Introduction
What makes valve amplifiers that special? Are it the valves, the special output transformers or is there also another reason? In the year 2004 I started an in depth research where I studied the effect of different topologies on the sound quality. I developed a kind of universal amplifier, an universal output transformer and power supply. Only by changing some internal wiring totally different amplifiers could be made with powers reaching from 5 up to almost 80 Watt. Essential in this research is that in every amp exactly the same components were used. So, only the influence of the schematics would be heard and nothing else. This complete research can be found at this site, see "The Project" and also literature 1) to 5). Last year I selected the best sounding and most powerful amplifier (amp 20) of this research, added some important improvements, and the PR20HE-S2 amplifier was born.

Audio schematics
Figure 1 shows the schematics of the audio part. The input on the left side is balanced and meant to be driven by a balanced output of a pre amp.


Figure 1: Audio schematics of the VDV PR20HE-S2

With a minor change of R3 and R7 also an unbalanced input is possible; see the remarks in the schematics.
The long tailed ECC81 input valve functions as pre amp and phase inverter and driver of the push pull power valves. The two cathodes are connected to each other, while R6 functions as a constant current source, fed by the negative voltage Vn = -73 V.
One might consider to limit the wide frequency range of this section, by adding a C1 = 100pF/500V silver-mica capacitor directly between the two anodes of the ECC81. This will suppress a small 90 kHz resonance in the output transformer. My experiments only showed a result on the oscilloscope, I could hear no difference between with or without C1.
After the ECC81 the audio signal goes through the coupling capacitors C3 and C4 to the power valves. The circuit is optimized for the famous EL156 power valves. They can handle a maximum anode dissipation of 50 Watt each, which enables a large quiescent current. The amp functions therefore mostly deep in class A, which also explains its great audio qualities. Also the famous 6550 valves with Pa,max = 40 Watt can be used, although with a little less quiescent current, see the table with measurement results for more details.
With the 10 turns trim pots P12 and P13 the quiescent currents Io can be set, and measured with a volt meter over the cathode resistors. With the EL156 the quiescent current are set at 50mA per valve, so the voltmeter should read 50 mV over R17 and over R18.
The anodes of the power valves are connected directly to the primary winding of the universal VDV-GIT80 output transformer. See "transformers" at this site for detailed specs and "order" for buying. Its primary impedance Zaa equals 8 kOhm in this application. The high voltage V0 = 720 V is connected to the center tap of the primary winding. The screen grids are connected with R16 and R17 to the supply voltage V1 = 360 V.
Now that the voltages and currents are known, the actual anode dissipation per power valve can be calculated. Using 720 V times 50 mA delivers a mere Pa = 36 Watt. This is smaller than the maximum value of 50 Watt, but ensures long valve life. However, if one wishes to go as deep as possible into class A, then the maximum quiescent current can be set at 69,4 mA. Probably the anodes of the EL156 will start to glow under this condition and please don't blame me that the lifespan of the two valves will be short.
The cathodes of the power valves are connected through R17 and R18 to the secondary of the VDV-GIT80. Make sure that the secondary winding tap point 2 (blue wire) is connected to the ground, else no current can flow through the power valves. This kind of local feedback at the cathode is used by Audio Research, and also proves in this design to have excellent qualities. The feedback raises the damping factor at the speaker and lowers the harmonic distortion in the power valves and the output transformer.
Loudspeakers with impedances of 4 (use tap 1 and 2) and 8 (use tap 1 and 3) and 16 Ohm (use tap 1 and 4) can be connected to the secondary. Never connect tap 1 or tap 3 or tap 4 to ground, else you will shortcut the secondary. Only tap 2 is connected to ground.

Figure 2: Pin numbers of the ECC81 and the EL156/6550-C.

Figure 2 shows the pin numbers of the EL156 valves with octal base, which are equal to the 6550 pin numbers.

Power Supply
Figure 3 shows the schematics of the power supply. The special power transformer VDV-POW80 can be ordered at this site (see 7 and 9) and be used worldwide because of its many primary voltages.


Figure 3: Power supply

Keep in mind that per voltage an optimal primary surge should be used, which is indicated as well. The mains grounding is firmly connected directly to the metal case, for optimal safety. Please pay close attention to this. The transformer has a metal shield (green-yellow wire) between the primary and the secondary windings, that should also be connected to the metal case with the shortest wiring possible.
The two 270 V secondary windings are separately switched with the stand-by switch S2 and surged and rectified and buffered with 330 uF in parallel with bleeders of 100 kOhm. This delivers two times 360 V and by placing these high voltages in series one gets the V0 = 720 V for the power valves and the V1 = 360 V for the screen grids.
The third winding of 50 V delivers after rectification and buffering the negative voltage Vn = -72 V. On purpose there is NO surge in this part of the circuit, to ensure that this negative voltage is always present. Else your precious power valves could burn in front of your eyes, caused by a stupid malfunction of a surge of 20 cents.
The filament winding of 6,3 V needs no further explanation, only that it is center grounded through two 100 Ohm resistors at star ground 2.

Specifications
They are given in the table and most speak for them selves. The -3dB power bandwidth start at the excellent low frequency of 18 Hz. At a low level output power of say 2 Watt (which is mostly where we listen), the low frequency bandwidth starts at 4 Hz. This very low frequency is caused by the extreme large primary inductance of the primary winding (Lp,max > 1000 Henry). The listening result is that you can hear, depending on the loudspeakers used, to low frequency breathing of the air, which gives much spatial information in the sound stage.
At the high frequency side of the power bandwidth I measured 26 kHz, independent of the output power level. This is caused by the leakage inductance Lsp and the internal primary capacitance Cip inside the output transformer. These results are in total agreement with my design goals of the special VDV-GIT80 output transformer, which should function flawlessly in our hearing range. Suppose one wishes to go up to 100 kHz bandwidth, then please use my specialist VDV4070-CFB toroidal output transformer.
The effective output impedance Zout (reverse with the damping factor) of the amp depends on the quiescent current and the type of power valves. See the table for details. Its value is small because (high damping) of the applied cathode feedback. Most dynamic loudspeakers will be totally happy with this damping and my listening results showed a quick and open and detailed bass character. No muddy sounds here.
The input sensitivity of the total amplifier equals 2 Vrms. Most CD-players deliver this signal level directly and then you only need a volume pot to set the desired loudness in your room. Use a stereo type when you work balanced, and use a single deck type for unbalanced application.
The hum level at the is remarkably small at 2 mVpp. This is caused by the magnetic leakage fields of the power transformer directly into the output transformer. By creating a larger distance between these two transformers, this level can be reduced, but I consider 2mVpp negligible, so why to make the amplifier larger? The tube circuitry delivers 1 mVpp hum because of the careful grounding method as described earlier. You can hear and measure no hiss, which is explained by the careful designed amplification factor, meaning no more amplification than absolutely needed for (which is a typical Menno trick).

VDV-PR20HE specifications	EL156	EL156	6550-C	unit
Quiescent current Io per valve	40	50	40	mA
V0	726	716	728	V
V1	363	358	364	V
Va1 upper EC(C)81	199	199	199	V
Va2 lower E(C)C81	196	196	196	V
Vn	-72	-72	-72	V
Vg1 eindbuizen (indicative)	-23	-22	-43	V
P-max @ 1kHz	71	73	75	W
-3dB power bandwidth	18 - 26	18 - 26	18 - 28	Hz - kHz
-3dB frequency range @ 2 W	4 - 27	4 - 27	7 - 27	Hz - kHz
Z-out at 4 Ohm tap 3	2,5	2,3	3,5	Ohm
V-in asymmetrical @ 70 Watt	2,0	2,0	2,5	Vrms
Hum at 4 Ohm tap 3 output	4	3	4	mVpp

Literature and sources:
1) www.mennovanderveen.nl ; the project
2) www.aes.org ; paper 6347 Barcelona 2005
3) AudioXpress January 2006, pp. 6 - 19
4) Menno van der Veen: "High-End Buizenversterkers 2"; chapter 9
5) see 1) project 28: the challenge
6) see 1) project 8: de VDV-GIT80
7) see 1) project 9: de VDV-POW80
8) www.addonamp.eu
9) www.mennovanderveen.nl ; transformers
10) Questions: info@mennovanderveen.nl