How does the power of a flash relate to the duration of the flash pulse?

Specifically for my Nikon SB-700, but also for flashes in general — how does the power setting relate to the duration of the pulse? Basically, is the power normally controlled by cutting down on the duration of the flash, or by using a more/less intense pulse?

I've not been able to find the duration of the flash pulse at each power setting anywhere. Also, I realize that the pulse isn't a perfect square, so is "duration" normally measured as the time from the beginning of any output to the end of any output, or (for example) just the shortest time window that contains 90% of the output?

Answer

This is one of the key differences between "speed light" hotshoe flashes and traditional studio lighting. With studio lighting, flash power is determined by the amount of capacitor charge, and the duration increases (slightly) for lower-powered flashes. For hotshoe flashes, the capacitor is charged to full but the burst of light stopped when the proper amount is delivered (traditionally, by an electronic component called a thyristor).

One thing that makes the Nikon flash system awesome is that Nikon makes by far the best manuals for their flashes. Only Metz comes close. You can find a chart of duration at each power setting on page H17 of the SB-700 manual:

Flash duration (approx.)

   1/1042 sec. at M1/1 (full) output

   1/1136 sec. at M1/2 output
   1/2857 sec. at M1/4 output
   1/5714 sec. at M1/8 output
  1/10000 sec. at M1/16 output
  1/18182 sec. at M1/32 output
  1/25000 sec. at M1/64 output
  1/40000 sec. at M1/128 output

As you note, this a curve, and you're exactly right that the normal means of specifying the duration do not measure the entire "long tail" of the curve. There's two values usually used: T.5 and T.1. The first, T.5, is the time that the flash pulse brightness is above 50% of the peak, and T.1 is the time that the pulse is above 10%. That's a little different from saying 90% of the output — which would maybe be more meaningful for photographers, but would be more calculation to determine. With these numbers, you just look at the peak and then figure out where the cut-off is on either side of the curve, and measure between those two points.

Generally, flash makers give T.5 rather than T.1, since that's a) slightly easier to measure and b) gives results which sound more impressive. For studio lights, the T.1 value is more useful and better reflects the real effect on exposure, but for speed lights, the abrupt cut-off means that at anything but full power, T.1 and T.5 are roughly the same.

Because these numbers are so low even at full power, I suspect that they are T.5 values (and I don't see Nikon saying either way, which is generally means T.5), but since Nikon flashes are so nice, it's possible that they are T.1.