3 Why DC-coupling? The fasted way to test a new concept is: build an amp, then you soon will discover the validity of the concept and its problems. 3-1 The first try-out amplifier For my TubeSociety students I bought amplifier kits, see picture 3-1. Our intention was to upgrade these kits during the lessons. You can find our results on my website, see (12). In 2013 I used this amp to do my first Trans experiments. In the schematics all has changed. Only the case and mains transformer and valves stayed. Figure 3-1 shows my first try-out.

Photo 3-1: The original Mable amplifier

Fig. 3-1: Schematics of my first TRANS amplifier

The two EL84 power valves are in push-pull class-A configuration. Their quiescent currents (35 mA per valve) are set by the current sources (BD139 and BC547). These are AC-decoupled by the two 1000 µF capacitors. This circuit guarantees equal quiescent currents per valve, which is mandatory for the new toroidal output transformers. These are small Amplimo 17010 mains transformers with primary 2 x 115 V in series and secondary 2 x 6 V in parallel. The primary impedance then is 8 kΩ at 8 Ω secondary load. See figure 3-2 for the specs of these inexpensive output transformers.

WIDE BANDWIDTH TOROIDAL PUSH-PULL TUBE OUTPUT TRANSFORMER

Type and Application Amplimo 17010 als OPT in Mable.

Primary Impedance : Raa 8.141= [kΩ ]

Secondary Impedance : Rls 8= [Ω]

Turns Ratio Np/Ns : Ratio 31.9= [ ]

UL-tap: tap 0= [%]

Cathode Feedback Ratio : cfb 0= [%]

-.1 dB Frequency Range [Hz to kHz] (3) : flf 7.313= fhf 3.578=

-1 dB Frequency Range [Hz to kHz] (3) : fl1 3.119= fh1 8.088=

-3 dB Requency Range [Hz to kHz] (3) : fl3 1.587= fh3 15.558=

Nominal Power (1) : Pn 6= [W]

- 3 dB Power Bandwidth starting at : fu 14= [Hz]

Total primary Inductance (2) : Lp 410= [H]

Primary Leakage Inductance : lsp 19= [mH]

Effective Primary Capacitance : cip 2.7= [nF]

Total Primary DC Resistance : Rip 86.3= [Ω]

Total Secondary DC Resistance : Ris 0.13= [Ω]

Tubes Plate Resistance per section : ri 4= [kΩ ]

Insertion Loss : Iloss 0.115= [dB]

Q-factor 2nd order HF roll-off (5) : Q 0.411= [ ]

HF roll-off Specific Frequency (5) : Fo 31.776= [kHz]

Quality Factor (5) : QF 2.158 104= [ ]

Quality Decade Factor = log(QF) (5) : QDF 4.334= [ ]

Tuning Factor (5) : TF 0.454= [ ]

Tuning Decade Factor = log(TF) (5) : TDF 0.343= [ ]

Frequency Decade Factor (4,5) : FDF 3.991= [ ]

(1): calculated under the conditions of balancing the DC-currents and the AC-anode voltages of the powertubes driving the transformer(2): measured at 54Vrms at 60Hz over total primary(3): calculation at 1 Watt in Rls; ri and Rls are pure Ohmic(4): defined as FDF = log(fh3/fl3) = number of frequency decades transfered(5): ir. Menno van der Veen; Theory and Practise of Wide Bandwidth Toroidal

Output Transformers; preprint 3887, 97th AES Convention San Francisco(C): Copyright 1994 Vanderveen; Version 1.7; results date 15-08-2011.

Final specs can deviate 15% or improve without notice

Fig. 3-2: Audio specs of the 17010 toroidal transformer

The high voltage B+ is 300 V and B1+ is 285 V. The input tube (ECC88) is a differential amplifier (phase inverter) with a current source (BSP135 + BC547) at the common cathodes. The anodes of the ECC88 are at +50 V. They are directly coupled to the gates of the BSP135 Fet’s which function as current sources. When their VDS is larger than 5 V, the IDS-VDS graph is a horizontal, indicating that their inner RDS resistances are infinitive. This is the mandatory condition for a real current source; see figure 3-3.

Fig. 3-3: BSP135 characteristics

Each source is connected to ground through a 100 kΩ resistor. The current through the first Fet equals 50/100k = 0.5 mA. The drain is connected to the 220 kΩ feedback resistor R2 (indicated with **). The DC voltage at the drain equals the voltage B1+ minus the voltage drop over 220 kΩ (0.5 mA * 220 kΩ) = 300 – 110 = 190 V. The calculation is valid for the upper and lower Fet. If an AC audio signal is applied at the input, it appears amplified at the anodes of the ECC88. This changes the gate voltage and therefore the current through the Fet’s. The “transconductance” “g” can be made larger by connecting the upper and lower sources through a resistor of 47 kΩ, see figure 3-1. This topology ensures that the AC current variations in both Fet’s will be absolutely equal, however, in opposite phase.

The simple arrangement will largely reduce the second harmonic distortion. An extra condition for minimal distortion is also that the two R** = 220 kΩ resistors are exactly equal. The Fet’s are protected by 12 V zener diodes between gate and source. The next two Fet’s function as source followers. Their gate impedances are very large, so, the first Fet current sources and the two 220 kΩ resistors are not loaded. There is none leakage current. The output impedance of the source followers is small. This largely reduces the influence of the Miller capacitance inside the EL84 valves. The AC signal from the source follower goes to the control grid of the EL84 through a coupling capacitor of 1 nF. Why such a small capacitor is mandatory will be discussed later.

3-2 General qualities of the first amp

The technical specifications of this first try-out amplifier are:

1) -3dB frequency range from 12 Hz to 30 kHz 2) Pmax = 4 Watt per channel at 1 kHz in 8 Ω load 3) Output impedance = 0,5 Ω at 1 kHz 4) THD and IMD distortions are smaller than 0,1 % at 1 Watt in 8 Ω 5) The amp has a tendency to low frequency oscillation 6) The low frequency amplification factor (linearity) behaves strange 7) The output impedance is frequency dependent; a maximum at 3 Hz

(resonance?). It also gets larger towards 20 kHz. This is caused by the leakage inductance of the output transformer. At this moment I pay no further attention to this leakage effect.

The observations 5, 6 and 7 will be discussed later; they showed me that not all was correctly designed. The first 4 properties are excellent. They follow exactly the predictions of chapter 2. The subjective qualities of this amplifier are extremely positive. I demonstrated this prototype to my students and to members of audio clubs in The Netherlands. All showed the same behavior: they got straight in their chairs, head a little bowed to the speakers, intensive listening. Remarks were : “this is as it should sound; what a details and space; this is better than SE”. Did not I tell you in the introduction that I digged gold? So, the first try-out was not that bad! But I need a redesign to optimize the specs and power and …. to remove the strange findings 5 and 6 and 7. What is their cause? This will be explained following. Because of holidays and negotiations with the publisher: the end.