@silversolder
I studied the diagram and did some tests on the actual unit. I do not measure 100V but 85V - I believe @Kleinstein mentioned that because of R1 and the capacitors the value was not going to be exactly VAC x 2.8 but more in the 80 range - which is exactly what I am measuring.
I now understand why R2 is reading 50V across it - the way you draw the circuit is very clear, thank you a lot!
Indeed I see the voltage slightly dropping when I connect a static load. Coincidentally, I played with Zeners just a few days ago on a breadboard and I know what you mean that a tiny change in voltage causes a large change in current - which is how Zeners work indeed. I have also tested with my thermal camera and indeed I can see the Zeners getting much hotter when I switch the unit off from the selector (which only disconnects the main PCB but not the PSU).
After replacing the resistors with a 5W and 2W versions and verified that the whole thing still burns like a star, I think I am happy to move to phase 3 and improve on the PSU as many here recommended
I understand the LM317 (and I guess LM337 for the negative rail) are the variable regulators of choice - Wouldn't a voltage regulator get very hot when dropping lots of volts though? . I do not have the skills to design something like that by myself so I would appreciate your help.
Based on the typical drawings found on the datasheets, it could be something like that - but this is the first time I am doing something like this so I am 100% sure that it'll make many of you cringe! Please be kind! (Yes, I know 3.024Kohm resistors do not exist, but I am struggling a bit finding the correct values so for now I put the ones given by the calculator!)
https://crcit.net/c/d71559e0902e4b2d84d78f4fad22bbb7
Thank you!
Learning by doing is the best way, well done!
For your improved circuit, the suggestions from @bdunham7 and @MarkF appear to be sound, although I have never actually used the LM337 or LM317 in a project - but these things tend to be quite docile.
The way these regulators work isn't like a zener diode. These things actually pass the entire current through a large pass transistor, which is controlled by a built-in op amp. This way, pretty much all the current goes to the load. Only a very small current flows in the "voltage setting" part of the circuit. So the power dissipated by them is equal to V*I, where V is the voltage drop across the regulator, and I is the current going to the load (I guess, 22mA max).
My penny's worth on the design discussion so far is that I would stick with fixed resistors instead of putting a pot or a trimmer in the circuit. The old design that is being replaced was not ultra precise, so using fixed 1% resistors will be more than good enough to set the voltage reliably.
The less parts used, the more reliable the result will be (all else being equal). The reason for disliking trimmers and pots is that over time, they have a tendency to go bad (usually by the wiper arm going open circuit or intermittent against the track). Depending on the circuit configuration, this can cause the voltage to either max out or drop to zero... there are commercial power supplies where it maxes out! - needless to say, that usually isn't good for the load connected to the supply...
The other thing to consider is that the improved circuit as it stands will be able to supply much more current than the 40 odd mA that the old one was capable of. It is probably wise to add resistors on the transformer outputs to the bridge as @bdunham7 suggested, in order to put a limit to this - making sure they are high enough wattage to handle a short circuit of infinite duration if necessary. Also, it would be wise to calculate how much power will be dissipated by the regulator(s) during a short circuit condition - this might influence your decision as to whether the regulators should be bolted onto the case for cooling, or if it is OK with a small heat sink internally, or even no heat sink at all!