PP2010 - Push Pull test

Follows the schematic of my first test to check the quality of the output transformers in the most simple configuration: class A, triode connection, no feedback.

The second stage is a cathode follower used to decrease the output impedance and to drive the final stage with sufficient current.

The result of this circuit are not bad:

Vout=14.337Vrms with only 2% thd so about 30w on 8ohm

Vin=0.767Vrms

Output impedance near to 3ohm

Frequency band 10-20KHz at -1db

Global feedback about 0dB

Va = 430v Ia = 42mA x 4

To reduce the output impedance it is possible use the transformer secondary on cathode feedback like use by AudioResearch and Bartolomeo Aloia.

Using this method the gain is reduced of 3dB and output impedance is near to 2ohm (see measurement test 2).

SCHEMATIC TEST 2

In the triode connection the power is limited and the damping factor is not enough good also if we use the cathode feedback of the transformer secondary.
So I have decided to switch to the ultra linear connection also if this mean go near to the pentode configuration with an inevitable increment of the internal resistance of vacuum tubes.
Using this connection we get also more voltage gain in the final stage so the cathode feedback of the transformer secondary compensate the increment of internal resistance of vacuum tubes so the output impedance without a global feedback is only about 4ohm.
If we decide to use global feedback it is necessary to get more voltage gain so I have changed the first stage to a hybrid cascode adding a fet on input.

This new schematic give very good performances and add a little complexity.

Measurement result:

Vout=20Vrms with only 0.3% thd so about 50w on 8ohm

Vin=0.700Vrms

Output impedance near to 0.7ohm

Frequency band 10-20KHz at -1db

Global feedback about 6dB

Va = 430v Ia = 42mA x 4

The previous schematic run very well but many persons can prefer a pure vacuum tube configuration so I have decided to create a new design inspired to the AudioResearch reference amplifier but maintaining the good input differential stage.
The less common tube 5687 has been used but it can be replaced with few modifications with the E182CC and the ECC99 present in the actual production.
The performances of this design has been optimized with electronic simulation using the WinSpice 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.
In the direct comparison with the previous circuit in the simulation is visible a little bit limited band due to the driver of final tubes with anode instead of cathode but the audiophile person will see in this new schematic many others benefits for the fet elimination and no strong local feedback in the driver stage.
The sound of this version is not pleasant and the reason of this should be the low anodic voltage used on both stages to allow the dc coupling.
Using this circuit there was as an oscillation of the amplifier also without global feedback so seen these problems I have decided to came back on the original idea.

Measurement result:

Vout=20Vrms with only 0.5% thd so about 50w on 8ohm

Vin=0.666 Vrms

Output impedance near to 0.7ohm

Frequency band 10-20KHz at -1db

Global feedback about 4.5dB

Va = 430v Ia = 42mA x 4

To try to design a pure vacuum tubes amplifier I have tested the following configuration where the fet used in the cascode has been replaced with another 12AX7.
After many measurement tests the best result are obtain using the lower 12AX7 at only 30v.
In any case the distortion and the sensitivity is worst of the version with the fet.