You guys are overthinking the problem. All a 936 station is just a circuit applying voltage to a low-resistance heater (a RESISTOR) and using feedback to control the power from a linear positive temperature coefficient sensor already conveniently installed in the handle.
Here's how I handle it: Make the idea of turning a few bucks of parts into a fun, software and hardware learning project, and ending up with a nice soldering station.
First requirement: Heat the heating element:
I have several orphaned laptop 18-21V 2.5Amp to 3.5 Amp rated power supplies sitting around and several old desktop PC power supplies and a lot of MCU chips. For power I decided to use two bricks in series, mounted underneath my work bench to be out of the way.
Heating problem solved:
I use two laptop power supplies in series to develop about 37V DC at 2-3.5 amps. I make a simple step-down converter to convert the 37V DC to approx. 30V DC. I do this using a PIC MCU PWM output and so I can directly control the power applied to the heating element by varying the PWM duty cycle of the buck convertor from 0 to 100%. A single NMOS (30N05), an inductor, a catch diode, a 18V zener, a small signal(1N4184 type) diode and a few small-signal transistors, a 78L05 regulator, a few resistors, and a few free capacitors, a couple LEDs, an Arduino or PIC MCU, a little slice of perfboard, and bingo, a precise, high-power buck converter/Lo-Side Switch/soldering station.
Using the bottom 18V brick of the two bricks in series, I can use this tap output to power a 78L05 regulator to provide regulated power and a 5V reference voltage for the MCU chip. Using a single transistor and a zener diode or LED and three resistors I create a precision and stable constant current source to drive the thermistor temperature sensor in the handle.
By biasing the positive coefficient thermistor with a small and constant current, you feed the soldering tip temperature sensor signal directly into an A2D input on any MCU. By then using the just about any potentiometer connected to the 5V supply and to an A2D input of the MCU (and conveniently mounted on a small case serving as the front panel that can also be used to mount a solder handle rest and a sponge and an on/off sw.), you then have a good-looking functional temperature control that controls the temperature of the tip with a tiny bit of software.
All that is needed now is very simple software code:
If the the iron is cold, increase the PWM drive to the tip slowly (It should take approx 15- 30 seconds to reach soldering temperature from the tip at room temperature while not overloading the power supply bricks.)
Once the tip approaches the setpoint, the software code reduces the PWM to proportionally approach the required tip temperature without overshoot. A simple proportional heating method is not so difficult to code nor does the code have to very complicated to accomplish this.
Once the tip temperature is reached, PWM duty cycle reduces to zero and increases again as the tip cools by the soldering operation or just cooling off by ambient air.
Here's where the used PC power supply comes in. A large 50W toroidial inductor is needed for the buck converter and it is already found and already wound and ready to use inside any PC power supply. The PC power supply also is the source of free high-current schottky diodes to connect the drain of the N-MOSFET to act as the required catch diode for the low-side sw/buck converter and the PC power supply also provides for free the small heatsink ans well as the output filtering capacitors for the buck convertor and control MCU as well.
Since a MCU PWM output can be controlled in very tiny duty cycle steps, it is easy to adjust duty cycle to match a wide range of inductance to form a buck converter, since PWM can be controlled in 0 to 1023 steps with a fixed 20KHz to 40KHz PWM frequency.
Since I have approx 38-V available from the two laptop bricks in series, I already have a convenient bias supply to drive the N-MOSFET gate. I just need voltage translation to control the MOSFET buck low-side switch.
I use a grounded base BJT transistor (2N3904) whose emitter is connected to a MCU I/O pin and a 470-ohm resistor to ground. The collector connects to the common bases of a NPN-PNP complementary pair (2N3904 2N3906) that is used to drive the gate of the MOSFET. The common-base BJT transistor's base is biased to 1.8V using a single LED and a resistor to the +5V.
A zener diode connected in series with a 1N4184 diode from the complementary transistor bases to the source of the N-MOSFET limits the gate to source voltage to <20V. The NPN-PNP complementary pair is biased with a single 100K resistor to the top of the brick pair output (approx 38-40V).
If you wanna little fancy, with just a little more code and three 7-Seg LED displays, add three resistors and just directly connecting to the 8-pins of a single output port on the MCU tied to the segments and three more I/O pins to drive the CA/CC's, and you have all you need to drive the MTPLX'd segments and common anodes/cathodes. The code needed is easy to add mtplx'd 7-seg digital readout of the temperature. The resistance v. temperature of the sensor in the handle can also be found on this board to eliminate calibration, but requires a little application of ohm's law to your MCU software code.
In fact, only a single LED on the "front panel" can show heat-on heat-off, ready to solder temp, with only a little code from the MCU.
The fact is, even without any calibration or readout, there exists a precise setting of a control pot that will give you the right soldering temperature as a setpoint that compares to the handle sensor output . You will easily know this "nice" setting by seeing how the solder melts and solders as you slowly increase the temperature by adj. the pot which controls the balance point of this feedback loop controlling the soldering tip's temperature.