LED photostrobe Alpha test

Just a video showing the Arduino switching 24 watts of LED's via Darlington pairs made of pn2222 and TIP41c.

The final project will have just over 100 watts of white LED's, as well as 20 each RGB. This current test is with 24 watts of warm white LED's, somewhere around 2000 lumens total light output, about one and a half times the light output of a 100w incandescent bulb, or about the same as a 100w quartz halogen bulb.

The flashes in this test are 1, 5, and 10msec in length (delay function), though the end project I want to be able to drive with pulses in the tens of microseconds. These units actually have current limiting onboard, 500ma. The remaining LED's will be driven from LM317 current regulators, I am considering running the LED's over their basic rating, say 150%, due to the short pulse lengths. At least initially, I plan to stay somewhat within spec.

Looks Good... Drive um up there! ... 150, 200, 300%... LED's are cheap, and getting cheaper all the time.
As long as the pulses give time between for thermal dissipation, you can probably pull it off. All the LED's I have burned out, took at least a second to self destruct, but that was continuous current +200%.

Added my ten watt cool white (12v @ 800mA), bringing it up to 2.8A @ 12v, a bit over 33 watts, yielding something like 3000 lumens. Going to set camera up later, and test it out to see how it does as macro flash, illuminating at a close distance. I'll use the arduino to run different pulse lengths to get sufficient exposure at close range. Reports will be made.

Nice and bright :slight_smile:
But why Darlingtons instead of MOSFETs? AFAIK, Darlingtons have higher loss/get warmer.

I'm planning on picking up some logic level MOSFET's as soon as I see a batch going for cheap. I've seen a few N-Channel models, logic level input, with max voltages around 60v and a current capacity of 10A.. those would be the right answer here I think.

Right or wrong, cleaned up the wiring, and up to 34 watts now..

EDIT>>Went up on ebay and got a batch of 20 N-Channel Logic Level MOSFETs for seven bucks shipped. 11A, 60v maximum. Time to get my feet in the waters of MOSFET... these will switch a LOT faster, and since it's Darlington, I'm never reaching full saturation on the output stage, right? Also a lot more headroom power-wise. Seems like I've been spooked by the scary looking symbol too long..


Reducing the pulse length from 10ms to 10us means 1/1000 of the light output all other things staying the same. Or with other words: 100W for 10ms is 1J while 100W for 10us is only 1mJ. If you want to get 1J during 10us this translates to 100kW. This is the reason why high speed photography is usually not approached with LEDs.

You might want to consider some focusing lenses to compensate for that. Or consider xenon flashes.

The core emitter has arrived... 3500 lumens of 3000K, a fifty watt warm white LED array. Eighteen bucks.

Guys, understand.. I know the physics I am facing.. But I am also depending upon the inverse square of the distance from the source. I am not lighting a room.. At best a field of a few square feet... And I already have the Fresnel lens for focusing this behemoth. This is an exercise in "let's see if I can make this work somehow".... And I do have a few tricks in reserve, including having harvested a dozen xenon flash units from disposable cameras. They are perfect for some things when triggered through a simple optoisolator. It is important to keep in mind that I am not trying to make something that's a normal use flash.. It'd be cool as hell for milkdrop photography, or a feeding hummingbird for example.. Very close, macro even. The power decreases by the SQUARE of the distance, so I believe that I have a fighting chance at getting at least some limited success with over 100 watts of LED's. Note that stage professional PAR cans are now 100w LED arrays already...that's replacing carbon arc in some cases.


That is a professinal stage flood, LED... Five hundred bucks. I am cautiously hopeful seeing these products.

I really am caught up with the 'If I can get this working at any level, it would be a very cool tool' thing. If it doesn't pan out as a strobe, it's still a solid state floodlight that is a consistent color every time.. That alone is worth the cost of the parts.. The total on this now.. Hmm.. Like thirty bucks.

It will probably be several days until I can hook it up, in any case.. Gotta now look at the best way to feed that LED. It'll be PWM'd with MOSFETS, but I will be taking a little time looking for a good current regulator setup for it... Though I understand there is a variation on LM317 thing with a feedback loop, that easily can do.

It's got to limit at 2.5-3amps at 16-18v, anyone got a particulary cheap and easy solution (I love LM317 regulators.. and I have a bunch...but current sources couldn't run tandem due to tbe feedback circuit, right? If they can, easy enough to tandem the little beasts.. Three would be plenty, an amp each...)

How about an LM350 for 3A? Still needs a fair bit of heat sinking. Or an LM338. Still needs heat sinking. If you want to get rid of that, you have to go with a switcher, which ends up requiring potentially bulky inductors, but perhaps that's not a problem.

I'm not sure if the current rating applies when you are wiring it as a current source rather than voltage regulator... The current rating is load max when operating as a voltage regulator. In the spec sheet for LM317, a circuit for a 1A current source is given, but I have no idea what kind of load is on the regulator when operating as a current source. It could even be that in that configuration it can carry more current, even..

If they can be used in parallel (Kirchoffs apply here?), then I'd just run three and call it good... I believe it is possible to run them side by side as voltage regulators, but in current regulation 'mode' I'll either need to research (or just try it).

I think the current rating is the same. As far as I understand it, a "current regulator" really just is a voltage regulator with feedback, which makes it very similar to an adjustable voltage regulator.

Hmm.. I was uncertain about that feedback loop, as there would of course be current from the other regulators. If there is a diode or transistor in the output stage preventing current from coming from the other regulators, it will work I would think. I even have some diodes that can handle the power level, though I still wonder if they would be unnecessary or just added drain.

Hmm.. Gonna need a test load too.... Maybe an incandescent bulb, 12v.. Close enough for government work... I may have a 55 watt headlight bulb spare to use as a nice load....

TheMOSFETs arrived in the mail today, in all honesty now that I have read about them, I am excited to give them a try. No more Darlingtons for me on heavy loads... so much better at carrying current.. And so much cleaner and faster switching.. Guess it was time for me to move out of the 1940s... makes it clear how far my formal electronics training went... basic theory, and I wasn't paying attention anyway. On the other hand, I am extremely proficient in home firefighting now :grin:

Okay, so I have looked over the specs on typical adjustable regulators, and it still looks like a reasonable way to provide current limiting. Since I have LM317T in hand, I want to make three of the current regulator circuits from the spec sheet (1.2 ohm between output and adjust makes for 1a regulator).

In thinking about this, and getting ready to wire up a test circuit, I was looking at the schematic in spec sheet for running the LM317 in tandem as a VOLTAGE regulator.. and it shows the setting resistor "shared" between the adjust connections on the regulators.. This makes sense to me, I think, as the resistor output voltage remains the same, even if the current is spread across the multiple terminals.. And as it only cares about that voltage, "sharing" the resistor ensures a single feedback loop to the regulators, keeping them balanced.

Given this is the case, does the same design consideration apply if I were connecting multiple LM317 current sources? I am thinking that a single voltage split three ways will be better than three sorta close voltages...

Darlingtons for me on heavy loads... so much better at carrying current.. And so much cleaner and faster switching.. Guess it was time for me to move out of the 1940s... makes it clear how far my formal electronics training went... basic theory, and I wasn't paying attention anyway.

This makes me wonder if you are driving the mosfets properly. Do you account for the gate capacitance?

How so? I thought darlingtons are inherently slower, in fact I was concerned if they were getting fully ON before going off again. I thought that MOSFETS are inherently better until you are into MHz switching rates,
Isn't the gate charge dissipated when the pin is taken low? I haven't even added a gate to source high value resistor yet, which I understand is typically how the gate charge is bled off.

Even without any additional components, no gate resistor, it seems to work great.. Though the gate resistor is a better idea so as to protect the AVR from the current draw of charging the gate. I have read conflicting rcommendations regarding the use of gate resistance, and even use of the gate to source resistance.

Even if there is a charging delay, that only shortens the length of the emitted pulse from the LED, I would only have to compensate by adding that time interval to the turnoff delay. If I have read properly, these should be able to switch in the tens of nanoseconds in their normal use.

Am I missing something?

I think that quote is out of context. The words right before that quote were "no more Darlingtons ..." and the "so much better at carrying current" sentence references MOSFETs, not Darlingtons.

@focalist: Was that one of those old LanCity cable modems in that first video (were you using as a heatsink?)...

Heheh yes it is. I never throw something like that out.. it's just one Ginormous heat sink, solid aluminum, probably a good two pounds.. with those nice vanes on the back and even a hollow interior (where I am hiding much of the sloppy wires. I see a heat sink like that, I grab it.

Remember my big old rack mount Telco power supply? It was time to sacrifice her, she's just too big and unweildly. But why would I do that, you ask? Simple. Inside of the twenty five pound steel monster is an array of Volgen industrial power supplies, each one with it's own heat sinking and case. I need 16-18v @ 3A for the fifty watter.. and one of the PS modules is a 24v@5A model. With the voltage adjust turned all the way down, I get 17.8v - Perfect! With the addition of a 2n3055 based current limiter set up for 3A- we are good to go. One question, if anyone has used them- there's an enable on the PS, open circuit the PS is on, closed circuit the PS is off. I am just planning to use a MOSFET like the others, but I am wondering if there's any advantage to using that enable circuit (prolly not).

After drilling a couple mounting holes, I soldered some leads on the 50watter and slathered it with compound, and screwed the beast right in the middle of that LanCity cable modem case.. and switched it on. I was foolish enough to actually be looking nearly directly at the LED from a foot away when I turned it on. I'll say this.. 3500 lumens of warm white LED is BRIGHT. Like afterimages for a good five to ten minutes bright. With that huge heat sink, I am able to place a finger right on the emitter and it's only comfortably warm. The resistors in the current limiter were far hotter. I actually have to return to that limiter circuit, because I substituted 240 ohms in the circuit for 220, and when metered out, it shows that the current actually is limiting around 2.25A while the emitter wants 2.5-3A. Not sure how much more light I can get for another 500ma, but I plan to find out. Since output isn't linear to input, I figure I am maybe losing a hundred lumens or so without that extra quarter to half amp.

So, I am putting together the MOSFET for this one this morning.. should be ready to start UberBlinken later today.

The current capacity of the system:

8x 3w Warm White, 200 lumens each, for a total of 1600 lumens
1x 10w Cool White, 1000 lumens
1x 50w Warm White, 3500 lumens

This gives us a total of 6100 lumens or thereabout, for a power load of 84 watts plus a little overhead for the regulators. With this mess hooked up, I am going to try to take a photo or two with it later today, just to see how it turns out. Overall, getting 20 lumens per watt isn't very efficient.. but I am not really looking to make this efficient. I want it BRIGHT, and able to be switched in the range of tens of microseconds. As I am trying to emulate a Xenon strobe in a way, I want my pulses to be at least as short.. Xenon tubes typically fire in the range of a millisecond duration.

I still have six more of the three watters, plus another thirty 1w discrete LED's, just about another 50 watts.. but I am going to hold off adding more until I play with it as built for a little bit. If I can get a usable image with the power I have at the moment, I have possible other nefarious uses for that batch. That being said, I originally set out for an arbitrary 10K lumens, and that mess would bring the total close..

However, I am now not powering up the system until it's facing the wall-- it's painful to look at, downright blinding...

So, I took the whole rig (much to my wife's glee) downstairs and just did a simple strobe program, 2ms pulses followed by a 10ms delay. This is only using the 50 watter. The results are pretty typical stroboscope shots.

50mm f1.8 Canon Prime lens - light source 6" from subject
ISO 1600
2 second exposures @ f/22!!

This works out to f/1.8 exposure of 1/125 second. That's with a dark subject. This, simply put, is going to be at least as good at close range as a xenon strobe, but FAR more controllable- there's no recharge time, I can set arbitrary intervals for both the pulse length as well as the interval between pulses. I still will be adding the 3 watters and the ten, almost doubling the light.

I was just pushing the shutter release then dropping a lighter on the table. There was a delay in dropping the lighter.
For code, I just modified the delay times in Blink.

I had far too much light to use the lens anything but stepped WAY down.

What this means is that I am going to succeed. Now, it's a matter of seeing how short I can get the pulses and still get enough light. The 50mm goes all the way out to a Beercan f/1.8, if I want it, I can borrow a friends f/1.4 but it looks like I may not even need it.

Got some things to do this evening, then I will hook up the other MOSFET and add in the three watters, and another for the 10watt. The ten watt is mounted on it's own heatsink, so I can use it as a synchronous strobe from another angle to give fuller illumination.

Macro is absolutely possible. We're going to see what it does on a full backdrop later on...

Sounds like you have fun! I love it when projects work out this well :slight_smile:

Yeah, I'm quite happy already, this is really looking like once complete, it may be able to do what I want.
If a project's not fun at any given moment, I'll shelve it and twiddle with another. I have several Arduino
projects going right now... ranging from CNC to Lighting experiments. I never stop tinkering.

Anyway, A cleaned up version, I may still get a chance tonite to rig up some more shots.

This was shot under tough conditions (f/22, ISO1600) intentionally to give a worst case view. Going to get MUCH better...