it meant losing the isolation from the power that my Power Follower has.
From the simulations I had verified that an inductance of about 100mH must be used with an air gap capable of manage at least 4A.
In the following years I started a lot of projects with the output transformers for valves and anode load inductors so I got the idea that the Inpol inductance had to be sectioned / stratified to reduce the parasitic capacities so the project has been forgotten.
I recently read Mike Rothacher's article about the Mofo project which got an excellent frequency response using a normal Hammond 193T, cost only 39$.
The Mofo project is very popular on the DIY audio forum (more than 500000 views) [Build This MoFo!].
The Mofo and Inpol are basically the same thing, a mosfet in the common source configuration with an inductance connected between the source and the ground. I don't think you can patent such a circuit because it is equivalent to any tube circuit.
For me it was necessary to check if these performances were confirmed even by using larger inductances with at least 100mH and capable of handling 5A like the Hammond 195T5.




This are the specifications of the choke used for this project.

Hammond 195T5

Inductance: 100 mH

D.C. Current: 5A

Resistance:  0.64ohm

Weight:  about 5Kg
Dimensions:  95(h) x 117(l) x 120(d) mm



In the Power Follower we have a typical source follower (as an emitter follower but with a Mosfet) working in pure class A with a current generator to set the bias current and to fix the output point to half of power supply voltage (in the following schematic is 20V).
In the Inpol / Mofo there is the same source follower but in this case there is big choke connected to ground and output point is at about 2V.
To have the same output power we need in the first a power supply voltage of 40V and in the second the half only 20V.
The theoretical efficiency in one case is 25% and in the second 50% like a pure class A push-pull.
In reality with a dissipation of 100W the Power Follower have an output power of about 20W and the Inpol / Mofo give 40W.
You must consider that a vacuum tube single ended amplifier with an output power of 20W, for example the my GM70 SE, have the an energy consumption about 120W per channel.


The D1 and D2 are 18V 1W diode zener used to prevent save the mosfet when it receive bad input signal like a vacuum stage startup, in some mosfet like 2SK1058 these zener are integrated.

The phase on output terminals should be inverted because I am using a voltage amplifier that reverses the phase.

In the Power Follower the quiescent current is set  by the resistence on current generator source pin, it can be changed with the simple formula Iq = 0.65 / R where 0.65 is the Q1 transistor Vbe and there is a trimmer to set the source of upper mosfet to half of power supply voltage.

In the Inpol / Mofo there is a single trimmer to set the quiescent current.

t this point we must consider 2 constraints: the maximum output voltage determined by the power supply voltage and the maximum output current determined by the bias current.

The output power will be limited by both these values so after choosing a power transformer and a bias current we can check the output power for each type of load.

There are some differences on mosfet specifications from one manufacturer to another, a low input capacity is crucial for having a good high frequency response.

For example the IRFP150 have these following differences:

The my voltage stage with the 6072A have an output impedante of 670ohm so the high frequency cut-off can be calculated with:

Ft(-3dB) = 1 / (2 * pi * C * R) = 1 / (2 * 3.14 * 1900pF * 670) = 1 / ( 2 * 3.14 * 1900E-12 * 670) = 125KHz

The my second voltage stage with the 12AX7 have an output impedante of 427ohm so the high frequency cut-off can be calculated with:

Ft(-3dB) = 1 / (2 * pi * C * R) = 1 / (2 * 3.14 * 1900pF * 427) = 1 / ( 2 * 3.14 * 1900E-12 * 427) = 178KHz

In the Power Follower 2019 I have used the IRFP150NPBF by Infineon (RS cod. 541-0856) with only 1900pF.

Many persons will think to drive this current amplifier with E88CC in SRPP (Totem pole)  or D3a/E182CC/5842/6C45 in single ended.

Always consider the high frequency cut-off because a D3a or a 6C45 have an output impedance near to 2Kohm so we will have:

Ft(-3dB) = 1 / (2 * pi * C * R) = 1 / (2 * 3.14 * 1900pF * 2Kohm) = 1 / ( 2 * 3.14 * 1900E-12 * 2000) = 41KHz


Follows the simulation of the current amplifier output impedance that is about 80mohm with IRFP150NPBF and it increase in the low frequency because there is the output capacitor.

The value of the choke is important to keep a good low frequency band, using 10mH instead of 100mH mean -0.5dB at 20Hz.

but the main problem with little choke is the distortion at low frequency, folllows the simulation with 50mH and 100mH.


Also the value of the output capacitor must be enough to have no loss at low frequency so the 10000uF until 4ohm load,
folllows the simulation with 3300uF and 10


Follows a table to calculate the transformer voltage (Vac), bias current (Ibias) and power to dissipate (Pd).

Load Vrms Pout Vp Irms Ap delta Vdc Ibias Vac Pd
8 11 15.1 15.5 1.38 1.94 1.15 17.8 2.2 15.9 39.8
6 11 20.2 15.5 1.83 2.59 1.15 17.8 3.0 15.9 53.0
4 11 30.3 15.5 2.75 3.88 1.15 17.8 4.5 15.9 79.5
8 12 18.0 16.9 1.50 2.12 1.15 19.5 2.4 15.9 47.3
6 12 24.0 16.9 2.00 2.82 1.15 19.5 3.2 15.9 63.1
4 12 36.0 16.9 3.00 4.23 1.15 19.5 4.9 15.9 94.7
8 16 32.0 22.6 2.00 2.82 1.15 25.9 3.2 22.6 84.1
6 16 42.7 22.6 2.67 3.76 1.15 25.9 4.3 22.6 112.2
4 16 64.0 22.6 4.00 5.64 1.15 25.9 6.5 22.6 168.3
8 17 36.1 24.0 2.13 3.00 1.15 27.6 3.4 23.9 95.0
6 17 48.2 24.0 2.83 4.00 1.15 27.6 4.6 23.9 126.6
4 17 72.3 24.0 4.25 5.99 1.15 27.6 6.9 23.9 190.0

Obviously for a 5A bias current the transformer must be min 10A.




The presented topology, has NO voltage gain (actually it looses something 1.3%) so it should be driven by voltage gain stage, with an output swing not lower than 10Vrms and Rout < 1000ohm. 

The input impedance of the current amplifier is 110KOhm - 1800-2800pF and its resistive value can be adjusted by a pretty wide range, just using a different input resistance (max 220Kohm). 

The current amplifier do not introduce any alteration on the signal so is very important take care of driver stage/voltage amplifier.

I thought of this voltage amplifier looking for something that would sound great without using anodic inductors and interstage transformers.

Compared to the 100W hybrid amplifier, in this case we don't need an extreme dynamic to drive a current amplifier, 17Vrms are enough.

Both the following designs give good performances (voltage gain and output impedance), the tubes used have a good reputation for sound quality and the vacuum tubes are in the current production.

Obviously we must not forget the connection of the reference of the ground filaments because each valve has its maximum Vkf (cathode - filament voltage) and in this circuit the second stage has the cathode at about half of the supply voltage therefore a reference will be chosen between ground and this cathode (so about 80V for the 6072A and 90V for the 12AX7).

6072 version

This 6072A driver powered at 280V and with an input signal of 0.58Vrms give about 18Vrms with 1% thd.


12AX7 version

This 12AX7 driver powered at 280V and with an input signal of 0.33Vrms give about 19Vrms with 0.56% thd.

Here the distortion is lower but has been apply a local feedback on first stage becasue there is no cathode capacitor.




I advise to build this project using a pcb instead of an air wiring because it is certainly more stable.

To support the high dissipation has been used 2 mosfet in parallel.


Mount the resistances in parallel configuration in opposite phase / direction.

R1,R2,R4,R5  1800ohm 1/4W 1%
R3           10Kohm  2W
R6,R7        220K    1/4W 1%
R9,R11       0.22ohm 5W       Mills RMA5 

C1           4700uF 16V   Nichicon  model UKA1C472MHD
U$1          0.68uF 400V  Clarity Cap CMR or other audio grade MKP

R8           10Kohm multi turn

IRFP240PBF (it would be desirable to make a selection based on current)

D1,D2        zener 15V 1W                 
D3           zener 10V 1W

F1           fuse 10A FAST with fuse holder Mouser 534-4628

Cout         min. 10000uF 50V Nichicon KG   Mouser

Isolators    Bergquist SP400-0.007-00-104  RS RS541-0856

The connections are 63862-1 (CUT STRIP) by TE Connectivity / AMP (cod. Mouser  571-63862-1-CT, cod. RS 718-7987)


This is the complete schematic of the voltage amplifier including the power supply section allocated in the same pcb.


Ebay shop for the PCB   same of the PF2019

This is a two channels pcb so the components list is complete.

Mount the resistances in parallel configuration in opposite phase / direction.

R1,R2,R18,R19     100Kohm 1/2W 1%                    Mouser 594-MBB02070C1003FCT
R5,R6,R22,R23     220Kohm 1/2W 1%                    Mouser 594-MBB02070C2203FCT
R3,R4,R20,R21,    1000ohm 1/4W 1%                    Mouser 594-MBB02070C1001FC1

R7,R8,R24,R25     5600ohm 1/4W 1%    <<< 5600ohm for 6072 and 4000ohm for the 12AX7
R15,R16,R31,E32    220ohm 1/4W 1%
R17,R33            3.3ohm 2W                         Mouser 660-MOSX2CT52R3R3J
R12,R30           470Kohm 1/4W 1%                   
R14,R29           150Kohm 3W
R13,R28            27Kohm 1W

C1.C3             220uF 6.3V   OS-CON                Mouser 667-6SEPC220M+TSS <<< not use for 12AX7
C2,C4              33uF 400V                         Mouser 647-UVY2G330MHD
C5,C6             100uF 400V                         Mouser 647-LGU2G101MELZ
CY1-CY8            10nF 440VAC                       RS 335-066 o 80-R474I210050A1K

U$3,U$6           IRF840                             Mouser 844-IRF840APBF
U$11,U$7          33uF 400V Solen MKP   

D1,D2,D3,D4,D6,D7,D8,D9  UF5406                      Mouser 625-UF5408-E3
D5,D10                   zener 10V 1W

KK1,KK4           Extruded Style Heatsink for TO-220 Mouser: 532-513102B25

Bergquist SP400-0.007-00-54        RS 169-2177

TO-220 nylon platstic Insulator hole size M3

The connections are 63862-1 (CUT STRIP) by TE Connectivity / AMP (cod. Mouser  571-63862-1-CT, cod. RS 718-7987)

The heat sinks have been grounded to avoid receiving radio frequency so these must be isolated from the transistor.

In order to keep the Vkf under the max value it is necessary add 4 resistors 2 x 100K 2W and 2 x 39K 1W (6072A) or 2 x 47K 1W (12AX7), see photo1 and photo2.

12AX7 - 6072A socket.



Obviously it is fundamental to use a high quality interstage capacitor and for this project I will test 4 different types of a good UK company.
This test was necessary to decide which capacitor to use because it is not true that it is enough to buy the most expensive one to be sure of having the maximum sonic performance.
It happened to me to discard many high-level interstage capacitors, also very well evaluated.
In all my sonic performance tests I always search to be sure of an objective result using more persons and these with different experiences.
The ClarityCap has been manufacturing high quality audio grade capacitors for over 30 years.

The low  frequency cut-off is determined by the value of this capacitor and input resistance of the current amplifier (110Kohm).

Ft(-3dB) = 1 / (2 * pi * C * R) = 1 / (2 * 3.14 * 0.68uF * 110Kohm) = 1 / ( 2 * 3.14 * 0.68E-6 * 110000) = 2.13Hz

I suggest to use values in the range 0.47uF to 2uF.



Here 3 different 6072A tubes of current production and of course each has a different sound characteristics.

  1. JJ
  2. EH
  3. TAD


This is the power supply design for the current amplifier, it use two MeanWell LRS-150-24.


I have add a capacitor on DC output the type Nichicon 4700uF 50V KG.

As output capacitor has been used a Nichicon 10000uF 35V LKG1V103MESBBK 30x50mm.

The power supply transformer for the driver stage is an 40W R-core model R26-09.

You can buy on Alixpress online shop.

The secondaries have 2 x 220V  50mA and 2 x 6.3V 0.8A.




You must consider the power supply capacitor on the signal path like the output capacitor so both must be Audio grade.

The Nichicon KG capacitors has been used on all my last hybrid amplifiers and in the my Amplifier End I decided to eliminate the bypass capacitors 47uF Solen MKP originally used because the sound is much better without these.

I have chosen for the output capacitor the value 10000uF because this give a low frequency cut-off very low also on 4ohm load.

Ft(-3dB) = 1 / (2 * pi * C * R) = 1 / (2 * 3.14 * 10000uF * 4ohm) = 1 / ( 2 * 3.14 * 10000E-6 * 4) = 4Hz

Using this value the output impedance give an acceptable 0.15ohm at 100Hz and 0.8ohm at 20Hz, this is much better than any SE tube amplifier but if you want to get a best damping factor use 33000uF to have 0.25ohm at 20Hz.




All the resistors on the signal path have 2 positions on pcb because ...

otherwise to get a slightly better result using the MK132 Caddock.



I am using this module got on Alixpress because it incluse a soft-start and a temperature protection at 75.



In order to dissipate all the heat generated by this amplifier in my case I chose this container by HiFi 2000.

Dissipante 05/300B 5U 10mm SILVER  
Product Code: 1NPD05300B

temperature coefficient 0,18 C/W per each side

Inner baseplate for Dissipante 300mm
Product Code: 1BASEPD300

 I used the HiFi 2000 company for almost all the mechanical processes and here there are some specifications used for this phase.

In a second phase I have add another hole on the heatsinks for the temperature sensor (see below in the photos section)

Here the cost of the chassie

1 x Dissipante 05/300B 5U 10mm SILVER (1NPD05300B) €187.00
1 x B) Drilling front panel 10mm (LAV10MM) €25.00
2 x A) Drilling panel 2/3/4mm (LAV4MM) €50.00
1 x Inner baseplate for Dissipante 300mm   €14.15

Totals Sub-Total: € 276.15
BRT for Italia (Weight: 12.00kg): €15.00
IVA 22%: €56.83
Total: €355.20


Vandal Resistant  Push Button Switches

Manufacturer Apem Manufacturers 
Part No. AV021003C900
RS  No. 174-6381

The panel cut out for thid button is 22mm, this is the same size of noval socket so this is the ideal choice for the 10mm front panel.



I have tested 3 different types of insulators for the assembly of mosfets.

Environment 2 x IRFP240 total 30.9VDC 3.66A 113W to dissipate on Hi-Fi2000 heatsink 3 units H120 (smaller than what will used).

The senson was in the center of heatsink.

Sil-Pad Bergquist SP400-0.007-00-104 (RS 707-3367)

time (min)  degrees (C)
   0          27
   5          42
  10          52
  15          58
  20          62

 Thermally Conductive Insulator Aavid 4180G + Thermal Interface Products Accessory / Grease Aavid 101800F00000G

time (min)   degrees (C)
      0          27
   5          42
  10          53
  15          57
  20          60

Kapton film + Thermal Interface Products Accessory / Grease Aavid 101800F00000G

time (min)  degrees (C)
   0         27
   5         42
  10         52
  15         57
  20         60

So the Sil-Pad it has been confirmed as valid and easier to use.



Obviously before the first start up the bias trimmers must be set for the minimum current so turn the trimmer like show.


I think the best mosfet for this project are:

I have found this values of voltage on Rs (source resistor) on my IRFP240 so I have selected the first for one channel:




2 x IRFP240 with Rg=900ohm and Rs=0.22ohm

total bias 4A 30VDC.

about 40W on 8ohm thd about 2% with perfect decay
about 40W on 4ohm thd about 2% with perfect decay

Rout about 0.2ohm

Frequency response 3.5Hz - 220KHz at -3dB on 8ohm with Cout=10000uF
Frequency response    7Hz - 220KHz at -3dB on 4ohm with Cout=10000uF




24 Oct 2020
Last evening the first listening session of the new version of my Inpol / Mofo with the 195T5 chokes and the MeanWell LRS-150-24 switching power supplies.
The result was surprising because with these switching power supplies improve a lot the result.
Probably the 500VA was not enough big for the both channels with bias 30VDC 4A but I have no enough space to use two 500VA transformers.
Now the Inpol / Mofo it has almost reached the levels of my reference the PF2019, however if the Inpol / Mofo has a lower depth of scene and some less micro details. 
Obviously now we have to ask ourselves whether if is better to build the PF2019 or the Inpol / Mofo.
The Inpol / Mofo with its doubled efficiency represents an obligatory choice if you need 40-50W but I think that if 20-30W are enough for your use the PF2019 is the right choice even if it consumes twice as much.
In the driver stage has been tested 6072A of two different companies JJ and EH.
The source used was HQPlayer + Network Audio DSC pure DSD DAC with permalloy output transformers + DACT 10Kohm.




It is possible to find some components at lower cost but the interstage, output and power supply capacitors are important for the final result.

My configuration
description unit price quantity total (euro)
Vacuum tubes 6072A JJ 20 2 40
Solen MKP 33uF 400V 10 2 20
Components + pcb 100 1 100
Mosfet 4 4 16
Interstage capacitors 40 2 80
HI-Fi 2000 chassie 350 1 350
Vandal Resistant  Push Button 20 1 20
Soft-start + termal protection 24 1 24
LRS-150-24 22 2 44
Hammond 195T5 100 2 200
4700uF 50V 7 2 14
10000uF 35V 8 2 16
Connectors 26 1 26