Flash Duration Measurement

I took a photo class where we used studio strobes. I was surprised when I set my shutter speed to 1/250s (the maximum sync speed — 4ms duration), when I looked at the image, part of the frame was black. With an on-camera flash, you never see this. The solution in class was to set the shutter speed to 1/160s (6.25 ms). I thought about why this was so and thought it had to do with the flash being of such long duration that the flash hadn’t delivered all of the light it was supposed to before the mechanical shutter started closing, causing part of the frame to be underexposed.  I decided to investigate the duration of the flash with different strobes. In order to do this I decided to measure the flash duration using a light sensor called a photodiode and a circuit I built which I’ll talk about at the end, if you are really curious.  Anyway, from searching the web it turns out that flash manufacturers typically quote a “t 0.5 time” which measures the flash from 50% of light delivered to 50% of light delivered.  Let’s look at one of my measurements:

This is a graph of the output of the photosensor that I built vs. time. It’s measured on something called a real-time oscilloscope, which plots voltage vs. time. The flash works by discharging a high capacitor charged to high voltage through a xenon tube. The shape of the graph should be familiar to most engineers.
I drew the 50% mark that most flash manufacturers quote. You can see that after 50% until the 10% mark there is a lot of light which hasn’t been delivered yet.  A more accurate measure of light duration is the t.1 measurement, which is when you measure from when the flash is 10% of the peak until it falls back down to 10% of the peak source. So, I measured the t.1 way, which people say is approximately 3x as long as the t.5 time. I set a trigger to capture the waveform, then I made sure the peak was not hitting the limit of the amplifier, and I called that the peak, then I measured from 10% peak to 10%peak using the cursors on the oscilloscope.

I have an older Dynalite pack I bought used. The Dynalite packs have kind of idiosyncratic notation, especially the old ones. It can either be in ratio mode, where each half of the pack is separate, or combine, where all the energy is available to both channels.

So, here are my results with a Dynalite 500XL strobe and 1 4040 head:

w/Pack set to Ratio and Variator set to zero:

62J   — 1/1355 s (claim from Dynalite: 1/4500s)

125J — 1/925 s

250J — 1/581 s

With pack set to combine and Variator set to zero:
312.5J — 1/464 s
375J — 1/460 s (why so little difference?)
500J — 1/342 s (claim from Dynalite: 1/1000s)

So you can see that increasing the light output increases the flash duration. Now something I have mentioned but not talked about is the variator which is a way of controlling the flash on a fraction of a stop basis.

Let’s look at when the we set the pack to 250J + 250J (combine) but the variator is set to -1 Stop
250J — 1/291 sec. Now, that is very close to our sync speed. If we have an wireless trigger for the flash, that probably adds some delay (maybe I’ll measure that next), and we could start to close the shutter when the flash is still on and delivering light.

In reality, in class we used 800J and 1000J Dynalite packs, which are slower than the 500J pack that I have. We were shooting with small f-stops, so the pack was almost at its full power setting. I am pretty sure that is what caused the black bars to show up on the camera.

I attached a report which has a graph of all measurements I took.

If you are shooting outdoors where you have to balance available light, or trying to capture a model in motion in the studio, try set the variator to the ‘0’ or no-trim setting. This will make the flash duration the quickest possible for the pack you have. If you want to estimate t.1 time for a flash pack, my measurements show that it is safe to use the rule of thumb that t.1 time is approximately 3x as long as t.5 time, as long as the variator is set to minimum at least with Dynalite flash.

How I built this thing:
Since a diode converts photons into electric current, I used a transimpedance amplifier to convert the current to a voltage. I used a TI 2380 Op-Amp which has a GBWP of 90 MHz. Originally I set the resistor to 10K but this provided too much gain, so I set the feedback resistor to 4.7K and increased the capacitance to 8pF. I carefully selected a photodiode which was sensitive in the visual range, but I think given the brightness of the light, I didn’t need such a specialized photodiode, and I probably could have used even an LED as the photodiode. The scope I used, an Agilent DSO9104A seems to be pretty noisy at 1V/div.

Great Moments in BP History

August 19, 1953 Operation Ajax

March 23, 2005 – Texas City refinery explosion

March 2, 2006 – Prudhoe Bay oil spill

April 20, 2010 – offshore oil well explosion