Spent four hours yesterday staring at an oscilloscope trace trying to figure out why the unlit segments on an old Futaba 12-digit panel were dimly glowing. Turned out to be the ancient UDN2981 driver board I pulled from a scrap bin taking forever to discharge the grid capacitence. People constantly blame internal grid-anode leakage on these tubes, yet they forget bipolar source drivers have atrocious turn-off delays that often stretch past four microseconds without aggressive pull-down resistors. Slapping 10k resistors across every grid to bleed off charge before the next digit cycle starts feels barbaric, wasting heat on a display setup that already pulls enough current. When you try pushing a tight blanking period at 500Hz to prevent camera flicker, the trailing edge from the preceding digit smears directly across the next active frame. Switching to discrete p-channel mosfets would clear it up instantly, though hand-soldering thirty SOT-23 packages on protoboard sounds like absolute misery. Modern dedicated high-voltage driver ICs are either permanently out of stock or marked up tenfold by surplus brokers, leaving few practical options besides babysitting thirty-year-old chips.