Thanks for the help. Especially taking time to run the simulations.

History: I should say, this is not a new design. It is typical of many vacuum tube radios in the 1940s - 1960s where making a line operated power supply without a transformer was very common and cheap -- many millions were made, they were 30-40 watts, it would not pass safety rules today. There was a point in time where the industry made a transition from using vacuum tube rectifier to using selenium rectifiers. That was a big advantage since there was no power wasting rectifier filament. The selenium rectifiers had a forward voltage of 5-7 volts for a 100 mA device, but the vacuum tube rectifiers had 20 or more volts lost. 1N400x had not yet been invented.

Clearly R1 dissipates much more power than R2, even though at first glance they both are in the same current path.

I am still somewhat confused about one thing. In the simulation I get the power pulse each cycle as you guys did, and when I measure as average I get about 4 watts which several of you did.

* The question now is whether AVERAGE is the right measurement for calculating the dissipation of the resistor, or should I use RMS power calculation which I get 9.6 watts ?

Also, in my start time and end time I used start = 400ms (integral cycle aligned) by which time things are stable, and end = 483.333 ms which is also an even 1/60 second frame. RMS calculations require a window of integral half cycles I think.

This result, even 4 watts, is surprising to me -- initially I was thinking in terms of I^2 * R . It was shocking, pun intended.