How to strobe a Lumiled with 100ns?

[Fixxxer]

Hi everyone,

I need to develop a driver to strobe a lumiled within 100ns. The LED needs to lit up for 100ns or less. During these 100ns the LED need produce it's max. power.

This application needs to be triggered with a 5V puls. Can some of you help me with this? Are there transistors/MOSFET's/IGBT's/... capable for this?

Thx

 

[Bunk3r]

you could cut the power to the LED with a FET on a resister driven setup, if using a driver then drivers often have a PWM pin that you can use for that (or failing that the shutdown pin). (as if you turn the LED off the driver would smoke as drivers need the load connected, if you turn the whole driver circuit off it may (or maynot) be slow on start up for each pulse).

 

[snarfer]

100 ns is pretty fast. You might get into some difficulties with propagation delays depending on how many components are involved.

There are certainly FETs available that will switch fast enough. Just try a parametric search on mouser.com or digikey.com or whatever your vendor of choice.

Really you would have to look at your system as a whole to figure this out though. What is the trigger input? Is there a microprocessor involved? Does the on-time have to be synchronized to some external event? Is this a one-time flash, or a series of flashes at a certain rate? And so forth...

 

[2xTrinity]

I looked at the output of my Liteflux LF2 keychain light on an oscilloscope using a photodiode, and on the lowest setting, the "on" portion of the cycle is only 250ns. That's in a fairly cheap PWM circuit able to fit inside a production AAA light. You should be able to find something to handle that switching time, especcially if space/cost isn't a constraint.

 

[frenzee]

Do you need just one pulse or continuous pulses? If the latter, then it seems to me you will need two circuits to accomplish this. First, you'll need a timer such as an LM555 to fire off a pulse of X duration and you can vary the off cycle to get the duty cycle you need. Then you will need to feed its output to a constant current driver with a very fast rise time. Offhand I don't which drivers would or wouldn't fit this criteria, but you can usually find this parameter in the spec sheets. If you are going to drive the LED at its upper limit, I really don't think you can get away with just a resistor/MOSFET combination to control the current AND get reliable performance out of it.

Another option I can think of is to use a charge pump circuit which is typically used in a camera flash circuits. Linear, National and others have dozens of these.

I suggest first running some simulations using a spice software such as SwCad and taking a look at the output curve. That will pretty much tell you what your rise times are going to be and what components you need to get the current and output characteristics you need. You will need to experiment of course.

The 5V trigger pulse is not a problem. Almost any timer can accomodate this and if not, you can always drop the voltage with a resistor bridge. I can't think of a timer that requries more than 5V to trigger.

 

[Fixxxer]

This is what I have right now. Just a simple circuit with a buffer, a filter and a transistor. I've got a 250ns strobe, but with low intensity. The intensity needs to be increased a lot. Can someone explain to me why the intensity is so low?

If I'm going to use a FET/IGBT/... would this improve the pulstime and the intensity?

OR

Do I need a different circuit.

I need it to be pulsed continuously. The DATA IN is the trigger input.

I'm not so into the whole electronic stuff, but I really need this.

Cheerz & thx

 

[frenzee]

You have a 22Ω resistor in line with the 5V supply. Wouldn't that give you a mximum of 200mA or so? Also, I don't understant your circuit. Isn't the LED supposed to be in series with the transistor's output, instead of parallel? Can you try it in a series configuration and replace the transistor with a power MOSFET or IGBT?

 

[Fixxxer]

When I set the LED in a serie config with the transistor, there are 2 things:
- The LED will be "OFF" for 250ns instead of "ON"
- When I remove the NOR/NOT-gate, & the LED is in serie with the transistor, the LED slowly fades out after the transistor closes and this needs to happen immediately.

In the config above the LED will be lit when the transistor is "open". When the transistor is "closed" the current will go to the GND without passing the LED.

I need a high current for a intense flash. Because this happens for a short period the LED won't be damaged. I need to have a lot of light during the flash of 100ns. Now I have a flash with low intensity during 250ns.

As you said, there will be a current around 200mA (= 5/22), but you need to bring the forward voltage of the LED (http://www.ortodoxism.ro/datasheets/toshiba/2589.pdf) in calculation. So the current will be around the 100mA. This is twice the max of the LED but still the Flash is weak.

Why is this? Because the transistor (http://www.ortodoxism.ro/datasheets/fairchild/BC548B.pdf) is to slow? Which power MOSFET/IGBT/... should I use or which specs do I need to pay attention to? Rise- & Fall-time? current???

I really don't know what the problem is and I even don't know if I can make it work with this circuit.

Thx for helping me out

 

[frenzee]

I see. Well, while that might work, if it were me, I would still put the LED in series and just change the phase by putting another inverter (like a 4069) back to back between the RC circuit and the transistor. I would also change from a bipolar to an N-Channel MOSFET on the output. If you happen to have a dead UPS laying around, those things usually have a half dozen or so power MOSFETs which you might be able to salvage, but look up the specs of course.

At 100ns you are talking about 10MHz operation, so I would carefully check the specs for all the components to make sure they can operate at this range. I would also hook up a scope, if you have one, to see the signal pattern at various stages.

As an alternative, I would also consider using a charge pump driver like this one.

 

[IMSabbel]

Hi everyone,

I need to develop a driver to strobe a lumiled within 100ns. The LED needs to lit up for 100ns or less. During these 100ns the LED need produce it's max. power.

This application needs to be triggered with a 5V puls. Can some of you help me with this? Are there transistors/MOSFET's/IGBT's/... capable for this?

Thx

Are you talking about white or colored here?
For whites in that timescales the speed of the phosphour might not be neglectable....
Thats doesnt matter for PWM (as the pulse will just be shifted and re-shaped, but might be a problem if its timing sensitive)

 

[snarfer]

The part number in the schematic is for an orange LED.

I'm not sure I fully understand your circuit. But if I were you I would first try removing the NOR/NOT gate and replacing the transistor with a high speed FET in line with the LED.

There are many that can switch on and off in under 10 ns. It will have to be a logic level FET with extremely low gate capacitance.

Just as an example, FDY301NZ might work. Well probably you will have to spend some time going through data sheets to find the right one. Maybe a radio frequency FET would work as well, if RDS_ON is low enough. Presumably you could also use a FET with high RDS_ON and remove the resistor as well. Or you could even use some fancy zero voltage switching FET from ALD.

Anyway frenzee is absolutely correct in suggesting use of a scope.

 

[Fixxxer]

Hey guys,

I'm want to use a MOSFET for this application, but I've got a couple of questions.

My thoughts are that I better can replace the 74HCT244, R3, C1, R4, the NOT-gate and R5 with a multivibrator and the transistor with a MOSFET and putting the led inline with the MOSFET.

The multivibrator needs to do the following:
The input will be a trigger signal of 5V. On the positive edge the multivibrator needs to generate a 100ns puls. The 100ns puls needs to switch the MOSFET to lit up the lumiled (LXHLLW3C). Therefore the 100ns puls probably needs to be 10V (or are there MOSFET's with a Vgs of 5V and a pulsed Id of 2A)

Where can I find a multivibrator capable of doing this and a suitable MOSFET? What are the most important specs? (pulsed Id = 2A, Vgs = 5V, ...)

Best regards,

Fixxxer

 

[snarfer]

This is kind of a strange idea, but the simplest solution of all might be to use a mosfet driver to run the LED. This would take your pulse sequence as input and drive the LED directly, instead of driving an external mosfet. Example of driver that might work would TC4421a. But of course there are many alternatives, almost every IC manufacturer makes one.

The mosfet driver is actually a pair of mosfets in a "totem pole" configuration that both source and sink current extremely quickly.

 

[paulr]

If you meam a white led I don't know if it's possible to light it up that fast, because of the phosphor persistence. You could do it with a monochromatic led. Maybe you really want a xenon strobe?

 

[SemiMan]

This is kind of a strange idea, but the simplest solution of all might be to use a mosfet driver to run the LED. This would take your pulse sequence as input and drive the LED directly, instead of driving an external mosfet. Example of driver that might work would TC4421a. But of course there are many alternatives, almost every IC manufacturer makes one.

The mosfet driver is actually a pair of mosfets in a "totem pole" configuration that both source and sink current extremely quickly.

By far the easiest way at least for the driver portion. Good idea snarfer.

 

[spencer]

Really off topic, but Fixxxer, what program did you use to make the schematic for the circuit in the above picture. Anybody else can chime in and suggest a program that does similar, preferably free.

 

[HKJ]

Really off topic, but Fixxxer, what program did you use to make the schematic for the circuit in the above picture. Anybody else can chime in and suggest a program that does similar, preferably free.

I have been using Eagle for some years, first the freeware version and later the paid versions. The freeware version is fine for small projects.

http://www.cadsoft.de/freeware.htm