[rfeenstra]

That looks like a bit more extensive of a radio than mine.  maximum DC voltage in this thing is about 120 volts so the electrolytic cans were rated at 200 volts.  Most of the sections were 60mfd.  I used 60 mfd Nichicon's rated at 250 volts.  I purchased the caps with the longest life and highest ripple voltage capability.  This radio uses a ground buss that is separate from the chassis ground.  Chassis ground runs abut 7 to 10 volts negative compared to the ground buss (I can vary that voltage by changing the pilot lamp) and this provides the grid bias for the 25L6's.  The ground buss is the negative side of C-157.  The negative line is the neutral of the AC in.  Notice C-158.  The A partition is the first filter cap for the half wave rectifier.  The B portion filters the chassis ground to negative line.  One would think that chassis ground would be riding some small positive voltage above the negative line.  It does not.  Chassis ground is about .05 volts negative from the negative line, putting C-158B slightly in reverse polarity mode  Very slightly.  When I turn the radio off, that voltage jumps to about 20 volts reverse polarity and then slowly dissipates.  I get a little concerned about what this reverse polarity will do to the cap over time.  Interestingly, there was a 47k resistor across this cap that is not on the schematic, nor is it called out on the chassis  as every other component is.  I wonder if that resistor was added reduce the effect of the reverse polarity.

Any ideas from some of you EE people who can analyze the design better than I can?