Bursting Balloons !

Bursting Balloons !
click on image for more

Friday, September 3, 2010

auto-fp flash

First of all what is "fp"-flash and why do we need it ?
FP stands for "focal plane"[ see the definition on the link ] .
At slower shutter speeds [ I'll refer to the limits of the D90 from now on ] up to 1/200th sec there is enough time for the shutter to open and for the flash to fire before the shutter closes without any problems .
But as you move into faster speeds the shutter starts using a slit to increase speeds and this means that the flash would only fire during the first stage of that slit and not show again on the image since the flash fires in between 1/700th and 1/41600th sec [ SB800 ] .
Here's an example , taken with the D90 and an old SB24 flash - since they don't speak the same language the camera fired the flash at any speed I chose .

At 1/1000th sec we can see when the flash fired while the rest of the image is black since that is the slit we see at the top . If you use 1/1000th sec the shutter can't move fast enough to do that speed .
How fast can the shutter move ?
We'll choose an easy figure to work with here so let's say the shutter curtain can only complete the trip across the frame in 1/250th sec[4ms] , so if you want 1/1000th sec it has to move a slit 1/4 of the size of the screen across the frame [ 1/250th/4 = 1/1000th ] .
As we see from the results the flash fired so quickly , 1/11000th sec actually , that it only lights up that 1/4 of the frame and then it's over.

So if we need to use a faster shutter speed , usually when shooting outdoors in bright sunlight and wanting to use a wider aperture , the flash needs to "stay on" for the whole time that slit is moving .
The manufacturers achieve this by firing the flash in a series of very fast pulses so the flash essentially becomes a 'continuous light' for the duration of the shutter travel .

The problem is that much of this light is wasted on the back of the shutter curtain and only a small amount reaches the sensor through that slit which is the major disadvantage of auto-fp flash .... it loses a lot of power . That's the trade-off you pay for using higher shutter speeds than normal flash can handle .
How much power ?
We'll work off the 'sunny 16' rule [ On the brightest sunniest day if your iso and shutter speed are equal then you will be at F16 for correct exposure ] .
I set my camera to F16 iso 200 1/200th sec and the flash head tells me I have 5.0 metres working distance . 
When I move into the higher speeds and fp mode the flash tells me I have 2.7 metres working distance at any setting that would give me an equivalent exposure for the ambient .
If F16 and 1/200th give us correct exposure then we would use perhaps F8 and 1/800th , F4 and 1/3200th etc .... For each increase in exposure via the aperture we decrease the exposure with the shutter speed so the ambient exposure is consistent .

Now since fp-flash behaves as a continuous light that should mean that in fp mode each equivalent exposure should give us the same results with the flash .
Actual tests prove different .

So why does the flash power drop off so rapidly and then climb again ?
It appears to be directly related to the concept of the focal plane shutter . Even though the shutter always moves at a constant speed , or more likely 'especially' since the shutter is 'limited' to a constant speed  , the slower speeds means it takes longer to for the shutter to complete the exposure .

Here's an image taken at 1/250th which is faster than the limit of the D90 . At low power and fast flash speeds there is no problem but if we tried to fire at full power the shutter would start closing before the flash was finished it's output and give us a dark patch at the top .
So the first curtain opens in about 1/250th sec[4ms] , then the second curtain closes and takes another 1/250th sec[4ms] to complete the exposure .... 8ms in total .
When we take a picture at 1/1000th sec the slit is 1/4 as wide and therefore only takes another 1 ms to complete its travel [ 1/1000th = 1ms ] . At 1/4000th the slit is so thin that it takes 4ms plus 0.25ms for the entire exposure , and here's a graph of the relationship between shutter speed and "total shutter travel time ":

Notice something interesting ? On both graphs it levels out at about 1/2000th sec . 
The only logical conclusion is that at slower shutter speeds auto-fp is limited to a lower power output to stay on for the longer time it takes for the second curtain of the shutter to complete its travel .

About 8.5 ms at 1/250th , compared to 5ms at 1/4000th [ it would have to come on slightly before and stay on until slightly after the shutter opens and closes ] This has recently been demonstrated by some tests on this blog . .
I am told that the graph will look quite different for a D700 and that the longest time for a D700 will be 4.5ms and the shortest time around 2.1ms . This would explain why similar tests done with a full frame camera show less power loss , the shutter moves much faster and the 'total shutter travel time' is shorter meaning that the flash can fire stronger for a shorter period of time .

So we learn from this that the nature of the focal plane shutter and how it works with fp flash means that the initial stages of fp mode , at the shutter speeds just past the flash's normal sync speed , give the lowest effective flash output .
Essentially it is the same amount of energy , only it is less power spread out over a longer time to accommodate the longer 'total shutter travel time' .
As we move into the faster shutter speeds we see an improvement due to the fact that the flash capacitor can be discharged over a shorter time allowing a higher output .
Think of it as a fast burning candle , at 1/250th we burn the candle which lasts 8ms . At 1/4000th we burn the same candle at both ends , it's twice as bright but lasts half as long .
With fx and a faster shutter travel we can almost start by burning the candle at both ends and advance to also burning it in the middle !
In practice this means that if you want to use a wider aperture don't think that staying closer to normal flash sync is going to put less strain on the system - it does the opposite so rather move as quickly as possible to the faster speeds when using fp mode .

You will have about half the working distance which is 2 stops less power using fp mode .
Initially though , you will lose around 3 stops of power at the slower fp speeds .



  1. Thanks for the feedback :) , it still needs a little refining , maybe some more descriptive graphs but I'm getting there .:)