Is it possible to build a reasonably high powered amplifier without using any resistors or any capacitors? This article describes a 10 watt single ended 845 amplifier that does that very thing: that is, there are no resistors, capacitors, batteries, solid state devices etc anywhere.
It uses the power supply previously described on these web pages that provides the necessary voltages to operate. The power supply likewise has no R's nor C's in it. The supply article can be reviewed by clicking here.
The input stage is a Class A differential 6DJ8 amplifier. The input transformer splits the phase of the signal. The stage is biased from 2 paralleled sections of a 6AL5 diode. This provides about 2 volt bias and at about 108 volts on the tubes, each section draws about 10 mA. The output couples into a Lundahl 1660 configured as a step up transformer.
Interestingly, the 6AL5 diode "biasing" seems to work out very well. This is similar to the other amplifiers described in the NO R NO C series.
The secondary of the driver transformer couples directly to the 845 grid. About 110VRMS is available before clip, at this point, which is more than sufficient to drive the 845.
Notice that there is an additional choke in series with the B+ feed to the differential amplifier. This provided better sonic performance, probably due to the additional isolation from the power source. Without it, the sound was rather -well- "hard" is probably the best description.
The 845 is biased with a 0C3 on the CT of the filament. This provides low enough impedance so that the 845 output resistance is not increased substantially. I tried both 0B3 and 0C3. The 0B3 caused a 65 mA current flow, and the 0C3 caused a 40 mA flow in the 845 from the supply discussed above. See below for the sonic differences.
The output couples through a Magnequest FS-100 output transformer to the speaker.
There were a number of variants to the circuitry I tried before settling on the schematic shown. I tried coupling the driver B+ to the 0C3 / cathode of 845, but the resulting amplifier was too distorted for my liking. I found I needed at least 110 volts on the differential stage to avoid "graininess" in the sound. The most meaningful measurement associated with this was the very high frequency response. The 6DJ8s needed that additional current from the higher voltage to drive the transformer and 845.
I also tried 0C3 (40 to 45 mA current in the 845) and 0B3 ( 60 to 65 mA in the 845). The 0B3 version sounded good at low volumes, but did not work nearly as well at higher volumes. This is probably a limitation of the power supply, as the ripple VOLTAGE at the 840 volt point (signal frequency ripple - not line ripple) increased substantially with the 0B3. What this seems to indicate is that we can indeed "hear" power supply effects. (there was much more of a audible difference than a measureable difference).
Listening to the amplifier as drawn on the schematic, tha amp is clean, punchy, very detailed. Suprisingly little hum. Measured response was -1dB at 19Hz and at 22kHz. Measured hum level was about 0.9 mV at the speaker. The low hum is attributable to only 2 reactive components in the power supply. This allows the phase of the "120 Hz" hum to somewhat counteract the filament 120Hz hum (quadrature) component.
This is not the best sounding amplifier I've built, but it is pretty good. I suspect it can be improved with a more exotic voltage regulator scheme, and perhaps providing a "no R no C" -105V bias to the grid and directly grouding the filament CT. This would remove any supply and/or VR tube coloration.
This thing is HEAVY, though; taking into account the amp and power supply.