Go Down

Topic: Camera strobe modification (Read 8729 times) previous topic - next topic

focalist

Jan 23, 2011, 05:29 pm Last Edit: Jan 23, 2011, 05:38 pm by focalist Reason: 1
Recently, I've been using the Arduino to do some high-speed camera strobe triggering to accomplish stop-motion photography.  Standard project done a thousand times- electret microphone, sound triggered.  Typically, when using an Airsoft gun to shoot Christmas Tree bulbs, I use a 75ms delay from sense of the sound to firing of the strobe.  The shutter is otherwise open.

The problem I'm running into is that the flash duration from my Interfit 150 studio head is way too long, even dialed back to minimum power.  I'm getting flash durations of around 1/2000sec, where I'd like to be dealing with flash durations more in the range of 1/10,000 or less, to freeze as much motion as possible.  As the flash duration IS the exposure time, 1/2000 sec is pretty slow when dealing with exploding glass.

So, I've done research and have thought of several solutions (short of glacialwanderer's spark-gap flash).

One is modification of a cheapo flash (like from a cheapo disposable camera).  Inherently the overall power is lower of course, but it's typically a shorter duration flash to get that.  As the trigger on these is high voltage, I'll probably use a 5v relay (they have 20msec switch time, which I can take off the strobe delay).  I don't think I've got a transistor or SCR handy in that HV range, but a PCB-mount mini-relay should suffice.

In reading, one method is to replace the flash storage capacitor with a smaller one, resulting in lower energy, which is shorter flash.  I may try this if i have a cap that's in the right range.  If I don't, I was thinking about being a real idiot:

In the flash, an oscillator is fed through a transformer to charge a HV cap.  Instead of reducing the cap, couldn't I instead place a bleeder resistor across the cap, and feed the transformer PWM?  In this way I could control the charge level of the storage capacitor by PWM.  Sounds good on paper until I start working the WATTAGE of that bleeder, I'd need a very high wattage (high voltage at high current storage) resistor, wouldn't I?

Thoughts?

Last but not least, also looking at trying to use a couple of power led's I've got around.. led's react in the nanosecond range...  I know the actual power is low compared to xenon- orders of magnitude, in fact- but proximity to the subject (inches) may compensate for that- nothing to lose in trying....
When the testing is complete there will be... cake.

Udo Klein

If you reduce the charge level (at constant capacity) you will reduce the voltage. Thus your flash might not trigger anymore.

I think this won't work. What might work (but I never tried it) could be to interrupt the current to the flash tube. No clue if this is easy. As I said I never tried it.

Udo
Check out my experiments http://blog.blinkenlight.net

focalist

#2
Jan 23, 2011, 07:27 pm Last Edit: Jan 23, 2011, 07:36 pm by focalist Reason: 1
Well, I've been looking at Sam Wasserman's strobe pages and links frome there, one of which details the modification of a kodak max disposable camera flash, by replacement of the storage cap- from 160uf to a 5uf (chosen because the author had it in junk box), resulting in a power reduction (duration reduction) to 1/36th the original.  He goes on to make a repeating strobe, but for the purposes I need, simply the reduction in flash duration is all that is needed.  For most photography, you'd like a longer flash duration.  For this, the power isn't as important as the speed.

Excellent discussion of abuse of strobes in general:http://repairfaq.cis.upenn.edu/sam/strbfaq.htm#strbnsc

Assuming we are talking typical cheapo, that's a 1ms strobe.. which already is faster (lower power) than the studio flash.  If the result is real- 36 times faster - that's a VERY usable interval.

For stop-motion, you don't need a ton of light- but you need that light to come on and off VERY fast.

What you've suggested is called a quenching circuit, and it's what is used in high-end strobes costing thousands... the damping circuit to interrupt the arc takes HUGE energy dissapation, methinks..
When the testing is complete there will be... cake.

mmcp42

I would think high power LEDs is the way to go
you have to watch colour temp as they are very monochromatic
(can you be slightly monochromatic?)

the big plus is you shoud be able to fire them rapidly and often with very little circuitry
there are only 10 types of people
them that understands binary
and them that doesn't

focalist

#4
Jan 23, 2011, 07:44 pm Last Edit: Jan 23, 2011, 08:11 pm by focalist Reason: 1
Well, I now have a reason to rip apart something I've been using for less nefarious purposes... Home depot sells a 10w CREE LED light- warm white, 2700K, CRI in the high 80's.  I got one and have been using it as a great little portable modeling light.  Coupled with a pair of Interfit 150's and some creative reflector work, it makes for a cheap (under $350) and very usable portable portrait "studio" setup.  I recently posted some shots taken with the CREE LED light.

However, there is a significant delay between applying power and the led actually striking.. enough that it's noticable.  I'm sure that's a function of the driver circuit.

I *should* be able to place some type of control circuit in the works, and be able to use it... yet another hack to try ;)

Here's a shot from that series.  No strobes were used for this shot, it was "Rembrandt" lit 45 degrees off subject on both sides with 75w of halogen at about three feet through silk softboxes for fill.. but the main direct lighting is that 10w CREE.  It's quite visible in the shadow tones and uplighting.. the halogens are at her eye level.  If anything, it's a bit too deep in rendering of the reds...  I am not displeased with the rendering, it works really well if used in concert with more standard light sources.

When the testing is complete there will be... cake.

mmcp42

a) once you know the delay you can compensate for it
b) if it's always the same delay it almost doesn't matter
c) try the old trick of not turning them fully off
d) implement c) using variable PWM
there are only 10 types of people
them that understands binary
and them that doesn't

wyager

#6
Jan 23, 2011, 08:58 pm Last Edit: Jan 23, 2011, 09:01 pm by wyager Reason: 1
Three of the ideas so far I like:
1.Smaller capacitor-this is probably the simplest and cheapest option.
2.LED-this is also a good idea, the problem is that LEDs are (unless you want to start spending thousands of dollars for a pre-built option) hundreds of lumens, while most camera flashes are many thousands of lumens (I think). To get more than a few thousand lumens out of an LED source, especially when timing is critical, the only practical option is to make your own device, which will probably end up being rather complex and expensive. You would need a very high sensitivity camera and a very dark environment. CRI is not so much of an issue for most photography IMO, you can always artificially re-color stuff in photoshop.
3.Control discharge from the capacitor. If you can get a proper transistor or MOSFET or something (maybe a cheap MPSA42/MPSA92 will work?), you can put that in series with the flash bulb and have 100% control over the amount of time the bulb gets energy from the cap. Just something like
Code: [Select]

digitalWrite(flashPin, HIGH);
delayMicroseconds(flashTime);
digitalWrite(flashPin, LOW);

u0421793

Smallening the capacitors will not necessarily of necessity result in a shorter flash duration. My studio flash system at the moment consists of a few Elinchrom D-Lites (4s and a 2) and somehow a BX250Ri (which I don't use wireless as I don't know the specs for the Skyport wireless trigger system, otherwise I'd build my own transmitter for it) and the BX is theoretically quite a lot more powerful than the D-Lites, yet far shorter flash duration.

If I were doing the "freezing motion" sort of flash trickery, I'd probably try my Metz gear at it to get a very small duration. Picking up a few 45CL1s these days can be quite cheap, even if only for this sort of a project.

focalist

#8
Jan 23, 2011, 11:53 pm Last Edit: Jan 24, 2011, 12:36 am by focalist Reason: 1
Well, been browsing over at the CREE site as well as several other makers.. CREE just released a document regarding pulse-driven power LED's at multiples of their rated values.. basically a document from their R&D department showing some basic overdriving info.  Interesting, and they take great pains in noting that overdriving will void warranty and probably shorten life of the emitter.  That being said, the document gives reference to pulse-running short durations at up to four times the rated current, giving up to three times the rated light output.  Heat is the monster, and the possible melting of the wires that go to the die.  Pulsed very quickly, like in the microsecond range, electromigration may be a bigger problem than heat.  The article also refers to overvoltages into the tens of volts, but overcurrent is probably a better option.  Maybe I'll just hook up two or three "D" alkalines with no current limiting, those pups can throw a couple of amps for short periods.  I have a 1w Luxeon that is begging for abuse..  I mean, let's say I can get 3x the rated output, but it will shorten the LED life to say, 10k pulses at that output.  Considering a new power led is only a buck or two, even a few hundred "flashes" would be plenty cost-effective to burn an LED out with, considering..

As for intensity, remember light decreases as the inverse square of the distance, or something to that effect.  I can get an LED awfully close, within inches, just out of frame... huge power isn't actually huge at all, if it's further away...

As i understand, there's potentially tens of amperes at hundreds of volts blowing through that xenon tube- can a transistor of reasonable price handle somthing like the spike that it would need to interrupt?

I'd love to have access to some of that kind of high-end studio equipment.. but I needed to save months to even get the Interfit el-cheapo setup.  $299 for two 150 w/s strobes with 100w halogen modeling lights, stands, soft box, bounce umbrella, and cables.  One of the cheapest setups on the market, but rated high for semi-pro usage.  If you are in the market for the cheapest possible (but stil usable) light rig, I recommend it heartily.  Slow to cycle, but the setup provides more than enough for decent head shots and small areas.  heads are adjustable from full to 1/8 power.  I use them with strobes for standard portrait shots, and with one additional modeling lamp (like the CREE) without the strobes for impact shots like the one above by using a good old Nifty Fifty.... f1.8 50mm, slow and noisy but try and beat the image without spending a grand on "L" glass.  At a hundred bucks new, if you don't have the "L" toys, you really MUST have a Nifty, IMO.   With that short DOF, you can really get tight on subject focus with a really nice softening.  Huge aperture = luv

The heads you are talking about start in the range of twice the price of my whole light kit- for a single head with no accessories...  modifying a $3 disposable camera flash seems a better idea than selling my soul (and still maybe needing to modify) for a pro monolight.....  As it stands, my next purchase needs to be a new body, my poor little 350D (Rebel XT) has seen a lot of milage, and Canon won't even fix them anymore under standard rates.  Dreaming of a pro body, but even one of the newer XT-series "Pro-sumer" models would be a big leap in gear for me.  The gig that shot above was taken for was defunded.. and that job was supposed to sock away most of the cash for the new body.. so I'll be milking life out of that 350D for a while yet.

:)
When the testing is complete there will be... cake.

Ran Talbott

Instead of modifying the flash, why not get one designed to work with TTL metering,  and have the Arduino pretend to be the camera?

My second choice would be to find a flash with built-in metering,  and hack into the circuit it uses to cut off the flash when it decides there's been "enough" light.

Go Up