both links dead. anyway, i've been browsing the web for SRD the whole night and found some pretty interesting infos. however, it seems that not a whole lot of people make them. it makes it hard to get a few samples to test on.
hopefully i am able to get hold of some, then build it to test its speed and step purity. most of the papers done are based on spice simulation or high voltage step. i need steps between -5 to 5V.
in the mean time, i will try to use analog switches and see if i can hit the picosecond risetime. so far, my step generator can only touch 1ns (4.5v) using cmos buffers. the settling is quite good though, less than 5ns.