Help with Flashing an LED

Status
Not open for further replies.

Thread Starter

JohnBee

Joined Jan 14, 2015
13
Hello
I want to flash an led as bright as possible. I believe a mosfet triggered to dump a capacitor onto the led
would be a logical circuit. My main question is: What kind of cap is best? I'd like to have very high capacity,
so a "super cap" or electrolytic comes to mind. I'd like it to deliver 10 amps at 5 volts, for a 1/2 millisecond.
Yeah, I know this is very high.
Any specific opinions on the cap? Recharge time not an issue.
 

MikeML

Joined Oct 2, 2009
5,444
What is going to limit the current through the LED?

Post a data sheet (or a link thereto) for the LED you are trying to flash.
 

Thread Starter

JohnBee

Joined Jan 14, 2015
13
What is going to limit the current through the LED?

Post a data sheet (or a link thereto) for the LED you are trying to flash.
thanks. This is a Cree white led with current 1000 ma at 3.5 V.
Size (mm x mm) 3.45 x 3.45
Maximum drive current (A) 1.5
Maximum power (W) 5
Light output Up to 456 lm @ 5 W, 85°C
Typical forward voltage @ 350 mA, 85°C (V) 2.85
Viewing angle (degrees) 115
Binning 85°C, ANSI
Thermal resistance (°C/W) 5




Note: I want to drive this at 10-50X its rated power for a very short time.

JB
 

MikeML

Joined Oct 2, 2009
5,444
Somewhere on the data sheet is a spec for pulsed operation. Since you didn't post the data sheet, what is the part number of the LED?

The problem with discharging a cap into the LED is that the current pulse shape is wrong. It would be better to pulse the LED with a rectangular pulse where the peak current is maintained longer.
 

Thread Starter

JohnBee

Joined Jan 14, 2015
13
Somewhere on the data sheet is a spec for pulsed operation. Since you didn't post the data sheet, what is the part number of the LED?

The problem with discharging a cap into the LED is that the current pulse shape is wrong. It would be better to pulse the LED with a rectangular pulse where the peak current is maintained longer.
OK - it's a XTE or XT-E white led. I am new to this forum, so haven't figured out how to attach a file yet.
Thanks John
 

Thread Starter

JohnBee

Joined Jan 14, 2015
13
Somewhere on the data sheet is a spec for pulsed operation. Since you didn't post the data sheet, what is the part number of the LED?

The problem with discharging a cap into the LED is that the current pulse shape is wrong. It would be better to pulse the LED with a rectangular pulse where the peak current is maintained longer.
Hi Mike
Here's a link to the xt-e:
http://www.cree.com/search-results?&lcid=9&q=xte&tz=America/New_York

You have to click on data sheet link, which is a pdf.
John
 

Thread Starter

JohnBee

Joined Jan 14, 2015
13
Bad news: I read the data sheet linked to from here. The maximum allowed current for this LED is 1500mA; no ifs, ands, or buts. Even pulsing it at 1500mA requires some serious heatsinking.
Hi Mike,
I probably wasn't clear - I want to just do single flashes like a photo-flash, not repetitive pulse operation. That kind of operation is "off the charts." The limit to single
flash current is not so much heat load as intrinsic mass action of electrons and electric fields operating in a small geometry. That's how a cell phone flash works - many times the
"normal" current of the led but for a millisecond- or so. The limit is mechanical: When does it blow apart?
Given that, how would you calculate the Joules needed for the led to conduct, let's say 10 amps for a short time at say 5 volts drop? I am a bit rusty, but I get energy in watts is just
IxE. In this imaginary case that's 50 watts. Joules are just watts x time or 50/1000 in this case - ,05 Joules. Now (here's where I am rusty) - the energy in a cap is 1/2CV^2. So if we charge the cap to 20 volts
we get algebra 200*C=.05 so C = .00025 Farad or 250 microfarad. Because of the cave man nature of this calce, just get a cap that is as big as you can - say 2500 uF at 50Volts.

Wear Goggles. :)
JB
 

Bernard

Joined Aug 7, 2008
5,784
As a start, I flashed multiple times a white piranah LED rated at 35 mA, at 900 mA, 500 us and it keeps on going. Used 2 ea 555's & n-ch FET.
 

Thread Starter

JohnBee

Joined Jan 14, 2015
13
As a start, I flashed multiple times a white piranah LED rated at 35 mA, at 900 mA, 500 us and it keeps on going. Used 2 ea 555's & n-ch FET.
Bernard,
Can you give details about voltage and how you measured instantaneous current?
Thanks
JB
 

MrChips

Joined Oct 2, 2009
30,714
I would not use a capacitor but a PWM MOSFET.
Use DC analysis to calculate the current and assume the PWM signal and resulting pulse is rectangular.
One way to determine your max current is to increase the current until the LED destructs.
Of course this is destructive testing.
Then set your safe operating limit to 50% of the destructive current assuming you keep the PWM frequency and pulse width the same.
 

Thread Starter

JohnBee

Joined Jan 14, 2015
13
What form of a trigger is proposed?
Might start with cheaper LED at 10 X rated current and work up to Cree.
The trigger I have thought of is just an mosfet driven with a pulse of voltage high enough to turn it fully on.
I work with picaxes, and they only work at 5 volts, so a way is required to get that higher, i believe.
Perhaps another mosfet that controls the gate voltage to the power mosfet is a good idea.
Bernard, what kind of FET did you use?
Obviously the led diode will turn fully on at voltages much above its bandgap. A laggard mosfet gate will just dissipate the energy
inside the mosfet, imo.
jb
 

Bernard

Joined Aug 7, 2008
5,784
Got brave, on R7 dropped from 10 ohms- 5.6- & 3 ohms., from 100 us to 500 us. Changed C2 from .o1 to .001 to keep trigger shorter than output. LED 2 is to monitor state of U1- on ready to pulse. Switches are self debounced.Pulse#3 00000.jpg
 

Bernard

Joined Aug 7, 2008
5,784
I was wondering why you were responding to an earlier post, then the little light came on & I hit " Post Reply" & there it was from yester day. Just another senior moment, just had another birthday aniversary, now 88 years old.
'Used the 3055V, not the best but was on hand. Could replace U1 with NPN transistor or LL N ch FET inverter fed into C2.
LED1, medium low power white # MVL-914 UW, 40 mA, 140 mW, Vf 3- 3.7, 5.2, 50 deg., 4 pin DIP, Super-flux, Piranha.
Rough calculation of I, ( 12 - 5 )/ ( 3 + .15 ) = 2.2 A ,@ 100 us to 500 us. Not able to view on scope but flash verry visable.
 

Thread Starter

JohnBee

Joined Jan 14, 2015
13
Hi Bernard
88? That's very good. Take any life extension vitamins? I have a list of some.

I will study your circuit. Think a Li type battery directly switched would work?

-the kid (70 yo)
 

Bernard

Joined Aug 7, 2008
5,784
Maybe, if internal battery R is less than an ohm & V is high enough to allow for some series resistance. Would be nice to know Vf of LED @ 10 A; have a memory scope ?
Not many vitamins, just fresh grapefruit juice & a shot of vodka.
 

Thread Starter

JohnBee

Joined Jan 14, 2015
13
Ha Ha - Bernard, I'm back, hope you are still there.
Say, what about a capacitor charged up to well over the band gap voltage of the led?
Then use a mosfet to switch that in?
Is that essentially what you did with that circuit?
In that regard, why both C4 and C5? C4 is 1 mike - Is this there for filtering?
I DO have a memory scope, but I forgot how to use it. :) Actually, it's a tds694 tektronix, and I've never used it - too intimidated by it.
JB
 

MrChips

Joined Oct 2, 2009
30,714
What! You have a modern $6000 3GHz digital oscilloscope and don't know how to use it?
Send it to me and I will make a video for you on how to use it.
 
Status
Not open for further replies.
Top