I have been toying with the idea of simulating/replacing a xenon strobe unit with a high power LEDs. Looking at DC strobe xenon drivers, they have a PWM chip with a current sense & voltage feedback that pulses a MOSFET which in turn pumps a step up transformer, a single diode then will feed a charge cap of say 220/330uF 400VDC and a thyristor will discharge it in to the xenon tube. A voltage feedback in to the PWM chip stops it pumping once the cap is fully charged hence reducing current consumption to only a few mA whilst charged.

I am thinking of modifying the above design to feed a 3A LED. Instead of charging to a 300V I will charge to say 14V (4 LEDs in series), voltage feedback stops the charge once at this voltage and reduces current consumption, the use a MOSFET to pulse the charge in to the LEDs.

Using this method, I am trying to reduce overall inrush and current consumption as the large charge cap can be charged and maintained and the strobes should take some of the charge which then will be automatically topped up by the PWM feedback. Does all this makes sense or am I just going off on one...?

 

hi sean,
If I understand your description, it seems an over elaborate way to pulse a LED chain.
E

 

Hi E,
As I said, maybe I am off on a wrong track completely, but If the goal is to reduce high current pulses and turn them in to a slow charging current as it were, wouldnt this idea be better than just pulse a high power LED with a PSU that needs to deliver say 3A on every pulse? I am only guessing the charge cap would have enough current in it for the strobe, as when shorting the cap it produces a big spark but thats when it is charged to 300V.

Any suggestion how else to do this?
S

 

hi S.
What if you charged a large cap via a resistor from a 14Vdc source.?
Chose the charging resistor to suit the pulse rate, once charged the charge current will be zero.

Trigger a NMOSFET at the required pulse rate and duration, add a current limiting resistor in the LED series chain so that you do not over drive the LED's

E

 

Read the LED data sheet carefully, as long as you don't exceed the non-repetitive peak pulse current, you should be ok.

It's all about managing the energy, if you want to make it efficient, avoid resistive capacitor charging.

 

hi Sensa,
I 'think' the TS has a 14Vdc power supply that cannot supply 3Amps pulse current.

Perhaps the TS could tell us the specification of his 14V PSU.??

E

 

E, yes psu is limited, as I mentioned the idea is to use lowest inrush and rms current to run this 'strobe' at high currents off the cap, flash rate needs to be up to 90 per minute... its all blue skying at the moment. PSU can do 1A or just over.

Sen, thanks, the LEDs that i am playing around with are 1A & 3A, generally you can hit them pretty hard 10:1 ratios.

 

OK Sean,
What pulse On period are you looking for, at say 0.66Sec rate.?

Corrected rate.

 

I thought of using LEDs in place of a xenon tube many years ago but when I looked at the peak current rating of white LEDs it was only a few times the maximum continuous current rating so the power of the flash would not be very high. The first photographic flash I has as a teenager was rated at 100 joules ( A joule is a watt second.) The flash duration was something like one millisecond. So the power of the flash would have been 100000 watts. So if you consider a white LED which needs about 3.3 volts then the current you would need to pass through it to get 100000 watts would be about 30000 amps.

Les.

 

My max pulse On period is 18mS for any flash rate. I am looking through LED datasheet to find out how hard I can hit it to get the max intensity for the period i.e. to make it look like a xenon flash.

 

hi sean,
This LTSpice shows a basic scheme,

The 4R is the LED path
E

 

Hi Les, I am looking to produce around 10-15 joule only. 1 LED has a 1A continuous rating and I was thinking I could hit it it at much higher rate maybe 10 times, with constant current control...

 

Thanks for the sim E, I suspect the brightness will be majorly affected by the very narrow /spiky current at the top.

 

hi sean,
This LTSpice shows a basic scheme,

The 4R is the LED path
E

Just an update on this, I knocked up a very simple prototype a PIC pulsing 20mS triggers to a N-MOSFET not the one you suggested as havent been able to source it from Farnell/RS (I know Mouser/Digikey have them). The one I used is IRF640 which has much larger on resistance ~ 0.15 Ohm, too large as the drop in charge cap C1 voltage is only about 3 volts.

 

Most LEDS do not take high current pulses. You do need to check maximum repetitive peak current ratings.
The problem is that once past their optimum working point, LED efficiencies drop off and series resistance increases the voltage drop. So hitting them with 10:1 current for a 1A continuous LED is more likely to fuse it. LEDs can work as flash light replacements for photographic use but need to be"on" for longer than a normal flash tube. Old flash tubes tend to flash for about 1ms. To get enough light from LEDs you need the full power (maybe 100W equivalent tungsten, perhaps 10W of LEDs) and just turn them on/off for 1ms or shorter if you need shorter times, but remember the total light output will be rather dim. While you can make strobes out of single LEDs you may consider wiring a bank in series-parallel (with current sharing resistors) to reach the power necessary. As a photographic light source you can pulse the LEDs on for the duration of the shutter- maybe 1/60s or so.

 

I was thinking I could hit it it at much higher rate maybe 10 times, with constant current control..

The problem is that the LED light is not proportional to current a high levels. The light output saturates, and further current is just wasted as heat.

 

I thought of using LEDs in place of a xenon tube many years ago but when I looked at the peak current rating of white LEDs it was only a few times the maximum continuous current rating so the power of the flash would not be very high. The first photographic flash I has as a teenager was rated at 100 joules ( A joule is a watt second.) The flash duration was something like one millisecond. So the power of the flash would have been 100000 watts. So if you consider a white LED which needs about 3.3 volts then the current you would need to pass through it to get 100000 watts would be about 30000 amps.

Les.

Isn't that an apples and oranges comparison? The light output relative to energy input is pretty different between LEDs and Xenon bulbs, right? So, you'd need to compare light output, not energy input.

Don't get me wrong, I suspect the end result is still that it's unrealistic unless you use large banks of LEDs.

I had to bring it up though - the math sounds fishy!

 

I agree that I have not considered the relative efficiencies of LEDs and xenon flash tubes. (I have not yet found comparison data.) The fact that the TS wants a much longer flash
(18 mS) than is typical for a xenon tube flash (1mS) means for the same flash energy the LED flash only needs to be 1/18 th of the power.

Les.

 

The fact that the TS wants a much longer flash
(18 mS) than is typical for a xenon tube flash (1mS) means for the same flash energy the LED flash only needs to be 1/18 th of the power.

Interesting observation!

 

I am open to experimenting with whatever LED within reason.Currently playing with CREE 1A white, next would be CREE 3A white and I am also ordering some fleeBay COB 10/20/30W LEDs.

The mobile phone market seem to use dedicated (sync) chips and super capacitors presumably to dump large current for a very short time into the LEDs, so I have ordered some of these 1F 5v5 and will start with 1 LED and increase the number and super cap in series to cover the voltage drop. Also some very low on resistance MOSFETs to do the switching.

The 18mS is the max pulse with and a ball park, I would be happy to use smaller periods but as we all know the light output will suffer as the on period is reduced. So my options are open for LED type, driver type and configuration. I am am sure I will be frying a few good LEDs in the line of development but I guess as they say you cant make an omelet without breaking some eggs!