The top reference tube amplifier


Pure Class-A 2A3 PSE - Parallel Single Ended

(An Andrea Ciuffoli's design, a 711's construction)


This is the version of the amplifier constructed by Andrea Ciuffoli with 300B fitted in instead of 2A3.



I classify this project as one of the best audio amplifiers, in terms of sound quality, that have ever been made. This is a TOP REFERENCE amplifier that has been designed with no compromise in mind. It is a VERY expensive project intended for the most advanced builder. You definitely cannot reproduce the quality of this amplifier with low cost components.

The main results match these items:

  • top sonic performances
  • two stages only from DAC chip to the loudspeakers
  • power about 7-10watt rms with 8ohm nominal load
  • very low distortion about 1% at the max power
  • good damping factor with an output resistance near 1.6ohm
  • very good low frequency response about 5Hz (-3db)

Although this is an Andrea Ciuffoli's design, in this page I present my approach, with some minor modifications that can improve the amplifier to my oppinion.




All the audio transformers used in this project are Lundahl products:

  • LL1627/120mA + PCB_B       Output
  • LL1660/10mA                        Interstage
  • LL1650                                  Power supply transformer
  • 2 x LL1638/10H 200mA         Power supply inductances
  • 1 x LL1638/4H 500mA           Power supply inductance
  • 2.5V 3A / 2.5V 3A                  Custom toroidal transformer

Lundahl is one of the world´s leading manufacturer of audio transformers used in many professional studio recording and radio stations so they have a very big experience and all these products use a custom audio C-core.

The total price of the transformers cost me about 1000 Euros.

As far as concern  the input stage I have tested many combination starting from parallel 6SN7 to the 6H30 Russian incredible tube but with the 5842 Raytheon the sonic result is much more higher. But note, the 5842 has a much shorter life (about 1000 hours compared to other 10000 hours of other tubes)

The other advantage using the 5842 is the high voltage gain that allows to connect the amplifier directly to the DAC chip so we will have only two tubes on the signal path. In fact I am proposing in this same page also a digital input.

I am using a DACT stepper attenuator instead of the normal ALPS or NOBLE solution, because the the sound is clearer. A good review of these components can be found on the article published in Hi-Fi World.

In this schematic also there are no passive components on the signal!

The output tubes used are the 2A3.

The tube diode rectifier GZ34 should be the Mullard version. Some DIY don't like the sound of the GZ34 because consider this to much hard near solid state diodes but this is not true because depend by the power supply design.  Also, 5R4 can be used. The appearance of the military 5R4 is nice and military types of this tube are long lasting.

About passive components types,  I don't leave many choices:  Holco or Caddock resistances on cathode for their excellent non-inductive and precision characteristics, ELNA Cerafine capacitors on cathode and power supply.
The ELNA Cerafine capacitors contain super fine ceramic particles which, through chemical reaction, improve the discharging speed between the anode and electrolyte with very low distortion.
Any other electrolytic or polypropylene capacitor give very lower sonic performances introducing some alteration. Only the Blackgate WKZ could give better sound, but I have not tested it and I know of some problems during startup phase.

Ciuffoli's quote about wire was: About wire forget the teflon! use only stranded tinned copper wire with pvc isulator and for loudspeakers I am testing the elevator cables (paralleled wire with a good section). A very good quality stranded tinned copper wire could be buy directly from E-Z-HOOK that carries an extensive line of fine stranded and extra flexible wire.

Well I have an objection on this. At least for the signal path, use silver plated or 99.9% pure silver wire, with teflon tube insulation. These wires are just excellent, and this can be practically proven from their use to critical performance radio coils. I do not believe that silver plated is worst than 99.9% pure silver wire. In fact you do not gain too much from using 99.9% pure silver wire, as the most signal is carried out the thin external layer of the cable. But the advantage of using 99.9% pure silver wire is that you further ensure that it's external layer will be the same as the inside, 99.9% pure silver material. I am saying this because some sellers sell bad quality silver plated wires that get some corrosion after some years. The insulation is mandatory as it prevents from accidentally touching cables. But it also has another advantage. It further prevents dust and humidity from reaching the surface of the cable, which will result in corrosion after some years. Also I do not agree with Ciuffoli about teflon. Teflon has the lowest relative dielectric constant (about 2) after the air (about 1) and it also has one of the lowest tangent losses (about 0.0002). It has also one of the best maximum temperature characteristics (about 250 degrees Celsius). Additionally, it is a highly non-chemical reactive material, it does not react with silver at all and it can be easily cleaned. So I think there is not better material than teflon.

Teflon is the choice for tube sockets too. This material is found to have the best sonic properties, since it gives very low acoustic coupling of the tubes to the chassis. This greatly reduces the microphonic effects, better than any other material can do. The disadvantage is that teflon sockets are not easy to find and they are very expensive.

For the I/O, superior quality plugs have been chosen. Audio Note AN-STR "MEISHU" silver plated speaker terminals and CMC brand pure silver RCA female connectors with teflon insulation.

The mains switch is a military one, pulled out from an old aircraft surplus device. A better switch couldn't be chosen. These are super long lasting and solid. The switch is double bank and it has three positions. Position SW1 (a and b bank on), position OFF (a and b bank off) and position SW2 (a and b bank on). By using this, I can choose manually the time for the protection of the anodes of the tubes. First switch to position SW1 to preheat the tubes for as long as you like. I always like things to be made manually so I can control them better. Then switch to position SW2 quickly (quick enough not to turn off the filaments) to apply the high voltage. It is a technique that I apply to all my constructions that require manually preheating and high voltage applying with just a single switch and no semiconductors.

Ciuffoli suggests not to use a mains filter because he does not like it's sound. Well, I disagree about this. A clean mains means clean power to the tubes which results always in cleaner sound. Much filtering is done in the PSU and many money is spent for a clean and stable PSU. So Why not having a cleaner mains too? Mains filters are bandpass filters that do not allow frequencies below or above 50-60Hz to pass into the power transformer. There is an even better solution (although much more expensive) that also keeps the frequency and the voltage of the mains stable. Power generators at power stations rotate at 50-60Hz but they are not stable neither provide a stable voltage. There are devices that can be connected in a way similar to the connection of a UPS at a computer and they keep the frequency of the mains stable. They also prevent from mains surges and they also keep the voltage of the mains stable. They provide more than one plugs, to plug in many devices. My cousin has tried such a device in his cheap Chinese EL34 push pull amplifier and he was surprised to actually listen to the great difference in the sound. Thus, I would not use a mains filter inside the amplifier, but I would definitely use one of these devices.


The main schematic

The LL1660/10mA in the 1:2.25 configuration have a right gap for 20mA.

If this amplifier will be used with a tube preampl. use a 100K 1/4W Caddock resistance instead the stepper attenuator.

If this amplifier is used after a CDplayer use the DACT 100K +100K stepper attenuator.



Power supply


The 0.56 ohm resistors could be adjusted to have correct 5V on filaments of tubes and a value range from 0.47 ohm to 0.68 ohm should be good to test and match any rectifier.

I have used 2 x 1Kohm MP816 Caddock resistors for the automatic bias of the 2A3 but a single normal 470ohm 25W could be used but sonic performance could be compromised.

Also this resistors could be adjusted to have the perfect bias current 120mA.



Digital Input stage

This project give a first class sonic performance but this can be easily compromised by a  bad input signal so I suggest you to don't use a preamplifier. Instead, take the output directly from a DAC, unless a phono stage must be used.
Using one of Ciuffoli's top DAC projects, all the DIYer's will reach the same sonic result.
The DAC normally included into commercial CD player (even these of significant price) have a rather low performance because the common usage of high NFB operational amplifiers "destroy" the sound given compression effects and poor soundstage stability.



Sonic results

The sonic performances are the best of all my projects!
This is a very reference amplifier and I suggest it to everyone if the power output is enough.



Layout views

click on the images to get real scale




Layout size

Attention:    some 100uF+100uF 500V ELNA Cerafine capacitors are higher so you should use an higher chassis or you can use 100uF+100uF 350V!

click on the images to get real scale


3 mm aluminum with external size 490 x 220 mm


3 mm aluminum with external size 490 x 120 mm



Project stage: Collection of components

Components photos below...

Big image    Big image    Big image

Big image    Big image    Big image

Big image    Big image    Big image

Big image    Big image    Big image

Big image    Big image    Big image

Big image    Big image    Big image

Big image    Big image    Big image

Big image    Big image    Big image
Big image    Big image   

You are free to copy any images and data in this site provided that you do not alter their content and you clearly define their source.
Copyright 711 © 2007 - All Rights reserved