All planned inertial confinement fusion (ICF) capsule targets except machined beryllium require plastic mandrels with tight requirements on which the ablator is built. In this paper, the fabrication of poly(-methylstyrene) (PAMS) mandrel is studied. PAMS mandrels are produced by using microencapsulation technique. This technique involves producing a water droplet (W1) encapsulated by a flourobenzen (FB) solution of PAMS (O) with a droplet generator, and this droplet is then flushed off by external phase (W2), forming a water-in-oil-in-water (W1/O/W2) compound-emulsion droplet, which is suspended in a stirred flask filled with external phase to cure. The encapsulation process is based on a microfluid technique, which can achieve the controlled production of millimeter-scale PAMS mandrels. In this work, capillaries-based co-flowing microfluidic triple orifice generator is designed and built to fabricate W1/O/W2 droplets. Two configurations of the droplet generator:one-step device and two-step device, are employed in this experiment. In one-step device, the end of oil phase capillary is located at the same position as the end of inner water phase capillary. So the core droplet and the shell droplet break off from their capillaries ends at the same time, forming a W1/O/W2 droplet. While in the two-step device, the W1 phase capillary tip is located upstream to the W2 phase capillary tip. As a result, the core droplet and the shell droplet depart from the ends of their capillaries respectively, forming a W1/O/W2 droplet as well. Differently, the shell droplet contains only one core droplet in one-step generator, while several core droplets are contained in the shell droplet in two-step generator. In this paper, the mechanism of the droplet formation and the effect of the flow rate on the size of the droplet are studied with these two configurations. Results show that tiny difference between the two generators will lead to great differences in droplet formation mechanism and size control. In the two-step generator, the inner phase flow rate has little influence in the outer diameter of the compound-emulsion droplet. The diameters of the compound-emulsion droplets have a similar change to the diameters of the single droplets (O/W2). In one-step device, the inner phase flow rate has a significant influence on the outer diameter of the double-emulsion droplet because of the existence of W1-O interface. Finally, the compound-emulsion droplets fabricated in this experiment are cured in external phase, after which PAMS mandrels are fabricated. The diameters of the final PAMS mandrels are measured with optical microscope. The distribution of the diameters well concentrates in an area of (200010) upm, which is favorable for producing the PAMS mandrels with a diameter of 2000 upm.