High power LED nanosecond pulser

Thread Starter

materialsengineer

Joined May 9, 2019
4
Hi all,

I am new on this forum and on electronics in general. I do have some experience with development boards but not with circuit design at all.
I am currently working on a project for which I need bright short duration pulses of light. I know this is possible using high power LED's and I am wondering where to start.

To give you an impression of possibilities have a look at these products:
https://www.vela.io/vela-one-high-speed-flash
https://gsvitec.com/en/light/value-line/
These light are often used for high speed photography or machine vision. My application is similar.

I am looking at pulse widths of ideally 100ns but 500ns will also suffice for a first try. I found some hobbyists doing something similar as well.
https://petermobbs.wordpress.com/20...d-based-flash-gun-for-high-speed-photography/
These LEDs are only turned on for a short amount of time and can therefore be overdriven. This results in a high light output for a short duration.

I do not understand the circuit above so maybe someone can try to explain each part to me? The comments state it can be upscaled to higher power LEDs but since I do not understand the circuit (or the basic electronics) I do not know how. I would like to trigger the whole thing with a microcontroller (probably arduino nano or teensy).

Could someone please help me?

Thank you in advance.
 

bertus

Joined Apr 5, 2008
22,270
Hello,

The datasheet of the leds should show the maximum pulse current.
Is there a datasheet of the leds you want to use?

Bertus
 

LesJones

Joined Jan 8, 2017
4,174
From what I can see in the link the pulses are nothing like as short as one nanosecond. The shortest time mentioned is 1 uS (1000 nano seconds) but it also said he could not get less than 5 uS from the Arduino.

Les.
 

Thread Starter

materialsengineer

Joined May 9, 2019
4
Thank you Bertus and LesJones for your replies.

I did not yet choose the exact LED to use. But I am thinking of high power COB LEDs like the luminous CXM22.
I am not aiming for 1ns pulses but in the range of 100-500ns (however, shorter would be better).

At the moment I am mainly thinking of how to generate the pulses and how to power the LED. I can imagine a pulse generator to trigger a mosfet allowing flow from a capacitor bank to the LED but I have no idea if this is feasible and how this is done in the circuits I attached.

EDIT:
This is a circuit from a scientific paper on which one of the examples is based. I do not understand exactly what happens here.
 

Attachments

oz93666

Joined Sep 7, 2010
739
This should be very easy to build yourself

First a SSR (solid state relay) 3 pounds ...

This can pulse /switch up to 60V at a max 25A .... over a KW plenty .... It's activated by a low voltage input , easiest way to get a single pulse is charge a small capacitor to between 3 and 20 V , and switch it on the input , it will power the SSR and when exhausted will switch it off ...if too long a pulse , next time charge cap to a lower voltage.....

The expensive part will be the DC power supply ... what do you have ?? a car battery?? or a DC mains powered supply ... depending on the voltage you select the leds to match ....

Best leds to use are these ...

34V ...100W s for less than a pound !!! That's a great deal of light , no need for heat sink or lens ....

The whole thing could be done for less than 10 Euros (not including power supply )...... everything on eBay.
 

oz93666

Joined Sep 7, 2010
739
Hmmm. I think an SSR is going to be way too slow.

It's going to be a bit tricky to get anything less than 100ns
Hmmm ... you may be right ... I've got the feeling switching speed doesn't have to be too low ...The guy in the video above is pulsing by pressing a micro switch.
 

Thread Starter

materialsengineer

Joined May 9, 2019
4
Hmmm ... you may be right ... I've got the feeling switching speed doesn't have to be too low ...The guy in the video above is pulsing by pressing a micro switch.
I think the guy in the video is triggering the flash with the micro switch. But he is actually using his arduino to pulse the mosfet (+-5us). Please correct me if I am wrong.
 

pmd34

Joined Feb 22, 2014
527
I worked a bit with pulsing high power led's for looking at luminescence response from minerals. Typically you can pulse an LED to the order of micro seconds, I would be surprised if the semiconductors in them can actually respond much faster than this.

Also be aware that the yellow phosphor you see on the top of typical white LEDs (which converts the blue LED light to a more white spectrum) will have some significant luminescence response - taking microseconds to fade away.

For powering the LEDs, you are often giving much shorter pulses than mentioned in the data sheets so you can really up the current - factors of 10 would not be unreasonable.

You can pulse them directly by using a simple (typically N) Mosfet or using a charged capacitor bank, but you should always have some form of current limiting resistor in series with the LED to give you some maximum current limit.

Bare in mind that you are talking about serious optical power, so make sure you never look directly at the LED itself!

(If you want to see awesome continuous optical power, you can stick your LED in a cup of liquid nitrogen and crank up the current- that is really quite something!)
 

MrChips

Joined Oct 2, 2009
30,714
If you use a 74LS123 mono stable multivibrator you can achieve pulse widths as low as 60ns. However, you will be constrained by the switching speeds of the rest of the driver circuit.
 

Thread Starter

materialsengineer

Joined May 9, 2019
4
Thank you all for the responses. They are really helpful.

I do not necessarily need white LEDs. Would blue LEDs be better because of the absence of afterglow?
And I can directly drive a mosfet using a signal from a microcontroller or is it better to use the 74LS123 mono stable multivibrator?
And I imagine it would be possible to use a capacitor bank in combination with a mosfet so that I can charge the bank and time the flash.
Again, I do have some experience with electronics but it is not my main area.
 

MrChips

Joined Oct 2, 2009
30,714
You can drive the 74LS123 from the MCU.
I would use a fast NPN transistor to drive the LED.
The gate capacitance of a MOSFET is going to slow down your pulse transitions.
 

oz93666

Joined Sep 7, 2010
739
I rushed into this thread without much thought ... lets try again .....

Capacitor !! ... Of course .. this is a job for a capacitor , and a small battery !!

The specs on the vela machine linked show a luminous flux of 500,000 to 1.000.000 lumens

Normal output for leds is 100lm /W ... that means the pulse is 10KW ! seems quiet large

They say they are driving the leds at 20 times continuous power ... that means the nine leds in the unit are each 50W or perhaps 100W

they have pulse length of 500ns, 1μs, 2μs, 3μs, 4μs, 5μs .... ...... 1 μs = 1 millionth

A 10 Kw pulse for a μs uses only 0.01 J !!! so only a small capacitor is needed ...but a fast one

1/2 xCxVxV = lets have voltage at 40 ......... 0,01= 1/2 C x40 x40 ....so C = 12.5μF

if use a 100μF this will give longer or multiple flashes and allow for losses ...

Why not use 4700μF @ 40V to cover all possible requirements ??? still very small and less than $1...

You charge this with a very small battery and a step up converter... the power supply could fit in a match box....

The size of the whole unit is determined by the leds , I don't know why vela used 50W and not 100W ??? They are the same price ...

Nine x 100W should do it ($9)....

The rest is easy.
 

DickCappels

Joined Aug 21, 2008
10,152
I think that transistors in the avalanche mode would require a very long transmission line or delay network to achieve pulse widths of 100 ns or more. Using a Monostable such as recommended by MrChips in post #14 is probably the most practical method for your needs, provided that a fast enough combination of MOSFET driver and MOSFET are identified.

I would like to suggest, for background reading and amusement, this paper: cmosedu.com/jbaker/papers/1990/RSI81990.pdf Which describes a MOSFET driven by a transistor in avalanche mode.

One other detail, from my experience in getting over 300 watts out of a 2N3904 in avalanche mode I have come to believe that wildly exceeding the manufacturer's specified maximum currents in semiconductors is ok for the most part for very small amounts of time -chip heating is not that much, but I suspect that the main concern, that of ion migration, will probably occur after a lot of cycles. Having said that, I have run 2N3904 and other transistors at many times their current ratings for million or billions of cycles and only had failures when exceeding the maximum average power dissipation ratings. This is weird science.
 
Top