iC-Haus makes off the shelf diode driver chips with the required bandwidth. So does Analog Devices.
The diode in question requires a current limited constant current source. It will faithfully follow a waveform, no need for light feedback.
Do not "spike" the diode by trying to drive it with a current limited bench power supply. Most constant current bench PSUs in labs overshoot like crazy on a microsecond scale.
The preferred hobby level test driver is an LM317 current source made with a metal film resistor. Lm317 has a slow enough start-up as to be very safe as a LD test driver for CW work.
Do not use a potentiometer with the LM317, the wiper "lifts" as you turn it.
The LD is sensitive to static discharge, it will need heatsinking, and collimating optics.
Laser hobbyists buy collimators from DTRs Laser Shop, Meredith Instruments in US or Roithener Laserteknic in Europe. Get at least a glass lens one.
If the module your building will be handled by humans, install a Lasorb across the diode within 4 inches of its body.
www.lasorb.com and a reverse protection diode in parallel. ESD kills little green diodes.
Laser Diodes drift in wavelength around 0.2 nm per degree C. The green ones are tunable over +/- 3.5 nm with grating feedback. Which OP does NOT need.
EOMs are expensive, the external modulator of choice would be an AOM, but OP should not need it.
Avoid bright back reflections into the laser diode.
Buy many extra LDs until you learn how to get the design right. Trust me on that.
50 mW is at the point where you need at least OD4 laser safety glasses,. At least until you learn how to contain the beam. Thorlabs.com has those, probably in OD6 or OD3 for "Alignment" glasses that allow you to see the beam. Watch out for "specular" reflections as you learn. OSRAM has a good laser safety app note, read it!
Optical bench plates from Base-Labs and optical specific parts from Thorlabs or Newport are a very good idea if precision exposing film.
You probably want to learn about first surface mirrors, half inch optical posts and mm1 or km1 mirror mounts.
Single mode green LDs have elliptical beams, down the road you may need correcting prism sets or cylindrical lens sets from Chinese Laser Show projector parts vendors [30$ a set]
For really precise work you may need to learn about Spatial Filtering, but get the current source working first.
Find a copy of Phil Hobbs book, Electro-optics, Making it All Work.
Your idea of using a DAC for current control will probably rapidly evolve to PWM in most applications, with switching the diode from just below threshold current to some desired level.
I used to use daisy chained 74hc85 magnitude comparators and synchronous counters to generate PWM for brightness control with on/off style AOMs. That way all you had to do was send a Byte to the modulator circuit, no DAC required. it's a neat trick for delivering controlled amounts of light, without struggling to write timer/counter code in a processor. That will also work with LD driver chips. You can build the same concept in an FPGA, but the discrete parts are far cheaper.
Analog current control is very possible.
Thorlabs Det-10A or their other models are great fast photodetectors for use with an oscilloscope. Don't shine the laser directly into cameras or photodides. Scattered light or a glass wedge, or AR coated plate as a beam splitter is preferred as a sampler.
LaserSteve