This project is a second revision of a power supply PCB for a 2x10 35 watt guitar amplifier. The supply provides 420V for plates and screens of a pair of 6L6GC's, 385V for the amp's phase inverter, and 10-15mA of supply current for preamp. This board is intended for use with an Antek 2T350 toroidal power transformer. One of the two 6.3V @ 4A filaments windings provides AC filament power for PI and PA tubes, while the other drives a voltage doubler and regulator on-board to provide 12VDC for preamp filaments. Also included is a fixed bias supply providing -35 to -45V DC for the amp's output stage; the bias supply is capacitively coupled to the HV windings on the power transformer. Finally, a power NMOSfet clamp holds the PI high-voltage supply to 400V during warmup to avoid overstressing the preamp tubes.
This revision makes several changes from the original: (1) A standby switch is added. Standby switches are a bit controversial, but they are cutomary and useful to guitar players, so needed to be added to this design. The HV supply is interrupted on the negative side of the bridge rectifier. (3) The dropping resistor, needed to tame the maximum plate supply voltage a bit, is included in the standy switch loop, external to the PC board, so as to remove the heat dissipation from the board in order to spare the surrounding capacitors. In this revision, the 75-ohm dropping resistor is mounted to the amp's chassis for more efficient heat removal. Additional air cooling space is also provided around the PI's HV supply dropping resistor. (4) The original connects to transformer leads, AC power line, and all loads were provided by PTH soldered connections. In this revision, screw-type headers are used to make these connections in order to facilitate assembly/disassembly and repair.
Board in fab 12/7/19..
Board installed into amp 2/25/20, and amp fully operational upon completion.
Had some second thoughts about this design, after the amp was put into service. It's doing well, and currently making some really great Funk sounds with the local band JULIA, but.. I found I'd designed in some issues arising from an evil interaction between the capacitively coupled bias supply and the negative-plate-supply-lead standby switch. The problem here is that, when in standby, the ground-facing diodes in the B+ supply are turned off, but the B+ facing ones are, of course, still connected to the B+ filter. The bias supply continues to operate, and over the space of a few seconds, the coupling capacitors charge up, partly through the B+-facing diodes in the HV bridge, to the peak AC voltage (about -540V) of the HV secondary, so that both HV terminals have an AC on them varying between -540V and -1180V! The Carling toggle I'm using for the standby switch is rated only at 250VAC/DC, but a careful reading of its spec (these switches have been used for over 5 decades in Fender guitar amps!) indicates that its dielectric strength is good for 1KV. So, the switch is a bit over-stressed, though not yet showing any signs of failure.
After thinking about this, and the issue that the bias supply is regulated and thus doesn't vary with AC line voltage like most conventional guitar-amp bias supplies, I've decided in might be wise to ditch the capacitively coupled design for the bias supply. Currently, I'm working on a 3rd revision that uses a separate tiny transformer for the bias supply (to deal with line variation, and to get the bias supply divorced from the HV plate supply). With that change, the standby switch can either stay where it is, in the negative part of the HV supply, or move to the conventional positive position, between rectifier bridge and capacitor filter input (to deal with the very high voltage on the switch in the present arrangement).
As an aside, I really don't like standby switches. They can cause, if left in standby for long periods, irreversible chemical damage to power-tube cathodes. It seems to me that a "mute" function that disables the audio signal chain, most likely at the phase splitter input, might be a better way to get the function that most guitar players actually want. The 'Cadillac' way to implement this is with a pair of LED/LDR soft switches. Next time!!