Pulsed LED driver. / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Pulsed LED driver.

Move Thread

LAN_403

iJim

Fri Apr 16 2010, 03:53PM

Registered Member #2662 Joined: Fri Jan 29 2010, 10:14AM
Location:
Posts: 36

Hi All,

What are the limitations of driving an LED in pulse mode? For an experiment I need 100ns bursts of light...

LED in question:
Power Dissipation 180 mW
Reverse Voltage 6.0 V
DC Forward Current 30 mA
Forward Voltage @ 20 mA 5.5 V 7.5 V
Pulsed Current (1 ms pulse with 1% duty cycle) 200 mA

Building the circuit is not a problem but the LED cost Â£100+ and I don't want to break it!!

The datasheet is thin on details, just the figures above and some optical info. If I plot the max current at DC and the max current at 1ms can I simply extrapolate back to find the max current at 100ns? I guess as long as the power dissipation remains below 180mW the LED should be fine. Also what are the chances of my 100ns current pulse actually producing a 100ns burst of light?

Maybe the wrong forum for these questions but worth a try...

Cheers,

iJim.

hboy007

Fri Apr 16 2010, 05:18PM

Registered Member #1667 Joined: Sat Aug 30 2008, 09:57PM
Location:
Posts: 374

What type of LED is that? You can look for a cheap diode or LED replacement to test the driver. As with electrostatic discharges, the limits are set by maximum local current densities (each material has a different damage threshold and it gets even worse with boundary effects and non-uniformities in the geometry). The circuit parasitics will introduce a pulse forming network that can wreck your emitter, too. the pulse current rating you mention may be derived from the thermal capacity of the semiconductor die itself whereas the repetition rate relates to heat transfer properties. You can run non-destructive tests on the device by measuring the center wavelength shift during different operation modes to monitor the die temperature.

I will have to solve the same driver issues with a 240mW AlGaInP laser diode soon (I need 250..500ns pulses).

Conundrum

Fri Apr 16 2010, 06:18PM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

hmm..
If they are cooled waaaaaay down using a Peltier stack and low melting point alloy casting , those "superflux" LEDs can be overdriven to absurd power levels.

according to the manufacturer's site some can safely be pulsed at 1.4* their rated current on a short duty cycle if cooled to room temp. if cooled to -25C and kept there the "destruct" current can be maybe 5-6* normal.
the actual failure mode is typically "cratering" of the ball bond and can happen with severe overcurrent in as little as a microsecond.

i did some work on this at 'uni and LED capacitance can be worked around in the same way as MOSFET gate capacitance, by putting a resistor of 5% rated current across the emitter.
this "damps down" the ringing so that the diode then behaves like a diode rather than a capacitor at high frequencies.

Please PLEASE test this with an "expendable" emitter not your Â£100 LED... luxeons work well for this as the die has large area so simulates a worst case load.

From the description this seems to be a UV emitter which is superficially similar to a Bluray laser diode. these are very sensitive to overvoltage and overcurrent but this can be mitigated by proper design of the driver circuit with Zener diodes and crowbar circuits.

Also from my own experiences adding a series resistor is always wise, as sometimes the diode can survive overvoltage ONLY as long as the current remains low.
Reverse voltage does however cause rapid failure as I discovered.

-A

klugesmith

Fri Apr 16 2010, 09:56PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

iJim wrote ...
If I plot the max current at DC and the max current at 1ms can I simply extrapolate back to find the max current at 100ns? I guess as long as the power dissipation remains below 180mW the LED should be fine. Also what are the chances of my 100ns current pulse actually producing a 100ns burst of light?

What would be the relative spacing of DC, 1 ms, and 100 ns on the X-axis for extrapolation?

Average power limits must always be respected, but additional rules are needed for single-pulse and low-rep-rate applications. A parameter that matters to bond wires, etc. is I2T integral -- the product of pulse duration and the square of the current. If that's the limiting factor, max current for 100ns is only 10x the max current for 10 us, and will at best give you 1/10 of the photons.

This product line includes a 7.5-ampere LED driver that can deliver 50 ns pulses.

These guys light up LEDs for 2 to 7 ns, and say the latter makes a pulse visible to the eye.

I have long wanted to know the phosphor persistence time of white LEDs; once found it to be unmeasurable (substantially less than 1 us) in a quick and dirty experiment.

I agree with others who said to practice with cheap LEDs of similar size and voltage, if possible.
Please tell us what you come up with.

iJim

Sat Apr 17 2010, 08:50AM

Registered Member #2662 Joined: Fri Jan 29 2010, 10:14AM
Location:
Posts: 36

Thanks for the replies guys...

The LED is a UV emitter, 315nm, quartz window hence the price! Destroying a few cheap LEDs is definately the way to go!

I'm not too concerned that I will exceed the maximum power dissipation as I can keep the repetition rate low, <10Hz, my camera only does 1FPS... but has a variable exposure time down to 5ns!

My main concern was the bond wires, these can be seen with a magnifying glass and they are literally microscopic! It's not going to take much to destroy them.

Interesting comment about the capacitance being like a MOSFET, perhaps using a mosfet driver could save me some time, something like a TC4452 would provide ample drive, 12A... 21ns rise time with 10nF load! Worth giving it a try methinks!

Cheers.

hboy007

Sat Apr 17 2010, 09:19AM

Registered Member #1667 Joined: Sat Aug 30 2008, 09:57PM
Location:
Posts: 374

iJim: that sounds very exciting. My first thought was that the device were a 375nm GaN emitter (the cool ones come in TO-3 packages, too) but then I looked at the forward voltage which would not fit.

Conundrums reply is very enlightening, he always knows the little details I can only make a guess about. So my assumption was that a limitation due to the "thermal capacity of the semiconductor die itself" exists, which can actually be identified as the P dt integral, P being normalised by some intrinsic resistance parameter so that P ~ I^2 and the dumped heat is RI^2*T. Google delivers nice REM pictures of cratering caused by wrong bonding parameters. This looks serious!

200 mA does sound like small bonding wires. If you're afraid the wire will blow up, consider it a fuse. You could have someone bond multiple contacts to the chip to overcome the limitiations.

Steve Conner

Sat Apr 17 2010, 09:59AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Also, increasing the current won't necessarily increase the light output in proportion. LEDs are optimized for a range of currents, and the efficiency falls off outside this. I'd take the 200mA figure as being a rough upper bound in this respect.

I once was asked to design an intrinsic safety barrier for a laser diode that would stop it generating enough laser energy to ignite methane, even if the mains were to short directly to the laser terminals. I had a lot of fun testing that with old reject lasers.

A MOSFET driver chip would be a good choice me thinks. The LED driver section of the Ronja free space networking device uses a similar setup.

iJim

Fri Apr 23 2010, 10:57AM

Registered Member #2662 Joined: Fri Jan 29 2010, 10:14AM
Location:
Posts: 36

Just a brief update on this project...

LED pulser: Made a simple test circuit using an IR4428 mosfet driver with a cheap 660nm LED in series with 10ohms.

Detector: I intended to buy a photodiode/amplifier unit, but at Â£300+ it's worthwhile trying to make my own. The circuit is based on a LM6171 op-amp, 30kohm feedback resistance and a pin diode reverse biased at -12v (actual diode in question Osram SFH213, peak response 850nm, 0.62A/W @ 850nm, 5ns rise time). I chose the diode because its cheap and I don't mind breaking a few, will need something suited to UV operation in the final setup!

Test run: I set the IR4428 input at ~40ns pulse width (worst case scenario for my real experiment) and 1kHz rep rate, the LED was hardly lit. The photodiode was held roughly 3cm from LED, definitely not aligned etc... Initially the output looked pretty terribly, lots of ringing. Time for some feedback capacitance, as I'd left no room on the board I had to improvise. Took a two bits of enamelled wire and twisted them together (about 5cm initially), soldered one to the output and the other to the input of the amp. As predicted the ringing had gone but the bandwidth had dropped considerably, hence I started to snip the twisted wires until a reasonable response was obtained (no ringing, decent rise time.).

(blue trace=voltage across LED, yellow trace=output of photodiode amplifier).

Unfortunately I have no reference detector to compare with so I don't know if the broader optical pulse with slowish rise time is because of a low bandwidth amplifier or an issue with the chosen LED. Either way, what I have so far should be sufficient to provide me with decent 100ns pulses for my experiment...

Cheers,

iJim.
(hope I uploaded the photos OK!)

IntraWinding

Fri Apr 23 2010, 12:47PM

Registered Member #2261 Joined: Mon Aug 03 2009, 01:19AM
Location: London, UK
Posts: 581

Might not matter to you at all, but if you manage to drive an LED at extremely high current you may start to get some Laser action alongside the usual purely spontaneous light emission. The currents for this may be way beyond practicality, I don't know, but could result in more directional light output, narrow spectral peaks, perhaps some polarisation too.

Alan

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob