"Burn-in" of SST-90 significantly lowers Vf

[Techjunkie]

After experimenting with different direct drive solutions and resistance mods in my SST-90 4500K torch, I noticed that even when the emitter has cooled down, it now draws more current at the same Vf than it did when first assembled. It's now possible for me to get 9A+ draw from 3 NiMH C cells. Even more after it has heated up.

I made a video demonstrating the current draw now that the emitter has been burned-in:

http://www.youtube.com/watch?v=C5vrmCCxwcQ

//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
///___UPDATE: Please watch this response video as well: http://www.youtube.com/watch?v=D9WaHtmDrFM&feature=response_watch ////////////////////////
///______________________________________________________________________________________________________________//////
///___In it, I explain and demonstrate how & why my tailcap current measurements are so high compared to the results that others get.//////
///___I also demonstrate the lower current draw measured using the method most others are using.______________________________//////
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////


By the way, the SST-90 4500K mounted with Arctic Silver Thermal Adhesive in the Brightlumens Deluxe Mag D heatsink starts to turn blue around 13A current draw, in case anyone was wondering.

 

[Linger]

Interesting. Is this first report?

 

[Packhorse]

Thats really interesting.

Has anyone else noticed this?

How long do you estimate the burn in process took?

 

[Techjunkie]

I searched for "LED burn-in" and discovered that wquiles reported a lowered Vf of SSC P4 emitters after a 24 hour burn in at 1000mA years ago. Here, the differences in this giant emitter seem much more drastic because I'm measuring the change in draw at a given voltage in Amps, not milliAmps.

Will's old SSC P4 thread here, and another thread started by Jerry B about a similar observation with MC-E here.

 

[HarryN]

This used to be a common aspect of power LEDs, and there were extensive comparisons done on Lux IIIs, Cree packages, etc. I used to commonly burn in Lux Vs for 24 hours to knock down the Vf a full bin code. A CPF member in ID used to post a lot of graphs on it - perhaps someone will remember his name.

This "feature" is not so common now in die produced by the leading firms (Cree + Lumileds) so it is an indication of where the die are likely not coming from.

BTW - if you "burn in" an LED die for 24 hours at full drive current, usually it will not drop significantly after that. The effect is a sort of annealing / recombination / crystal defect related.

 

[Packhorse]

So we are talking about a 1 off 24 hour burn or will 48 half hour burns do the same thing?

 

[Techjunkie]

Who knows what's the safest or most effective. In the case of my experience it was much more aggressive. More like a series of near death experiences for the emitter. It worked though. I just took it for a spin outside and on the three C cells, the image it casts on the bare tree 450 feet to the North, looks as bright as it ever did with the Lithium cells that now overdrive it.

 

[Linger]

Harry N, any suggestions what variables are for this?

Techjunkie - help me out, so if 'burn in' is successful the vf should drop and lumen output maintained or increases right?
As opposed to 'hurting' the emitter where we're going to have a series of mishaps and output degrades?

Is heat a variable in this or is 'burn in' a bit of a misnomer?

I recall a thread a few months ago where a person had 1 dead die in 4 on an mc-e: cpf'er suggested hitting it with a whack of over-voltage and it activated and began normal function.

 

[Techjunkie]

Harry N, any suggestions what variables are for this?

Techjunkie - help me out, so if 'burn in' is successful the vf should drop and lumen output maintained or increases right?
As opposed to 'hurting' the emitter where we're going to have a series of mishaps and output degrades?

Is heat a variable in this or is 'burn in' a bit of a misnomer?

I recall a thread a few months ago where a person had 1 dead die in 4 on an mc-e: cpf'er suggested hitting it with a whack of over-voltage and it activated and began normal function.

I'm not 100% sure, but I suspect heat is the catalyst for whatever change the emitter undergoes to cause it's Vf (even while cold) to be reduced after the burn-in. (Reduced Vf means that less voltage is required for the emitter to draw the same amount of current as it did peviously.)

The result of lowered Vf should be the same light output at the same current as before but now with lower input voltage, or conversely, more light output at the same input voltage as before, but now with higher current draw. It does not increase the maximum possible output, but does reduce the total Wattage necessary to achieve maximum output, due to the reduced input voltage at max current.

In support of my suspicion that heat is a factor, I offer that the torches in which I have surface mounted MC-E emitters to MCPCBs myself by using the hotplate method, all appeared to have the lowest Vf MC-Es that I had encountered. I had originally assumed that was because of the source of those emitters, but now I suspect that the heat applied to the emitters during my crude surface mount soldering may have had a burn-in effect.

Also, from Wikipedia: For electronic components, burn-in is frequently conducted at elevated temperature and perhaps elevated voltage. This process may also be called heat soaking. The components may be under continuous test or simply tested at the end of the burn-in period

As for the MC-E resurrecting the dead die with a blast of power, I don't think that phenomenon applies to burn-in.

 

[ma_sha1]

The result of lowered Vf should be the same light output at the same current as before but now with lower input voltage,

Lowering vf is great news for single cell DD of SST/SSR.
But you will reduce lumen/amp also, that's hard to measure unless you do a controlled experiment. There are some article talk about head speed up the reduction of lumen maintainance (Eventually LED will put out 70% initial lumens at the same drive current vs. when new)

http://lsgc.com/downloads/AN_01_Life_Lumen_Maintenance_and_Reliability.pdf

 

[Techjunkie]

Accellerating reduced lumen maintenance is expected whenever running LEDs at max current. To me, that's a fair price to pay, especially considering that there will be more powerful LEDs out every year.

I'm just happy that now that this LED has been broken-in that it's much brighter running on 3 NiMH batteries than it was when I first got it.

Lowering vf is great news for single cell DD of SST/SSR.
But you will reduce lumen/amp also, that's hard to measure unless you do a controlled experiment. There are some article talk about head speed up the reduction of lumen maintainance (Eventually LED will put out 70% initial lumens at the same drive current vs. when new)

http://lsgc.com/downloads/AN_01_Life_Lumen_Maintenance_and_Reliability.pdf

 

[Techjunkie]

I thought it was important to point out that because of the drastic effect that the smallest amount of resistance has on voltage these currents, and in turn the drastic change in the amount of current that is drawn by the emitter when voltage drops due to that slight resistance, that I have taken special steps to ensure as little resistance as possible is introduced into the circuit while taking current draw measurements.

In a second video (http://www.youtube.com/watch?v=D9WaHtmDrFM&feature=response_watch), I demonstrate the difference between taking the direct drive tailcap measurement with an in-line DMM using 16AWG copper leads (good), versus taking the measurement with a clamp meter with that same low resistance lead (better), versus taking the measurement using the clamp meter and replacing the 16AWG stranded copper lead with a sheet metal conduit (best).

The difference with a fully charged 4.2V pack is literally 8A->9.4A->11.9A by varying the method. Even the best method does not capture the actual current draw with the tailcap in place.

By rough estimate, I'd say most people are direct-driving their SST-90 several amps harder than they are measuring with their DMM at the tail.

The true test will be a clamp meter test around one of the wires between the emitter and the driver (or switch in the case of direct drive). If I ever remove my heatsink again, I'll grab a video or snapshot of that clamp test.

 

[HarryN]

In the past, 24 hours at the factory rated LED current was nearly ideal - very well heat sinked. There was very little change after that. In electrical devices, there is plenty that happens from electrons passing through crystals - heat or no heat. I don't have any idea about 48 cycles of 30 minutes, but if you think about it, that is a pain.

There could also be a thermal effect, and here is why. At the end of growing the LED at the wafer level, the final step is a thermal anneal. Some firms are better at this than others, and some simply do this to a minimum level to save money.

Do not let the LED overheat - that just causes LED death. Guess how I know - lol.

I still do this on my Lux Vs, although I am not sure if it is really necessary. In the meantime, when I am burning in the LEDs, my garage is really bright.

It would be interesting to see what a modern Cree or Lumileds Rebel does under those conditions. I don't have the equipment for that testing, but if you have a proper voltage / current power supply it it is not that hard.

I am pretty sure that the main CPF poster of this info was evan9162. If you do a "theads started by him", it will likely be in his list, assuming that info has not been lost over time.

 

[flashfiend]

OMG! This thread is awesome. Now to temper my excitement although there will be some difficulty, are you guys trying to tell me I can get my sst-90 to draw more amps after a 24-hr burn in process? I just purchased a nailbender direct drive sst-90 and was a bit underwhelmed by the output. After further digging on this forum I discovered that my sst-90 may only be pulling 3.5-4.5 amps from my 26650 Battery Space battery. But now with this thread, it seems you guys are saying I can get it to pulls amps at spec or even more. Granted with a great increase in heat. I am very interested in how to do this and am wondering is this just a matter of amassing 24hrs of runtime on this LED or is there a specific process I need to follow?

 

[JohnF]

OMG! This thread is awesome. Now to temper my excitement although there will be some difficulty, are you guys trying to tell me I can get my sst-90 to draw more amps after a 24-hr burn in process? I just purchased a nailbender direct drive sst-90 and was a bit underwhelmed by the output. After further digging on this forum I discovered that my sst-90 may only be pulling 3.5-4.5 amps from my 26650 Battery Space battery. But now with this thread, it seems you guys are saying I can get it to pulls amps at spec or even more. Granted with a great increase in heat. I am very interested in how to do this and am wondering is this just a matter of amassing 24hrs of runtime on this LED or is there a specific process I need to follow?

I wouldn't get too carried away... I have a nailbender SST-90, and have never seen more than 3.2a at the tail with a fresh 26650, and it drops quickly to 2.8. There are other limitations. I believe the direct drive PWM driver he uses max's out at 6a, but other factors come in to play - the tail switch & spring of the host, etc.

On the bench, I don't see enough difference in output running the SST-90 at 9a over 4a to make it worth it - but the heat gets out of hand unless you are using a great heat sink.

John F

 

[flashfiend]

I understand there are many other factors at play here and I didn't realize nailbender used a pwm driver, but seriously I would be very happy to get my sst90 to pull just 6amps. Heat be darned I want my output. Will burn in help me get closer to achieving this? And if so, how do I go about burning in my LED? If not I may change to a different LED.

 

[Techjunkie]

I understand there are many other factors at play here and I didn't realize nailbender used a pwm driver, but seriously I would be very happy to get my sst90 to pull just 6amps. Heat be darned I want my output. Will burn in help me get closer to achieving this? And if so, how do I go about burning in my LED? If not I may change to a different LED.

Based on my observations, yes, burn-in will get your emitter to draw more current. Also, being that a PWM only driver is in effect direct drive, reducing resistance at every contact point will raise Vin sent to the LED which will also increase current draw. Try using a deoxidizing contact cleaner on all the battery contact points and inside the switch. (I used Deoxit, but it's expensive.)

I wouldn't get too carried away... I have a nailbender SST-90, and have never seen more than 3.2a at the tail with a fresh 26650, and it drops quickly to 2.8. There are other limitations. I believe the direct drive PWM driver he uses max's out at 6a, but other factors come in to play - the tail switch & spring of the host, etc.

On the bench, I don't see enough difference in output running the SST-90 at 9a over 4a to make it worth it - but the heat gets out of hand unless you are using a great heat sink.

John F

How are you measuring current draw? Unless you're using a clamp meter on the wire going directly to or from the LED, then you're introducing additional resistance into the PWM/Direct Drive circuit, which is lowering Vin and reducing current draw because of the lower Vf.

You would not visually notice a difference up close, but the amount of brute force throw is noticibly increased at 7A vs. 4A. Also, I've noticed that the color of the 4500K tint emitter loses its sickly yellow look at higher currents. I've heard others complain that the 5700K version also has a yellow hotspot in most reflectors. I imagine that would also be remedied with burn-in and higher currents.

 

[flashfiend]

Thanks for the help and very informational videos techjunkie. Now this is where I'm a little confused because I'm running my sst-90 direct drive so in effect there should be no PWM unless I'm using a dimmer mode. Is this assertion correct? Btw, I should be getting some DeOxit today. Also as far as current measurement, I would seriously be happy 'just' getting 6 amps which I don't believe I'm getting from the visible output from my sst-90. I won't know until later this week when I get my new multimeter. If I am getting 6 amps then I am very disappointed with the sst-90.

 

[Magic Matt]

The burn in is lowering the Vf, and so pulling more current.... and I realise that more current increases the output and also affects the tint slightly according to the specs. Is this burn in also affecting the tint and brightness, or is it just pulling more current to achieve the same output as it was new? - if it were restricted, to the same current as before, would it have dimmed down and colour shifted slightly after burn in?

 

[Techjunkie]

Thanks for the help and very informational videos techjunkie. Now this is where I'm a little confused because I'm running my sst-90 direct drive so in effect there should be no PWM unless I'm using a dimmer mode. Is this assertion correct? Btw, I should be getting some DeOxit today. Also as far as current measurement, I would seriously be happy 'just' getting 6 amps which I don't believe I'm getting from the visible output from my sst-90. I won't know until later this week when I get my new multimeter. If I am getting 6 amps then I am very disappointed with the sst-90.

I would assume, as you have, that on the brightest mode, the PWM that controls the FET is at 100% duty cycle, so the circuit is constantly closed. The FET itself adds some resistance though, and as we've seen in the video, the slightest amount of resistance changes the numbers significantly in this high current, low voltage application. In other words, compared even to your high mode, direct drive without the driver would draw even more current.

I built a few keychain torches with MC-Es and observed the same difference between the direct drive ones without the PWM only driver and the ones with it (and an MCE only pulls 2-3A at the same Vf that the '90 pulls 5-9A).

Short of unassembling your torch and getting a clamp meter between the PWM driver and the heatsink, the closest to actual measurement that you can take (by introducing as little extra resistance into the circuit as possible) is to test as I have with a piece of Aluminum sheet metal or some very heavy gauge copper and a clamp meter. Whatever number you measure, you can safely assume that without the test equipment in your circuit, the emitter is drawing even more current than you measured.

If you find that you are seeing 6A or more to the emitter and you are still underwhelmed, then you should consider trying different reflectors or optics. That big giant emitter makes getting a tight focus a real challenge (especially with the dome in tact).