# Macro the 1/f rule and flash duration

Started Sep 8, 2008 | Discussions thread
I'd be interested to know...

I'm not sure what method the flash makers use to officially characterize their flash pulse widths.

I explored this a bit in this post to that other thread:

In that post, I speculated about how this should really be done.

From the little bit of reading I did online, I found that (just as with a lot of other things), optical pulses are typically characterized using their full width at half maximum (FWHM).

And the same is generally considered to be "standard practice" for oscilloscope work in general.

But my problem with using the FWHM is that for some of the waveshapes that we get from these flash units, I feel that we'd be underestimating the "blur potential" if we thought of the FWHM for the pulse as being the "virtual shutter speed" of the pulse.

For the very rectangular pulse shapes, the FWHM would be very accurate for our purposes. But for the cases where there is a considerably longer "tail" to the pulse than there is time spent above the half max value, I feel that we'd get more blur in the image than we might expect.

So I invented another arbitrary method of characterizing the pulse widths. I decided (arbitrarily) that anything above 10% of the peak value would contribute to the exposure and the "softness" of the image in the case of a moving target or shaking camera/lens.

So in that post, I gave figures for my measurements both using my "10% rule" and the FWHM rule.

For pulses that are rectangular in shape, both figures are about the same. But for the pulses with long, drawn-out capacitor-discharge "tails" on them, the 10% rule gives us a much more pessimistic figure.

It's all quite arbitrary, I guess.

But the sad fact is that because the pulses are not rectangular, we do need to take all of that into consideration.

From the point of view of exposure, all of the area under the curve results in light delivered to the sensor. So I'm sure that Canon, in the design of the speedlites, employs some clever algorithms for the flash so that it can deliver a known power to the scene.

I know, for example, that the 580EX is able to account for varying states of charge in its flash capacitor by varying the flash pulse width so that when you pop off two or three flashes in quick succession, the exposure will be the same for all three shots even though the flash capacitor may not be fully recharged for the 2nd and 3rd shots.

So the exposure is held constant, but for our purposes, we can assume that we'll have more blur in the 2nd and 3rd shots of a fast sequence. So all of my testing is even more useless because all I'm testing is the pulse shape/width with the flash capacitor fully charged.

Too many variables

-- hide signature --

Jim H.

Complain
Post ()
Keyboard shortcuts: