The power supply uses one RCA-83 mercury rectifier. An Omron timer is used to delay the B+ in order to allow the rectifier to warm up. The last 2 RC sections (3k6, 220uF, 3k9, 10uF) are separate for each channel, but in future I plan to remove that and use a CCS for each channel instead. The 10uF capacitor is polystyrene. Yes, those are 50H chokes!
The RCA-83 rectifier uses 5VAC for the filaments. The heaters for the 5687 tubes are provided from a full bridge of MUR860 diodes, followed by 5 x 10,000uF Elna caps. 830mA of current regulation is supplied using a LM317 regulator and a 1R5 resistor between ADJ and VOUT pins. Votlage to the tubes is about 11.5V, so I am actually running starved filaments.
The preamplifier uses the double section of a 5687 tube for each channel, with each section using it's own grid stopper (220R) and it's own pair of green LED for a 4V bias. The double section is used so as to lower the effective tube impedance, resulting in better compatibility with the Tamura A4717 5k output transformers. The 5687 tubes are operated at 115V, -4V and 25mA, which is one of the more linear regions for a 5687. Initially, I tried AC for the filaments and the result was far too much hum, so DC is used for the heaters. The schematic of the preamplifier section is shown in Figure 01.
