Ultra Short, Ultra Fast LED Flash

[JAndrew]

At a previous job, I did some high speed stop action photography. (See Attached Pictures)

The way this is accomplished is to use a dark room, with a long exposure on the camera, and to control the exposure with a short duration flash. The shorter the duration of flash the faster an object can move and still capture a clear image. Its important that the flash is bright, short and abrupt. We used Air Gap flash that generated a flash duration of 500ns. Hack-a-day explains how to build one but ours was a commercial one. (Uses high voltage caps, something I'm not real comfortable toying with)

I'd like to try and take photos like this on my own. I could go out and purchase the commercial unit for $1000+ or I figured I'd see if I could build it using RGB LED's. The reason I'm using RGB LED's is the phosphorus layer on a "White" LED Glows for a unspecified amount of time after the blue LED shines on it. This may extend the length of the flash beyond my target of 500ns.

The issues I'm having is trying to build a driver for the LED's. I plan on triggering it with a 5v high pulse driven from an ardiuno or some other simple electronic trigger. My thought was to use 25 of these 3 Watt LED for a total of 75 Watts:

https://www.superbrightleds.com/moreinfo/high-powered/vollong-3w-rgb-high-power-led/899/2214/

Any thoughts on how I might go about building a driver for this?

EDIT:

The direction that I think I am heading in, is to purchase some power MOSFET's. Probably one for each leg of the LED's. The larger power MOSFET's would be driven by a single smaller MOSFET. As far as switching goes it seems like MOSFETs are fast and can handle higher amps and voltages without being expensive. 

Thanks in advance,

Jay

[dmendesf]

You should take a look at this:

https://www.analog.com/en/analog-dialogue/articles/led-driver-for-high-power-machine-vision-flash.html

Enviado de meu moto g(6) plus usando o Tapatalk

[David Hess]

Many years ago I designed a low power LED flasher like that which easily generated pulses below 500 nanoseconds.  It used +5 volt logic to drive the emitter if an NPN transistor with the base tied to the +5 volt supply and operating as a high voltage cascode so transistor drive was ideal and the resistance between the emitter and gate output controlled the current very precisely.

Scaling this up would not be difficult with a power MOSFET on the low side and NPN RF power transistor as the cascode.  The advantage of running at higher voltages is that unavoidable parasitic inductance has less of an effect.  In practice the cascode is probably not required although it makes current regulation more precise.

If you operate with such short pulse widths, then the power can be much higher although this makes the power switching more difficult.  If the pulse width is only 10s of nanoseconds, then a charged transmission line and bipolar avalanche transistor might be worth doing; the voltage with that many LEDs in series is high enough.  But my guess is that you do not need to go below 200 nanoseconds.

[StillTrying]

There are a few threads on fast LED flashes on here.
https://www.eevblog.com/forum/beginners/4-microsecond-high-power-pulses-through-led/msg1414506/#msg1414506
https://www.eevblog.com/forum/beginners/microsecond-led-flash-ringing-issue/
https://www.eevblog.com/forum/chat/20w-halogen-bulb-viewed-by-a-photodiode/msg1751210/#msg1751210

If you go by the half light point the rise and fall times of medium-white LEDs is quite fast at about 100ns.

[Zero999]

I've done this before. The requirement was for a 1µs. Phosphor LEDs are variable. Some are fast and others are slow.

Attached is a schematic of the test set-up I used to drive LEDs with 1µs pulses of hundreds of amps, hence the low value current limiting resistor. It's possible to overdrive LEDs by a factor of 10, for short pulse lengths. Note transformers are used to drive the MOSFET and measure the current, to avoid large currents passing through the grounds of the test equipment.

[ogden]

Any constant current *switching* regulator will be too slow to produce speeds you need. Flash capacitor bank with current-limiting resistors and solid state switch (MOSFET or IGBT) is way to go. For design phase I would not use 5$ LEDs but cheapest I can find at given rating, initially use shorter string at lower voltage. You may find design shown in following post relevant. IMHO few tweaks needed and you are ready to go:

https://www.eevblog.com/forum/beginners/4-microsecond-high-power-pulses-through-led/msg1417075/#msg1417075

[Siwastaja]

Most LEDs have rise and fall times below 30 ns, usually below 20 ns. Similarly, fairly easy to find a power MOSFET that can switch in 20ns or even faster. Hence, you requirements are easy.

Constant current driver may pose a problem - the inductor slows down its response time. Some special design is required.

Although not perfectly optimal, the overall easiest solution, by far, is to use a bog standard regulated voltage supply and a plain old series resistor - the only downside is, to maintain some current regulation, you need to drop some meaningful voltage over the resistor, let's say at least 20-30% of the total supply voltage. So it's not as efficient as a constant current drive, but can be acceptably efficient. For even faster switching, you can use a small bypass capacitor parallel to the series resistor, but this shouldn't make a difference with your fairly slow pulses. Even without the small capacitor, you are already at full LED power after about 20-30ns, which is a small percentage of your 500ns pulse.

Do beware of stray inductance. Keep the physical loop as small as humanly possible. If you have a lot of LEDs, make the LED series layout a tight, thin "U shape" so that the opposing currents go as close to each other as possible, and Vdd and GND are right next to each other (preferably sandwitched on a multilayer PCB):

VDD-Res-LED-LED-LED-LED-|
GND-FET-LED-LED-LED-LED-|

Spray ceramic capacitors (snubbed with a larger electrolytic) between your Vdd and GND.

If your duty cycles are high, you'll have two opposing constraints, thermal design would benefit from loose layout, while you need tight layout for low inductance.

[StillTrying]

My thought was to use 25 of these 3 Watt LED for a total of 75 Watts:

You could try wiring up just one or two of the RGB LEDs with 1 to 3 Watts worth, and use its light in a darkened room to take a normal photo of the objects and look at the exposure time. It'll be a slow 1/5 sec-ish, but will give you some idea of how much more light is needed for a 1us exposure. You'll find for a 1us exposure you'll need to multiply the light/power by about 50,000.

Hence, your requirements are easy.

LOL.

[ogden]

to maintain some current regulation, you need to drop some meaningful voltage over the resistor, let's say at least 20-30% of the total supply voltage.

20-30% of waste power? Please don't... Somebody may take your joke as an advise. Example: Osram 24V strips uses 7 white LED's in series.

[JAndrew]

Thanks for all the replies. I've got a lot of reading and studying to do.

Duty cycle would be pretty low.  At most we would have 1 shot ever few minutes. I thought about over powering the LED's and wondered how well they would handle it for such a short duration and if I would need any sort of passive or active cooling. From the sounds of it, I don't think I will.


The machine vision is something I hadn't even considered. I know high speed flash is used for inspection on turbine and other high speed equipment. I should study a few of those circuits.

 

[Zero999]

to maintain some current regulation, you need to drop some meaningful voltage over the resistor, let's say at least 20-30% of the total supply voltage.

20-30% of waste power? Please don't... Somebody may take your joke as an advise. Example: Osram 24V strips uses 7 white LED's in series.

For short pulses, at a low duty cycle, the efficiency is often unimportant, as the total power is very low. If the LED is being over-driven the resistance of the bond wires starts to become significant at high currents and an additional series resistor is unnecessary. I've recently tested a white COB LED, the Cree® XLamp® CMA3090, with a nominal forward voltage of 45V and maximum current rating of 3.6A, with 1µs 105V pulses, no resistor, and the bond wires limited the current to 43A: well over the absolute maximum rating but it ran with a 100Hz repetition rate for an hour or so, with no damage or measurable reduction in the light output. Note that although the instantaneous power was 4.5kW, the duty cycle was only 0.01% so the total power was just 0.45W. The efficiency is probably much lower than non-pulsed operation but who cares if it's only on for just enough time to take a picture?
https://www.cree.com/led-components/media/documents/ds-CMA3090.pdf

I agree about not using a conventional SMPS, as it will be too slow to regulate.

If the delay isn't important or can be factored into the timing, one possibility is to charge up an inductor to the required current, then dump it through the LED. If the inductance and current are known variables, then the energy per pulse can be accurately controlled.

[ajb]

Many switching constant current LED drivers support PWM dimming by using a transistor to short the LED(s), so you could do the reverse here.  Establish your target current in the inductor, then open the switch across the LEDs for your desired duration, then close it again.  If the pulse is long enough and the response of the switching converter is fast enough the converter will pick up the load to maintain the LED current, otherwise you can just use a big enough inductor to sustain your required pulse.

[ogden]

I do not see inductor as better solution than simple resistor. I would just make 3-channel system with individually programmable voltages and timing for each color, obviously with (inrush) current limiting resistors to drop like 3..5% of voltage on each string. Not for god's sake 30%. It is not needed to increase voltage specs of whole system just because someone here in the forum said that ballast resistor shall drop 30% of the voltage which for high power LED's shall be considered as insanity.

[Zero999]

I agree 30% voltage drop is quite a lot.

The big advantage of using an inductor is the voltage can be boosted, so a lower voltage can be used to drive a long string of LEDs. It's also good if the LED has a much lower forward voltage than the power supply, so a power wasting resistor can be avoided. Note that in boost configuration there needs to be something to protect against open circuits, otherwise the high voltage spike will destroy the switching transistor.

I agree that it's better if the power supply voltage can be set to a suitable value and the current limited with a resistor.

As I said above: it's very important to keep the grounds for the logic and power separate: use an isolated MOSFET driver or pulse transformer for the transistor and another transformer or Hall effect sensor to measure the current.

I experimented with several different pulse and current transformers. Keep the burden, the secondary side resistor, as low as possible to avoid core saturation and improve accuracy. It's fine to overdrive current transformers, with low duty cycles, as long as the burden resistor is proportionally reduced to avoid core saturation. I've had quite a lot of success with winding my own on ferrite ring cores, but I found off the shelf ones work quite well. In the schematic I posted previously, I used the Murata 54100C with both primary windings in parallel and a 1R terminating resistor for a 10mV/A output. At 250A it worked perfectly fine and gave accurate measurements, when compared against a resistor.

https://docs-emea.rs-online.com/webdocs/0eb1/0900766b80eb1ad9.pdf

[Marco]

With RGB LEDs you'll need resistors for current limiting for white balance, even if they can take just dumping the full voltage in.

Personally I'd try 74LVC1G123 to generate the pulse, MCP14A0301 gate driver, 12 V power supply, random low voltage low rds MOSFET, resistors for current limiting and balancing, a lot of LEDS&resistors in parallel, feed the power&signal to the gate driver/mosfet/flasher section through two common mode chokes. If it generates too much noise you can always go the GDT route later.

[ogden]

a lot of LEDS&resistors in parallel

No, no, no. Better use (inductive switching) converter to pump energy into (flash) capacitor (48V or even 100V), to *avoid* lot of LEDs in parallel, but use single or few (long) strings. White balance using programmable voltage.

[Marco]

He only wants 500 ns pulse width. Avalanche transistors, massive overvoltage to overcome inductance, worrying about a little trace length are all overkill.

Meanwhile low voltage has the advantage it can't kill.

[Zero999]

One thing to note is the LED(s) should have a reverse parallel diode. I didn't include one in my schematic, because the LEDs I tested all have one built-in, which is very common for high current LEDs.

[Seekonk]

to maintain some current regulation, you need to drop some meaningful voltage over the resistor, let's say at least 20-30% of the total supply voltage.

20-30% of waste power? Please don't... Somebody may take your joke as an advise. Example: Osram 24V strips uses 7 white LED's in series.

Resistors, we don't need no stinking resistors.  I hit the LED right from the FET at about four times the rated voltage.

[schmitt trigger]

Jay

Doc Edgerton himself would be proud of this photo

[JAndrew]

Jay

Doc Edgerton himself would be proud of this photo

Thanks man, I’ve got a collection of similar photos, this is probably the one I am most proud of.


Sent from my iPhone using Tapatalk

[T3sl4co1l]

The biggest problem is getting enough peak power.  LEDs aren't good very far beyond ratings, unfortunately.  (The efficiency tanks, and then eventually you get accelerated damage, and outright failure.)

Tim

[Giaime]

Many switching constant current LED drivers support PWM dimming by using a transistor to short the LED(s), so you could do the reverse here.  Establish your target current in the inductor, then open the switch across the LEDs for your desired duration, then close it again.  If the pulse is long enough and the response of the switching converter is fast enough the converter will pick up the load to maintain the LED current, otherwise you can just use a big enough inductor to sustain your required pulse.

Been there, done that. This is the best way. Use Ti parts like LM3409 or TPS92641, TPS92515, etc, all support shunt PWM dimming. You can get 300-500ns pulses like that if you control the circuit inductance carefully.

[ogden]

Many switching constant current LED drivers support PWM dimming by using a transistor to short the LED(s), so you could do the reverse here.  Establish your target current in the inductor, then open the switch across the LEDs for your desired duration, then close it again.  If the pulse is long enough and the response of the switching converter is fast enough the converter will pick up the load to maintain the LED current, otherwise you can just use a big enough inductor to sustain your required pulse.

Been there, done that. This is the best way. Use Ti parts like LM3409 or TPS92641, TPS92515, etc, all support shunt PWM dimming. You can get 300-500ns pulses like that if you control the circuit inductance carefully.

Why, why do you people suggest such  ?

First one, LM3409. Datasheet states that buck regulator frequency above 1MHz is hard to achieve. This means that one buck regulator pulse alone is longer than required, not to mention that first pulse most likely will not reach nominal current. Following waveform from LM3409 DS clearly shows problem why buck regulator and inductor-based approach as such is not the best choice (to say it politely) for this application:

[StillTrying]

buck regulator and inductor-based approach as such is not the best choice (to say it politely) for this application:

Yes, I think a charged cap is the only way the energy can be stored for many minutes while waiting for the trigger, storing it in an inductor adds all kinds of timing problems.
But the biggest problem is still getting the LED light up to 10s of kW for very short exposures.

This ~5.7kW for 4us one mentioned above is about the best I've seen anywhere on the web.
https://www.eevblog.com/forum/beginners/4-microsecond-high-power-pulses-through-led/