Fried Green LEDs

A project log for Bullet Movies

Using red, green, and blue LEDs to capture short movies of very fast objects

Ted YapoTed Yapo 11/20/2016 at 23:2820 Comments

I knew it would happen at some point. I killed some 30x30 mil LED chips in this stock 10W green package. The remaining good ones still light off DC:

Actually two others in the left string also appear at higher currents; before the "event," they all had similar forward voltages and would all light about equally at a few mA.

What Happened?

I built this new flash circuit using a NTD5867NL MOSFET driven by the NCP81074B gate driver IC. The circuit has the same topology as I used before - this time, I substituted a 100 milliohm 1206 for the current sense resistor, and started with the 40uF photoflash capacitor. The idea with this test was to see how high I could get the current at reasonable voltages. The MOSFET is rated for 60V Vdss, so I figured 50 might be a decent de-rating for tests.

With the 40uF cap in the circuit, I tested the pulses up to 50V. This is the voltage across the 100-milliohm sense resistor as measured with my soldered-on 2x Z0 probe. I believe that ringing is really present in the circuit, since the Z0 probe terminates the coax at both ends, and has shown clean pulses before. Accounting for Rsense and the 2x probe, the cursor measurement is at around 23A. With a 50V supply, this isn't great.

The optical pulse, as measured with my biased photodiode looks good, shape-wise:

the cyan trace seems to confirm the ringing issue - the light is ringing, too! The optical pulse would be fine for a camera exposure, though, capturing a clean 1.086 us instant.

I wasn't satisfied with 23A - that's less than 8A for each LED chip, and I had pumped 12 or so into the smaller ones before. I knew the 40uF capacitor had a high ESR limiting the current, so I replaced it with a 100uF photoflash unit, expecting a smaller ESR but perhaps a larger ESL. With the 100 uF capacitor in place, I turned up the voltage, eventually reaching the 50V mark. Before I could measure the current or take a screenshot on the scope, though, the current waveform began to droop - pulse by pulse (about 1Hz), the waveform fell little by little until it came to rest at less than half of what it had been. At the same time, some of the LEDs stopped working.

The Cure?

My suspicion is that the LEDs were killed by overstress transients caused by the ringing. I had recently found the paper Pulsed operation of high-power light emitting diodes for imaging flow velocimetry by Willert, et al. In their driver circuit, using a similar MOSFET/capacitor arrangement, they place a BYT 01-400 rectifier diode reversed across the LED. They state that:

Diode D1 protects the LED from potentially damaging reverse currents that arise during the rapid switching transients of the circuit.

sounds like what I've just experienced. I'm not sure about their choice of diode, though - that rectifier diode has a reverse recovery time of 55ns, which sounds too slow. Maybe the LED-killing transients are at the end of the pulse? You can see the red curve above spike negative when the LED is turned off. At the end of the pulse, you don't care about the trr of the protection diode.

My next step is to replace this LED with a fresh one and add a protection diode - but which one? Jack Smith has done some great measurements on forward and reverse recovery time in various diodes. His studies indicate that even the lumbering 1N4007 with it's 3us trr switches on in a few nanoseconds to eat transients. Luckily, I have more 1N4007s than I will ever use.

I am a little concerned by the long bond wires in this LED package, though. Maybe it would be better to go with the 5W LEDs that have (4) 30x30 mil chips in a smaller package? I'll have to try some things out.

Part Two

I replaced the first LED and added a 1N4007 diode. This one lasted long enough to make some interesting measurements before failing in the same way after about 400 flashes. After failure, some of the LED chips only partly glow. Notice the ones in the lower left:

While it was still working, I was able to grab some data about the crazy transients in this circuit. The ringing during the pulse is reduced by the protection diode, but I finally got a look at the transients after the falling edge:

The forward current pulse is around 27A with a 50V supply, but look at that ringing at the end! Zooming in, we get a better look:

That first cycle peaks at 40A! I can see how that could fry LEDs - but it should be going through the protection diode now - unless the 1N4007 just isn't suitable here. Another possibility is adding a RC (or RCD) snubber to damp these oscillations, which look like around 70 MHz.

Even though the LED didn't survive, the optical pulse shape verifies the ringing during the pulse has been cleaned up:

Next Steps

I'll try another protection diode.


esot.eric wrote 11/21/2016 at 16:06 point

Interesting... zooming in on the image after "Part Two" it's clear to me now why they use several small LEDs rather than making one large one. Weird!

  Are you sure? yes | no

Ted Yapo wrote 11/21/2016 at 16:13 point

When I get a chance, I'll try taking a picture under the microscope.  That one is actually of the flashing LEDs - I had to crank the camera f/stop and ISO all the way down an use a multi-second exposure to get the flash and the board all at the same time.  I don't know if the LEDs will light the same way under DC.

You can also see the "blue shift" between the two images I posted - the first one at low-current DC looks green; the high-current one looks cyan or blue.  I have been thinking I might need to use red/yellow/blue or even red/yellow/green LEDs to get good matches to the RGB sensor after the blue shift is accounted for.

But, that's a problem for another day ... hopefully next week :-)

  Are you sure? yes | no

esot.eric wrote 11/21/2016 at 05:00 point

Wow! 40A reverse! Have fun with that :) Got any Schottkys for voltages like these?

This may be obvious, but it just occurs to me you've got several LEDs (or strings thereof?) in parallel...(?) So manufacturing-differences might render some more-likely to fail than others, right? Or, more-importantly, maybe, some might be taking more of the surge-current than others. Some may have slightly lower forward-voltages, etc, causing those to turn on first, at a much higher surge-current than the "average." I have no idea how that'd be dealt-with in a case like this... is the bond-wire-resistance in each path enough to even that out a bit? Or maybe it's irrelevant when working with voltage-surges like these... can't quite wrap my head around it.

  Are you sure? yes | no

Ted Yapo wrote 11/21/2016 at 13:40 point

I have a lot of small-signal Schottkys, but I'm not sure I have anything that would handle this.  I need to dig through the diode box.

Yeah, the parallel LED strings bother me, too.  But a lot of companies make LED arrays like this, so they must work to some degree.  Maybe if they've taken all the LED chips from the same batch, and the case holds them all at the same temperature, and the series connection within each parallel string averages out differences in forward voltage, and the internal resistance of the LED chips works as a ballast - maybe then you can get away with it.  At DC, anyway.  In this high-current pulse regime, all bets are off, and it's trial and error (so far I'm two for two trials/errors :-)

The other thing that bothers me with this particular package is the long internal bond wires.  There may be significant energy storage inside the LED package just from those - and the only way out is through the LED chips.

  Are you sure? yes | no

esot.eric wrote 11/21/2016 at 14:17 point

LOL I like your "Maybe if they've..." statement is about a paragraph long of "ifs" :)

It's my understanding that the reason I've gone through so many LED-flashlights that allegedly were supposed to last for "100,000 hours" is because they keep insisting on installing only one resistor for *all* the parallel LEDs. 

And, yeah, in this case, you're pretty much going to abuse them until they blow to figure out how much they can handle. That's science for yah :)

Interesting point about the bond-wires and energy-storage... Looks like you've got the right team discussing that, below.

I dunno if you've scrap circuit-boards laying around, but I've found a wealth of beefy Schottkys on old laptop motherboards, alongside all the switching power-supplies. Always handy to keep that stuff 'round for when you don't want to wait for delivery-times...

  Are you sure? yes | no

Ted Yapo wrote 11/21/2016 at 14:26 point

I have a pile of old laptops, LOL.

I also just opened up an old UPS to decide if there was anything interesting to keep from it.  I spotted the magnetics right away, not I'll have to go look for diodes.

  Are you sure? yes | no

esot.eric wrote 11/21/2016 at 14:40 point

I tend to acquire electronics which suffered ill-fate... some of those laptop-mobo's've been quite-literally pho'd, others died from tin-whiskering or deseated BGAs... I've only lucked-out on fixing one of those, despite all the internet-rage about how easy it is. The functional ones I try to maintain...

I'd be curious about the results with a small-signal schottkey... I mean, essentially you're doing all this to "small-signal" LEDs anyhow, maybe throw a bunch in parallel ;) Hah, or maybe throw in an anti-parallel LED!

Not sure about all the RC-snubber stuff they're talking about, below, but this looks a bit like a solenoid-driver in need of a flyback diode, to me. Though, I can see how diode-switch-times obviously are a much bigger concern here, and R/C circuits wouldn't have that. Got some reading/learning to do.

  Are you sure? yes | no

Ted Yapo wrote 11/21/2016 at 14:45 point

>Got some reading/learning to do.

You and me both, brother...

I thought about slapping even a 1N4148 on there - there's no reason not to try; the LED is already busted.  None of these parts are really rated for these brief pulses of high current, so I think it's all a matter of experiment.

  Are you sure? yes | no

K.C. Lee wrote 11/21/2016 at 14:52 point

A 3A schottky can handle 100A (absolute max) non-repetitive pulses.

Old PC PSU has high current schottky rectifier on the secondary side for 5V and 12V.  They are mounted on heatsink.  They ranges from 20 - 60A rating.

Broken switch mode wall warts would have smaller rectifier on the secondary side.  1A diodes can handle 40A (rating given for 60Hz).  They can handle much more current for higher frequency.

  Are you sure? yes | no

K.C. Lee wrote 11/21/2016 at 15:20 point

Your external current loop is going to be much more dominant.  Red line is the current part.

A: the area enclosed by the coil which in this case is the shaded area.  l = circumference, N=1

L = Inductance, 
N = Wire Coil Number of Turns, 
µ = Core Material Permeability, 
A = Coil Area, 
l = Average Coil Length.

So try to keep the loop small!

  Are you sure? yes | no

Ted Yapo wrote 11/21/2016 at 15:34 point

Yeah, I'm in synch with you on the loop area thing.  In the PCB designs I've sketched, the LED is on one side, and everything else is on the other side, right behind the LED, to keep the area as small as possible.  The size of this LED package is ultimately the limiting factor here.  Another reason to re-consider smaller packages with fewer LED chips.

The prototype layout isn't ideal, but, I think I should be able to tame this amount of inductance with some circuit tweaks - it's not *that* bad.  Some LED flash circuits I've seen have the LEDs on long wires - not even paired as a transmission line - a horror show of random inductance.

I guess I could build the prototype on both sides to minimize the area.  When I make the next one, I think I'll try that.

  Are you sure? yes | no

K.C. Lee wrote 11/21/2016 at 16:18 point

Even if they are on the same side, you can reduce the loop area by having the break out inside the package.  The break out can be in a L shape to squish the loop to something narrow.  The mutual inductance between two very closely placed opposite current would help to reduce the inductance.  A/l is what you try to reduce.

  Are you sure? yes | no

K.C. Lee wrote 11/21/2016 at 16:24 point

You could try something like this for a single sided proto:

Having the breakout under the package.  The two vertical pieces forms a loop with the LED package, so the area is very small.  The horizontal branches out runs close together carries currents in equal and opposite direction, so part of the inductance cancels out.  The area is small while l is long, so this can help to get the inductance down.

Once you can visualize it and understand how the parameters interacts, it should be easier to do this.

BTW my favourate way of prototyping this to to use a layer of copper tape on top of high temperature kapton/yellow/polyimide tape.  That's in some cases has better performance than a PCB.

  Are you sure? yes | no

Ted Yapo wrote 11/21/2016 at 16:38 point

Oh, yeah, that's a good idea!  I could also put a reverse-protection diode right where the l-sections meet under the LED (if there's vertical room).  Or, where the L's meet at the LED package edge.

  Are you sure? yes | no

K.C. Lee wrote 11/21/2016 at 03:31 point

Looks like all the energy stored in parasitic L have no where to go when you switched off the current.   That and the parasitic L&C is what causes the ringing (or oscillation). Try adding a RC snubber between the Source and drain.  TI's app note on buck converter suggest a snubber of RC time constant equal to the period of that ringing.  Start off with the R in the 100 ohms range.

Might also want to add a series resistor between the gate driver and the MOSFET.  This might slow it down the switching a bit and could help on the rising edge.  I would start with something around 10 ohms.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 11/21/2016 at 04:00 point

Then a less sophisticated/powerful gate driver might also work :-)

  Are you sure? yes | no

K.C. Lee wrote 11/21/2016 at 12:36 point

For good signal quality, you always pick the slowest part that can do the job because fast edges causes problems.

When you switch around 20A of current, there is quite a bit of energy that is stored in parasitic inductance.  You either give the current a different path/bleed off when you turn off the MOSFET (snubber) or slow down the transition so that that energy can slowly bleed off.

They recommend 1.5A for 1000pF gate capacitance (20-25ns rise/fall), 3A for 1800pF (23ns), 6A for 2500pF (25ns) and 9A for 10000pF (60ns).

Judging by the D-PAK package size of the MOSFET that should be around the 1000pF range and could be slightly higher.  10nF is getting to IGBT territory.  

Last time I used a 3A driver was for driving H bridge in a switcher  (Ron is 1m ohm) for 50A range.

  Are you sure? yes | no

Ted Yapo wrote 11/21/2016 at 13:52 point

Ha! It's a 10A gate driver into 675 pF of Ciss.  Too fast?  Almost certainly - a gate resistor to slow down the edges is a very good idea. While I have a busted LED on there, I might as well try some experiments.

I got the same ringing to happen in SPICE last night; once I sanity check it, the simulation might be useful for some other experiments - one thing I tried was dumping the inductive surge into a second capacitor through a diode - I guess it's an RCD snubber without the R.  In simulation, it seemed to work pretty well.

I originally bought the 10A gate drivers thinking that I could use those alone to flash the LEDs; at that time, I thought 10A pulses would be enough.  If I end up using smaller LEDs with fewer chips each, I may end up going back to that idea - a PCB with an array of individual LED packages, each with it's own local capacitor and driver.  For now, though, I like the idea of larger LED packages.  

Thanks for the references, I'll have a look!

  Are you sure? yes | no

Ted Yapo wrote 11/21/2016 at 13:56 point

Since the gate driver has a split output, I might want to use different resistors on the H and L sides to just slow down the falling edge; the rise time seems to be limited by the same stray L that's ringing on the falling edge.

  Are you sure? yes | no

K.C. Lee wrote 11/21/2016 at 15:04 point

675pF is nice for this size.  Sound like it is a good part for the top switch for a buck converter.  

In your case, the MOSFET is only going to switch a few times so the amount of losses isn't going to be that bad (vs a SMPS).  10 ohms series resistor would reduce the current to ~1A which is just about the right size.

  Are you sure? yes | no