Hello smart peoples,
I'm an electromechanical engineer who likes to thinker a bit as a hobby now and then. I'm not the guy who comes asking for help easily, but right now, I feel like i'm overthinking, so here I am.

I would like to make my own LED camera flash circuit, a microcontroller-triggered one-shot.

what it should do
be ON briefly for about 1m - 2ms.
be as bright as it gets.
be as fast as it gets (response from mcu trigger to maximum brightness).
oh yeah, I would be using about 5-10 leds.
it should not become hot (which even at high currents, which such low duty-cylce, I guess this won't be a problem)
ultimately, it should be able to get 30hz or more.

what I've been thinking of
should I not overthink and let the MCU switch a transistor to turn on the LEDs in parallel from the battery + to ground? -> is it fast/bright?
should I use the same method as above, but with a MOSFET instead?
Can this be improved with a pre-charged ###µF capacitor to boost some current through the leds when the transistor/mosfet turns ON or am I talking crazy?
should I use a led driver (eg: RY3730) -> is it fast enough? If i understand correctly, these use a chargepump system, building up to the required voltage when the EN pin is HIGH = slower? There is a feedback on this circuit detecting open-loop, so i can't pre-enable this circuit to charge and enable the led string with a separate mosfet... or can i cheat this feedback?

I've attached the LEDs ( UV SMD 5050 type) and LED driver datasheets.

Thanks in advance!

 

If this is to be used with a camera, how will it be synchronized with the camera shutter?

I once built a circuit that triggered a Xenon flash unit by using a photodiode to sense the camera flash which triggered the flash unit and augmented the camera flash illumination.

 

If this is to be used with a camera, how will it be synchronized with the camera shutter?

I once built a circuit that triggered a Xenon flash unit by using a photodiode to sense the camera flash which triggered the flash unit and augmented the camera flash illumination.

Hello crutschow

About the same application I'm going for. The first photodiode of a chronograph would also be the trigger of the LED flash.
So no synchronization with a camera needed.

 

Any flash requires power in a short burst and so to keep the battery size reasonable there is needed a storage capacitor of some variety, Multiple LEDs would most easily be in series both to keep the current down and also to have all of them at the same power level. This also reduces the current that the switch must carry.

 

It's not critical whether you use a BJT or a MOSFET to trigger the LEDs, since the LED flash time is quite long (ms) compared to the circuit response time (µs).

There is a current limit for an LED beyond which it doesn't get much brighter, so there's no need to go past that.
That point may have to be experimentally determined.

 

Any flash requires power in a short burst and so to keep the battery size reasonable there is needed a storage capacitor of some variety, Multiple LEDs would most easily be in series both to keep the current down and also to have all of them at the same power level. This also reduces the current that the switch must carry.

Hi MisterBill2,

so to be clear, you suggest using the LED driver method in a default configuration (no pre-charge hacking)?

 

Any reason your LED choice is a purple LED? My read on the data sheet points to stuff like sterilization equipment. I would think you would want a bright white LED for a camera flash or strobe. Matters not as to a viable circuit. If you want white I believe Osram markets several bright white LEDs suitable for what you may want.

About the same application I'm going for. The first photodiode of a chronograph would also be the trigger of the LED flash.

Pictures pf things in flight, I like that. I really don't get triggering or the use of a micro-controller for projects like this.

Ron

 

Any reason your LED choice is a purple LED? My read on the data sheet points to stuff like sterilization equipment. I would think you would want a bright white LED for a camera flash or strobe. Matters not as to a viable circuit. If you want white I believe Osram markets several bright white LEDs suitable for what you may want.


Pictures pf things in flight, I like that. I really don't get triggering or the use of a micro-controller for projects like this.

Ron

Hi Ron

the complete end application is a tracer unit for airsoft. These things blast light on glow-in-the-dark BBs flying at 350fps or more. The UV spectrum is optimal for exciting this material. The mcu is to log and send the chronograph data, such as speed, fire rate, count, battery level to an external bluetooth device.

ah yes, important, battery powered! 3.7v lipo. meaning if anyone suggests 10 leds in series, i either have to precharge a capacitor with a chargepump to 20-30V, and dump that energy on the LED string when the projectile passes. or use the LED driver i was talking about, hoping it will be fast enough.

I simplified the question to the LED circuit only since i have no questions with all the other circuitry and programming related stuff.

 

OK, now we have more information. Really, though, even with just a single LED you want a capacitor because it will serve as the close by power source for that short burst. Firing from the capacitor avoids the issue of battery internal resistance, even if the cap is only charged to the battery voltage.
But now the application sounds like a repeating strobe-flash, not like a normal camera operation. At that rate heat can become a problem.
There are some incredibly bright LEDs available and so it will probably be good to contact some of the LED manufacturers to get their advice. They will be happy to explain how to make their products do their very best.

 

OK, now we have more information. Really, though, even with just a single LED you want a capacitor because it will serve as the close by power source for that short burst. Firing from the capacitor avoids the issue of battery internal resistance, even if the cap is only charged to the battery voltage.
But now the application sounds like a repeating strobe-flash, not like a normal camera operation. At that rate heat can become a problem.
There are some incredibly bright LEDs available and so it will probably be good to contact some of the LED manufacturers to get their advice. They will be happy to explain how to make their products do their very best.

I'll look up some strobe-flash circuits and contact a manufacturer or two thanks

 

That LED in the link in post #1 is certainly able to do what is asked, and with a forward current of 200mA will deliver a lot of UV light. That flash rate of 30 Hz is definitely possible, although probably not the intended application. Certainly some heat-sinking will be required.
The driver circuit may look a lot like the drivers for laser diodes, so there is another place to look.
The driver will need to both supply the drive current and then remove it quickly, so it will not be just a simple switch. For a photographic strobe the light pulses must terminate rapidly.
So the projectisnot being "over thought", rather it is a bit more complex than might appear at the initial glance.

 

so it will not be just a simple switch. For a photographic strobe the light pulses must terminate rapidly.

Can you explain this? Why would a MOSFET switch not turn off fast enough? 1usec should be fairly easy to achieve. For a pulse of 1 to 2 msec that is very sharp.

Also, there are small LiPo batteries that can deliver 30A, and he is talking about 1A. Don't see any need for a capacitor.

Bob

 

To very rapidly terminate a pulse it helps to intentionally remove any remaining charge, rather than letting it decay through the load. It is not simply "turn off " the device that is powering the load, but also removing any charge remaining. This is usually done with a "totem pole" arrangement, a second switching device connected across the load device to discharge whatever charge iis remaining. Most LED applications do not require it but the larger high brightness LEDs may have a larger capacitance and so a slower brightness decay with just switching off the power.
If the flash does not end "instantly" then there will be a bit of a tail on the image recorded. That may, or not, be a problem. That is why some photos of fast events show tails on the moving items.

 

A single Li-Po Cell will probably not provide enough Voltage-Overhead,
especially if You want to connect your LEDs in Series, which would be the recommended arrangement.

Here is an extremely fast Current-Regulator-Circuit that may give You some ideas.
It uses a "One-Shot" Circuit that can be re-triggered substantially faster than what You need,
and always maintains the exact Pulse-Width that You set it to.

An appropriate Input-Circuit will have to be substituted.
.
.
.

 

To very rapidly terminate a pulse it helps to intentionally remove any remaining charge, rather than letting it decay through the load. It is not simply "turn off " the device that is powering the load, but also removing any charge remaining. This is usually done with a "totem pole" arrangement, a second switching device connected across the load device to discharge whatever charge iis remaining. Most LED applications do not require it but the larger high brightness LEDs may have a larger capacitance and so a slower brightness decay with just switching off the power.
If the flash does not end "instantly" then there will be a bit of a tail on the image recorded. That may, or not, be a problem. That is why some photos of fast events show tails on the moving items.

The TS says he wants 1 to 2ms pulses. The turn off time is going to be miniscule compared to that. About tails: the TS quoted 350 FPS. At 1 ms, that is a 4 inch tail. Another couple of microseconds is not going to matter. I think he probably wants the tail effect. Think of the videos of tracer rounds.

I would guess that the micro itself, with no gate driver could turn it off fast enough so that it would not matter, but an external gate driver would certainly switch it fast enough. @MisterBill2, I think you ARE overthinking it.

Bob

 

Hi guys, thanks a lot for thinking along.

I'm mostly interested in a fast trigger-to-max-brightness circuit. the pulse time should be as long as the projectile is flying between the two gates of the chronograph. imagine that being about 50mm or so, which would equal ten 5050 size LEDs next to eachother.

projectiles travel at 250-500fps, which would translate to 75-150mm/ms

so even the slower projectiles would only be flying for less than a millisecond above the LEDs.

if the leds would be on for a bit longer, that wouldn't hurt the application too much, except for shortened battery life because of energy wasted.

everything being said here makes so much sense, i'm already so rich on information right now, for which i thank everyone of you!

 

So I've read up about led drivers and boost converters. The first seems to be an adaptation of the latter specifically for LED applications. So to keep it as simple as possible, I will go with a standard boost converter IC to charge a capacitor. The capacitor will then dump it's charge on the LED chain via MOSFET. Things to think about: size of the capacitor and, i guess, the frequency of the boost converter IC -> faster is better for high RPM burst fire.

I go with the LEDS in series method because most airsoft tracer units have a High Voltage warning sticker, indicating that they use this approach as well...

 

So I've read up about led drivers and boost converters. The first seems to be an adaptation of the latter specifically for LED applications. So to keep it as simple as possible, I will go with a standard boost converter IC to charge a capacitor. The capacitor will then dump it's charge on the LED chain via MOSFET. Things to think about: size of the capacitor and, i guess, the frequency of the boost converter IC -> faster is better for high RPM burst fire.

I go with the LEDS in series method because most airsoft tracer units have a High Voltage warning sticker, indicating that they use this approach as well...

Some LED drivers might include a buck/boost converter but they are otherwise unrelated. The important bit about LED drivers is constant current which is necessary because LEDs are non-linear with respect to voltage.

In order to avoid destroying the LEDs you will have to ensure the pulse length is very short, something on the order of μs. You can use a gate driver like this one to take a very small 5V pulse (like from an MCU) and use it to gate the MOSFET up to the supply voltage.

I'd read about someone experimenting with this but I can't seem to find it. I will keep looking, they might offer more information than my recollection does.

 

the complete end application is a tracer unit for airsoft. These things blast light on glow-in-the-dark BBs flying at 350fps or more.

I guess I'm pretty dense. Why would this even need to "flash"? The UV could be on when ever the trigger is pulled and the pellets are flying. Since the pellets "shine" when exposed to the UV, they should be the only thing seen. The only reason I can see to make it flash would be to emulate the commonly used every fifth round being a tracer. Doing that would be done by the UC to be figured out by the length of time the trigger is pulled.

 

I guess I'm pretty dense. Why would this even need to "flash"? The UV could be on when ever the trigger is pulled and the pellets are flying. Since the pellets "shine" when exposed to the UV, they should be the only thing seen. The only reason I can see to make it flash would be to emulate the commonly used every fifth round being a tracer. Doing that would be done by the UC to be figured out by the length of time the trigger is pulled.

Well, if you flash the LEDs you can get a much higher output from the LEDs for the short period. Also, UV LEDs emit some visible light so you would be giving away your position before firing.