The previous part of this series showed a scaled down version of this concept based on the 6CB5A which minimized cost mainly by reduction of the iron content. But most people who were interested in an amp with the 6CB5A asked what can be done to get the best possible sound out of it:
The most obvious way is of course to choose the best possible quality of interstage and output transformers. While the Lundahl transformers which were mentioned in previous posts will provide excellent sound quality, this can be improved by the use of suitable transformers from the Tango range. These come at a much higher cost though.
The very early prototype of this amp has been built with the Tango XE20S which is an excellent performer. One level up is the type FC30-3.5S. Tango makes one of the best interstage transformers available, the famous NC20, which is especially well suited for the highish plate resistance of the 6N7. A change to these transformers does not require any other change to the circuit.
The photo above shows the most elaborate implementation of the 6CB5A amplifier so far with Tango NC20 and FC30-3.5S transformers.
Another possibility to scale up is the power supply. Rather than splitting the amp up into mono blocks, a split between amp and power supply is the better option in my opinion. Not only does it provide good isolation of signal section and PSU but also more room to use a heavy duty power transformer. The external PSU also provides enough space for 4 tube sockets which enable the use of an all vacuum tube rectifier bridge instead of the more traditional full wave rectifier scheme with a center tapped secondary.
The schematic above shows the power supply as I use it in amps with external PSUs. It has a LCLC filter section as the previously shown supply. The transformer secondary only needs to be about half the voltage. No center tap required. The advantage of this rectifier scheme becomes obvious when you compare the voltage waveforms after the rectifier, or the secondary voltage on a scope. In the common full wave scheme the two diodes 'fight' with each other during the switch over between phases. This causes a distorted waveform. The full wave 'Graetz' bridge as shown above has a much smoother switching behaviour which means less potential interference. The rectifier bridge is composed of 4 6AX4 TV damper diodes. The 6AX4 deserves a separate post on it's own, so I'm not going much into detail about it here. All heaters of the 6AX4 can be wired in parallel and fed from a dedicated heater winding which should be referenced to ground, simply connecting one end of the winding to ground is sufficient. All other aspects of the PSU are the same as in the previously introduced version.
Check out the blue glow of the 866As in operation and you understand why. Using mercury vapour rectifiers adds some complexity to the circuit since their filaments need to be preheated before the high voltage can be applied.
So a separate filament transformer and a delay circuit was necessary to implement this preheating mechanism. A manual override switch for this mechanism is added on the back to allow longer preheating of 866As when new tubes are used the first time. All connections are brought out on the back side of the chassis.