The XRF is an 'oldie' -- a Philips 1410 spectrometer interfaced for data collection with an old Apple -- but it's a sturdy instrument, and with some operator patience, it provides good whole-rock compositional data. My approach is to use glass disks for Si, Ti, Al, Fe, Mg, Ca, K, and P, and to use pressed powders for Na, Rb, Sr, Ba, Zr, Nb, Y, La, Mn, and Ni.