OK Samuel, for the page 3 schematic, it's pretty simple , start our journey with the two LEDs
V15 and
V16.
-------- current control mode -----
if
V15 is illuminated, then you are in current mode, current through V15 drags down the drive to the pass transistors (v13,14,21,22). So how does it get into current mode? simple!, the voltage across the current sense resistor R10 exceeds the voltage across the current pot RP3. The difference in voltage is sensed by
N2 (a garden variety LM741).
--------- voltage control mode -----
if
v16 is illuminated, then you are in voltage mode, this happens if the output voltage is greater than the voltage (at C13) on the wiper of the voltage setting pot RP4 , RP2. , The difference in voltage is sensed by
N1 (a garden variety LM741).
--------- floating bias supplies ----
What make the whole circuit confusing is
the "gnd" of the circuit is actually the positive output terminal, and there are floating voltages referenced to this node, e.g. the TL431 generates 2.5000 V reference above the positive output terminal , there is +12v on C3 , and ~ -15v on C2 .
-------- tap changing ----
There are two relays K1 and K2 that change transformer taps according to the output voltage. M4A and M4C compare the output voltage with two setpoints one across C15 , the other across C17 , these two caps slow down the voltage levels so the relays don't switch unnecessarily with small voltage dips. R29 and R28 provide hysteresis so the realys don't chatter back and forth if you are close to the switching threshold. Note that the relays perform as primitive analog to digital convertors, so K1 adds say 5v to the transformer voltage, K2 adds 10v, and both add 15v. To get the A-to-D effect , V28 shorts out R23 , which adds 10v to the reference voltage of the 5v relay driver.
----------- stabilities --------
Almost all power supplies are basically unstable, as you have two or three lag terms in the feedback loop , there is a really good tech-note out there written decades ago that explains all this, I can't find it, but this isn't too bad, from about page 5 , this is the topology commonly used.
http://www.ti.com/lit/wp/snoa842/snoa842.pdf go ogle "voltage regulator dominant pole"
In any case the best solution to the stability prooblem is to use a SINGLE DOMINANT POLE , i.e basically make one part of the circuit so slow, that it swamps the delays in other circuit elements, in your circuit it's C7 in voltage mode, and C10 in current mode (you then need to add C8 and C11 to "compensate" the LM741 , because it is prone to oscillate internally with low local loop gain)
If you design your own PSU , you will also need to add a dominant pole somewhere, the mistake most newbies make is to have too much gain in the loop , you only need unity closed loop gain to make a PSU.
------------------- pg4 ---------------
Page 4 looks like a higher current (and maybe higher voltage) version of the same PSU , it adds an extra pass transistor , and an extra voltage tap , so the relay driver is now a 3 bit A-to-D, so you might have
taps at 5, 10, 15, 20, 25, 30, 35, 40v
for K3,2,1 = 000 001 010 011 100 101 110 111
(I'm just guessing these voltages BTW)
--------------------- fan --------
Most PSU's have a thermally activated fan on the heatsink
-------------------- accuracy -------
All the components used are very cheap, garden variety, so you can expect the last digit of a 4 digit LED display for volts and amps to be fairly meaningless