I use a type S-11 light bulb, 120vac, standard base, round white globe about the size of a golf ball, in a light-tight enclosure with a ~1/4" aperture. I use it suspended about 30 inches above my work table in the darkroom. This permits me to use preflash exposure times of around 5-10 second range, compatible with my old analog Gralab timer. Being incandescent, the filament has to warm up to get its full spectral response, which it can't adequately do with really short exposure times.

To calibrate the amount of preflash, I did a series of tests in 2-second increments. I don't use the additive method as most people do with test strips, but rather each test increment gets its own full exposure time; this accounts for filament warmup effects. The test paper is then fully processed as per my standard processing. The shortest exposure time that shows a visible density change is what I use for that paper's default preflash time.