After a few weeks of waiting, I managed to get all the components I need for start.
For example, this nice transformer. Yep, it's 5kV.
I'm going to inscribe the long period grating (LPG) into fiber using electric arc and this guy (and I have it's identical brother too) should help me with it.
Except of that, box of cheap chinese boards that are sometimes quite handy for quick experiments, some stepper motors with huge gear ratio (for scanning spectrometer), party galasses with polarization foil for spectromter difractor and plastic bag with IR photodiode.
I took out one of my projects I did many years ago, when PIC18F2550 was absolutely new hot chip. That PCB's in that ugly plastic enclosure are stepper motor controller with some regulating circuitry. Nowadays, I would do it completely differently, but nothing can stop me from repurposing this for initial experiments. I expect final version to be built on modern components.
Fortunately, I got my hands on precise and quite expensive translation stage (above the ugly plastic enclosure) and this will be good start for experiments with LPG inscribing (I'm going to need precise fiber movements for that). This translation stage will be definitely replaced by something low-cost, as it's price is one order of magnitude higher than whole budget.
Some optic fiber patchcords, PCB with monostable multivibrator to control arc timings, plastic bag with wolfram electrodes (replacement parts for proper fusion splicer) and finally - DVD ROM. No DIY optical project would be complete without parts from those.
Now I'm going to print some plastic parts on my MendelMax and proceed to practical experiments.