rico wrote:
No PiPE for the D4, thank god. The D4 as spec'ed, and as tested by me, operates exactly as an IGBT design: faster discharge at lower energy settings, invariant color temperature at all energy settings, independent energy settings of .1 EV for each of the four outlets. Finally, the pack doesn't dump energy when settings are lowered: the capacitors retain the existing charge!
Good to know. Thought all profoto systems used that weird pipe thing
AFAIK, the Einstein doesn't dump charge either...that's why it gets more pops from a VML compared to a B1600...
Yesterday, I read an article about photo diodes. It says a regular photo diode will have a response time of less than 1 ns! 1 nano second response time is very fast, so it should be enough for my test on measuring the flash duration of different speedlights.
Here is the photo diode I bought yesterday. This is a small diode very similar to regular LED. In fact, most LEDs can be photo diodes, too!
Now, I used my oscilloscope to record the response curve from the photo diode. This oscilloscope has 1 giga/sec real time sampling rate with 2 mega points memory depth.
The whole set up looks like this. A small box is used to direct the light from a speedlight to the photo diode. (It can also block the ambient light.) The speedlight is one the left triggered by hand. The photo diode is on the right .
The test subjects are Canon 580EX & Metz 54MZ4-i speedlights. The later has 1/256 ratio which is not included since the 580EX only offer 1/128 minimum ratio. Each image has 2 rows with 4 curves, the upper left is the response curve of 580EX of the entire flash duration (well, almost). The upper right is the magnified view of the highest curve which also shows a sign of cut-off point (I believe). The 2nd row represents the same curves of Met 54MZ4-i. For anyone who is not familiar with oscilloscope, please pay attention to the voltage (vertical) & time (horizontal) scales below the curves. For instance, the upper left of the following image, the time scale is 250us (0.25ms) PER division, and the voltage scale is 200mv PER division!
At 1/128, the Canon has steeper curve on fall off. The Metz is less steeper than Canon with longer cut off time (70us vs 56us). Note that I moved the cursor to the point where the power is close to 10% of the entire flash duration. I assumed this is the point to measure the t.1 duration. So, the Canon has 860us = 1/860us = 1/0.00086 = 1/1162 sec t.1, and the Metz has 970us = 1/970us = 1/0.00097 = 1/1030 sec t.1.
1/1 Full. At full power, I am surprised by the flash duration on both units. Canon has 10ms burst duration with 2ms fall off. Metz, on the other hand, has double the burst duration - 20ms! Canon t.1 = 11.88ms = 1/0.01188 = 1/84 sec, Metz t.1 = 21.25ms = 1/0.02125 = 1/47 sec!
I don't know if this is the correct method to measure flash duration, but it does show some interesting results. The t.1 measurement on both units are much longer than I expected. I suspect the photo diode has some kind of ceiling on voltage level. It looks like the maximum voltage is 800mv. If this is true, then my t.1 measurement will be incorrect. Next time I will try my Elinchrom's. BTW, any comment are welcome.
First of all, you're my kind of maniac. There may be a logic analyser in my office suite, otherwise I should get one. The use of a photo-diode is clever, but you should confirm that maximum voltage ceiling from specs or, better, experimentally. I'm really surprised the gradual decay plots from both flash units - was expecting more of a chop.
As rico pointed out, your curves are more than a bit odd in shape. Your curves don't display the "interesting" part of the discharge, where most of the light is emitted, in a meaningful way, because in most of your tests, your detector seems to be in clipping until well after t.1
(t.1 is the point at which 90% of the energy has been discharged; t.5 is the 50% point. As a rule of thumb, you expect t.1 to be around 3x as long as the t.5 value. When flash manufacturers publish their specs, they usually give t.5 values, because that's what the ISO/DIN standards ask for, even though the t.1 time is probably a more meaningful number to use when comparing flash durations and camera shutter speeds.)
Your calculated times for the flash duration are off by an order of magnitude. Here's an example of what the discharge curve for a 580 at full power looks like with a detector/scope setup that's not clipping:
Surprisingly, someone has done the same thing! A photo diode is really a tool to measure light intensity. In my previous test, I measure the light intensity passively that is called photo-voltaic mode. A photo diode can also work on reverse-biased mode which is the method that Andy Gock has chosen. The reverse-biased mode is current based, so we need a resistor for i/v (current to voltage conversion). Under this mode, the accuracy and linearity are better, and most importantly, no ceiling problem. I followed his schema and chose a similar 64.5 ohm resistor for i/v. Here is my test set up again.
http://kspqm.kymco.com/roller/index/resource/johnny/ex1-2.jpg
At full power, t.1 is 6.120ms = 1/0.00612 = 1/163 sec. 6.12ms is slower than Andy Gock's measurement. (4ms) A possible explanation will be the capacitor inside 580EX. A speedlight uses electrolytic capacitor which has a tolerance of 20~40% in capacitance. If my copy has more capacitance, a full discharge will be longer. Another possible reason may be the life span of main capacitor. My 580EX is hardly used, no more than 500 flashes. As discharge-charge cycle builds up, the capacitance will decreased and results in shorter discharge period.
The following is the result by Andy Gock. They are very close to my measurement, especially at 1/4 & 1/2 ratio!
Canon 580EX
Power us s
1 4000.0 1/250
2 1088.0 1/919
4 484.0 1/2066
8 266.0 1/3759
16 166.0 1/6024
32 105.6 1/9470
64 71.6 1/13966
128 50.4 1/19841
Looks like your results are good this time.
At 1/2 power and lower, you can see clearly where the cutoff thyristor in the flash switches off the pulse. At full power, the curve has a long tail, which I guess is because the thyristor isn't used, and the capacitor is allowed to discharge almost completely.
I wonder what your 600RX's would look like under similar test conditions.
Thanks for the update: those plots are very satisfying! As expected, the discharge rate is identical for all settings. Except at full power, current is quenched by the thyristor, and in 50µs or so! I integrated your full-power curve: t0.1 delivers 96% of the total light, while t0.5 delivers just 2/3 total. The t0.1/t0.5 factor is roughly 3x.
http://kspqm.kymco.com/roller/index/resource/johnny/rx600-32.jpg
1/32. t.1 = 1/621s
As expected, there is no cut-off circuit on 600RX. This monolight is purely controlled by voltage. Each flash is a full discharge on capacitor with different voltage level.
http://kspqm.kymco.com/roller/index/resource/johnny/rx600-1.jpg
1/1. t.1 = 1/328s
3.5v peak (a bit on the low side, another ceiling?)
It's very strange that t.1 is getting slower and slower. It's 1/328s at full power! I guess t.5 is a better scale for non-IGBT or non-cut-off monolights, just like 600RX!
Finally, this is the measurement for t.5 at full power. It failed to comply with the manufacturer's claim - 1/2050s @ t.5.
I got 1/793s t.5 on my 600RX. Now, I wonder how I took the ice cube splashing shot by 1/8 ~ 1/16 power ratio! http://kspqm.kymco.com/roller/index/resource/johnny/rx600-1-t5.jpg
Interesting stuff. I think your measurements are good. Maybe the difference between the manufacturer's specs and your numbers is a question of how you define t.5 and t.1.
As I understand it, t.5 is the time that corresponds to the point at which 50% of the total flash power has been dissipated, and t.1 corresponds to the point at which 90% of the total power has been dissipated. Given these definitions, I can't see a quick and easy way to determine the voltage that corresponds to t.5 or t.1. Since the pulse has a somewhat complicated shape, wouldn't you have to make a series of measurements from each of your stored traces to figure out what times correspond to 50% or 90% of the area under each curve? Or, does your test gear have some sort of built-in integration function to make this easier?
Latest results are strange. If t.5 is time while illumination is 0.5 of the peak (Wikipedia), then the 600RX rates around 1ms for all energy settings. I also integrated the 1/2 and 1/1 curves, and full power yields just 40% more light than 1/2. I presume the diode isn't wobbling around in that snoot.
Interesting observation. I wonder what the dark voltage is for the photodiode setup. Phuang3 took precautions to shield the photodiode from the ambient light in the room, but maybe ambient illumination is still influencing the measurements.
I don't know what electronics are inside a 600RX, but I'm fascinated that all the curves are roughly the same shape. There's no sign of the pulse being quenched early at low power settings like you'd see on a thyristor controlled flash. They just look like the sort of curve you'd expect to see if you discharged a capacitor through a resistance and measured the voltage across the resistor over time.
I still like phuang3's 580EX numbers - they're not exactly the same as other people's results, but they're in the same ballpark. But, I think that the 600RX numbers are somewhat suspect.
PS:
I looked around for definitions for t.5 and t.1 and came up with some contradictory ones. The glossary of terms on the Paul Buff web site (manufacturer of Alien Bees, etc.) defines t.5 as time after which 50% of the flash power has been dissipated, and the wikipedia defines it as the point at which the light intensity has dropped 50% from its peak. The definition on the BPPA web site is different from both of them: "[t.5 is] a time measure taken from a flash between it reaching 50% of its peak value to the point when it has diminished to the same value." That seems better than the wikipedia definition, as it would give you meaningful numbers for flash bulbs, which have a slow rise time, as well as for electronic flashes. I don't know which of the three definitions is the "right one".
During the measurement, I turned the light off, so the ambient light shouldn't be a problem. 3.5v is probably not the ceiling because I still got 3.68v form 580EX. This photo diode has a very small die area, I think its angle to the flash tube is probably important. If I have time, I will try a different test set up.
aborr,
I also got confused with the definition of t.5 or t.1. However, I prefer the terms of Paul Buff's.
Here is another ice cube splashing by 580EX today. My 580EX is about 80cm behind the glass with a 2mm thick white acrylic in between. Power ratio is 1/16.
Agreed, action is nicely frozen. Looks like the ice cube is in my court. Just one light, correct? You seem to be using a softbox, and len FL somewhat above 200mm. Is that Cognac in the tumbler?
Yes, just one 580EX. I don't use softbox this time (In fact, I can't. Beware of the water spill!), it was a large acrylic panel for diffusion & background. I prefer a 60mm macro lens for this shot since I got a cropped body. The ice cube and "Cognac" are all faked. I made the "Cognac" with water and dye. The dye is safe, it's mainly used for drinks. I bought it from a food store. Just few drops, you can make litres of "Cognac".
Just wondering how would a Bowens 500 pro perform regarding freezing motion. Apparently it has the fastest flash duration of the Bowens line up of studio strobes. Appreciate comments.
pokemanyz wrote:
Phuang3 how are you timing the shot? I'm using a remote and just using trial and error, mostly error.
There is no secret. It just needs some practices before getting a good shot. I have to say this is a tedious task. For every try, I have to clean the the background, the glass and refill the liquid...and so on.
The D4 as spec'ed, and as tested by me, operates exactly as an IGBT design: faster discharge at lower energy settings, invariant color temperature at all energy settings, independent energy settings of .1 EV for each of the four outlets. Finally, the pack doesn't dump energy when settings are lowered: the capacitors retain the existing charge!