The power transformer of the first prototype of the power amplifier, was delivering too high a plate voltage. I needed a power transformer better suited for EL84's. I also wanted a neater chassis.
So I bought a set of Geloso transformers through our national second-hand-stuff site. The seller said these were from a 6V6 push-pull amplifier kit. The output transformer had a lot of secondary taps, so I thought it should be easy to get the right impedance ratio to use it with a pair of EL84's. Their specified plate-to-plate impedance is similar to the 6V6 anyway.
Trying to find the more information on my transformers, I browsed through Geloso amplifier designs. Finally I found my transformers looked like the ones from the Geloso G215AN amplifier. They have identical connections and the wire colours correspond. By the way, the G215AN has EL84 output valves, not 6V6's ! It has fixed bias, no cathode resistor and the output stage is supplied directly from the buffer capacitor, not through a choke. Both these design choices will increase the available plate voltage. The separate 6.3 V heater windings on the G215AN power transformer were used to power the heaters of the EZ81 rectifier separately from the output valves. That is prudent: although the EZ81 is specified for a cathode-to-heater voltage of 500V, it is known to short occasionally. The enigmatic 36V winding was used to obtain a DC voltage for the heaters of the two pre-amplifier valves. These are put in series, each 12.6V, 150mA. Using DC for the heaters will reduce hum. Neat. The bias voltage is also taken from the DC heater voltage. Smart. Since the 6.3V heater windings originally were delivering less than 1.5 A and look like they cannot supply the full 4 A the amp will need, I decided to use the 36 V winding as a DC heater supply for three valves in the pre-amplifier, using a modern 3-pin voltage regulator.
I decided to roughly follow the overall set-up of the AC15. That one has two subchassis, bolted together in an L-shape. One subchassis holds the rectifier and output valves while the other one carries all the low-power valves. The chassis is mounted in the top of the cabinet, with the controls on top, at the rear. This way, it is easy to slide it out but it is more difficult to create an air flow to cool the output and rectifier valves. VOX had two vents in the top of the cabinet. On first sight, I don't like that because that may create an acoustic short-circuit hampering bass performance. And I don't have the nice grills they had to cover the vents. It probably isn't all that bad because although the path from the vents to the front is shorter than from back to front, the distance from these vents to the back of the speaker is larger than the distance from the back of the cabinet to the speaker, the speaker being placed towards the bottom of the cabinet. Well, VOX must have known what they were doing. Who am I to criticise them? Nevertheless, I decided to reverse the chassis and place the pre-amplifier frame at the front top of the cabinet. This way, it is easy to create an air flow through the back and do away with the vents. The chassis will be placed upright like the original. I will mount the chassis on a board that will be fixed to the front and back of the cabinet.
I also made a change to the distribution of the components over the two sub-chassis. I didn't like the phase splitter valve living on the pre-amplifier chassis, separated from the output valves. And this way, I could build and test the power amplifier as a self- contained unit before adding the pre-amp, an approach I had found useful for my first prototype.
Then I designed the tagboard that holds most of the components. I prepared the cabling in a cable harness, stringing the wires together using wax cord. After mounting the tagboard and the cable harness, I wired the valve sockets and soldered the remaining components. Then I placed the choke and transformers. The power transformer has a peculiar double mains voltage selector, which I mounted on a piece of aluminium L profile at the side of the chassis. The output transformer has an impedance selector. I mounted it on a second aluminium L profile on the opposite side.
The output valves were taking 68 mA, as I deduced by measuring the voltage drop over the cathode resistor and over the power supply choke. The phase splitter was biased as it should.
I inserted an input signal. I got a clean output of about 8 W. Increasing to the point where distortion became visible on the scope, I got 10 W. At that point, the current through the output valves had increased to 87 mA. With the output stage badly overdriven (last oscillogram) the output valves were drawing 106 mA.
So this amplifier was working fine. I experimented to increase output power. I tried a different set of EL84's, which actually decreased output power. I replaced the choke, which had a resistance of 260 Ω by one with only 150 Ω. This raised the plate voltage by 8 V. Now I had 8.4 W. Experimenting with the load resistor, I found out that at 14 Ω I got a clean output power of about 9.4 W. So my amp was not optimally adapted to an 8 Ω load. Not necessarily a bad thing since I also can use a 15 Ω speakers.
Time to start building the pre-amplifier.