AG9018 (same amp, but wooden box):

Notes:

-In all schematic R36 (and R136) is 100K, while my is from factory build with 220K. In case of 100K resistors the C28 (C128) 8uF has higher voltage rating 400V instead of 350V and is a little also bigger sized phisicaly.

-The R2, R4 shown in the AG9015 drawing is in reality 2 parallel resistor, which became incorrectly corrected in the AG9018 schematic, as the shown R62, R64 common part is not connecting to the R2-R4 common part, while in reality they do…

-B1 and B101 tubes in all the amps I was seeing was mixed compared to the assy drawing. The right tube was for the left channel and the left for the right, while all the rest was logical. The cable color is also very bogus. Red is mainly the Left, while white is the right, but not consistently (typical mix at height treble and at phono). Tape in was mixed up in all what I see, you need to rotate L and R to match with standard. Phono input is rotated to the Tuner (so Red is Left), but if you fix B1 – B101, then the color becomes equal to the Tuner. Some small pic to explain the issue:

In general about the circuit:

 The output stages of the AG9015/AG9018 are equipped with an ECC83 double triode. The first triode acts as a preamp, the second triode as a phase-reversing tube. The latter controls the “top” EL86. The power tubes work in class A setting, and each receive an anode voltage of 210 Volt at a current of 55 mA.

The end amplifier is using a special circuit with two EL86 tubes in cascode to secure low output impedance of 800 ohm. The low impedance is achieved with the circuit, where from AC perspective (sound) the tubes work parallel, while from DC perspective they work in series. (OTL mode is with speakers AD3800AM or with the power more matching AD9710AM). In the HF302, HF303 (preamp HF306), AG9014 Philips used a special bifilair choke to drive the screen grid of the pentodes. The choke secures efficient and equal distanced voltage from the anode of the corresponding tube (the 2 current flows in opposite direction, where current matches!) This is really much more efficient then burn energy on alternatively use resistors, where small resisitors would shortening music, while big resistrs would limit amplification… The choke resolves this issue. … This choke became part of the output transformer in AG9015. The 2 reverse direction winding with DC is not magnetizing the core as current flows in reverse direction. So the screen grids of the power tubes are each fed via the power transformer, so the efficiency of the amplifier is higher than with a circuit with screen grille resistors. The yield is close to the theoretically achievable (50%). The two inductors are combined on one core, which prevents premagnetization. The inductance of the coils is so high (60 Henri) that the effect of this is only visible below 20 Hz. The auxiliary winding (S5a) on this choke / transformer is connected to the output of the amplifier. The winding ratio of the transformer is 1: 1: 1 (in AG9014!). Due to this construction, DC voltage is never present at the output (transformer, capacitor limits it).

Positive feedback is generated through the common cathode resistance of the triodes. The negative feedback is created by returning the output signal via a resistor to the cathode of the first triode.

Effect of capacitance from the plate of the driver stage to ground: Since the plate voltage of the driver tube is taken from the output, the output voltage is fed back to this stage. This feedback is negative, and it can be considered useful for reducing distortion. However, when it is desired to avoid the
associated loss in gain, the feedback can be minimized by using a pentode driver.

The pentode works in triode mode as screen is tied to the corresponding tubes anode. The lower screen is on DC (as the cathode ), but the upper is via C29 following the waves, so following it’s cathode.

The increased power of 15 watts per channel was obtained by increasing the supply voltage to 420 volts (was 320 volts with the Philips AG9014). The EZ81 rectifiers from the AG9014 are actually a bit too tight for these voltages, so that with the AG9015/AG9018 Philips has opted for a heavy duty GZ34 rectifying tube.

You can read about the output transformerless version here (is basicaly an HF303 or AG9014), page number 85: