How high can you push the P7?

[IMSabbel]

Well, i was asking myself this...

I had that old P7 that got a bit damaged in that other thread.
And right now i was trying to destructively test one.

Too bad it failed. It was still running at 10A @ 4.7V. (my PSU doesnt go higher)

The weak point where thermal issues: The P7 was just put on a heatsink with some thermal epoxy, without enough pressure to ensure a thin epoxy film. After reaching 8A, it started to get bluer and bluer. I let it 15 seconds at 10A and then turned it down again, which promptly revered the blue taint when going below 8A again.

Took a look under very low power, and the die is the same as before, no broken bond wires, no additional black spots besides the one that was already there before.

Ah, and it DID get brighter up to the point at 8A where it started to turn blue. Significantly compared to the rated 2.8A.

Dumb thing was my digicam ran out of memory during the movie i made, so it cuts off at 9A (but the turning blue is still noticable). Will link it as soon as i find my stupid micro usb link cable

Anybody elso pushed the p7 to that level or higher? I would be interested if a reflow soldered p7 without that thermal ballast to the HS could go higher before overheating...
(Or is it simply the phosphor getting too hot because of the stokes losses in there. After my calculations, that should have been at least 3-4W at that high power, and the phosphor has a bad thermal interface...)

 

[saabluster]

Anybody elso pushed the p7 to that level or higher?

I did but I couldn't tell you by how much. I was using it to test different switches and their ability to handle high power when the switch failed. Normally this would not be a problem but I was still learning how to operate my PS and had the voltage side cranked all the way up and was using the current side to control what the LED saw. When the switch broke the PS tried to keep the current up by sending the voltage up real high which then jumped the small gap that had formed from the melted contacts and blew the LED. I got it up to 6A before all this happened and did not see any problem running it at this power.

 

[IMSabbel]

Hm. Interesting.
I also ran that LED at 5A for extented periods of time before.

I was just wondering because i remember people claiming that the P7s would be bond wire limited well below 10A.

 

[Der Wichtel]

I was driving the P7 with 7V!!!! But only with 1 to 0,5 microsecond

It flashed with 1khz with an on time of 1/256 of the period.


It worked fine until I went higher up tp 7.4V.

I had to use thick wires and a few (10) extra low esr capacitors otherwise the voltage drop through the wires where pretty high.

Now I'm driving them with less voltage. about 3,7V. It was just a test because I wanted to know how high I could get with the voltage in a strobe mode

 

[qwertyydude]

I'm driving my P4 led at 1.6 amps, with good heatsinking it handles it fine so 6.4 wouldn't be unreasonable so long as it gets really good heatsinking, even better if you fan cool it.

 

[saabluster]

I'm driving my P4 led at 1.6 amps, with good heatsinking it handles it fine so 6.4 wouldn't be unreasonable so long as it gets really good heatsinking, even better if you fan cool it.

6.4?:huh: You mean Amps? Thats enough to blow that P4 up three times over.

 

[Hack On Wheels]

6.4?:huh: You mean Amps? Thats enough to blow that P4 up three times over.

Yes, he means amps. I believe he was saying that judging by how his P4 (1 die) is fine with 1.6A, a P7 (4 dies) should be able to handle 6.4A (1.6 * 4 = 6.4).

 

[Gryloc]

Just for the record, I ran my two P7 emitters at 5280mA several times for several minutes at a time (can't remember those specifics), and the emitters fared well (no tint shifts or permanent reduction in brightness). 5280mA is the max current that my CC/CV power supply can dish out with the fine and rough knobs cranked to their max (it is rated for a max of 5A). My P7 was held down on to an early stock Intel P4 heatsink (with a thermal paste interface) with the stock fan. So, that is ~1320 per die (assuming the current is shared nicely).

Well, I wish I knew what the max current rating was for the gold bond wires were used for Cree's, Lumileds', and Seoul Semi's emitters.
-Is there a standard used, and does someone know the common rating?

I pushed all of my TFFC K2s (0200) to 2.5A briefly (only once each to test their worthiness) after finding a couple would handle that current level for several seconds. :sweat: It was not uncommon for me to test my K2s at 2A for a couple of minutes. Newbie, evan9162, and jtr1962 both did lumen testing of their LEDs up to 2A. jtr1962 even ran an R2 up to 2.8A.

What is important is notice how the lumen output levels off, or begins decreasing, given the current input. As the gains decrease, the advantages of the added lumens, compared to the risk of shortening the lifetime, slims. jtr1962 had superb cooling measures on his R2 and the output leveled off at ~2.5A and was decreasing by ~2.8A. Obviously, I was not measuring the output of my K2s, which were attached to a large aluminum heatsink, using a thermal paste interface. The output of my K2s probably leveled off much sooner than 2.5A.

This is why I wished evan and jtr, or someday, Newbie, would chime in and explain every detail of "overclocking" these quad-die monsters, even with good heatsinking. The thermal resistance plays a big role, so there is a chance that the output would level off way before you reach the fuse current rating of one of the bond wires.

On the subject of fusing, I did test this on an old, gen-1 K2 that I had laying around, but I cannot remember the exact current that I powered the LED before it failed (something above 3A). I need to start writing these things down! Maybe, if time allows, I may have to do a destructive test on another pre-TFFC K2 (still a sad loss of life). I have two Cree emitters (XR-E and XR-C) that I recently sacrificed to test output reductions when the dome and retaining ring is removed (I still have to post the results -oops), so I can test the fusing currents on those emitters as well. The nice thing about the Cree XR-E is the use of two bond wires each for the anode and cathode. I am worried that my power supply will not be able to supply enough current to fuse both bond wires (would I need over 6A?).

So, for the P7, with awesome cooling (since there is a larger power density), you should be able to drive each die to 2A semi-comfortably, right? This is a total current of 8A, assuming there isn't any current hogging with any of the four dies (thanks to the lacking of perfect Vf matching). I ran my TFFC K2s at 2.5A, so if the bond wires are the same (which they should be better since there are both two anode and cathode wires), then you could run the P7 at over 10A. However the dies will be screaming in pain I bet. :mecry: So, if the P7 has the four bond wires per die, then theoretically, if everything stays cool, couldn't the P7 be driven at 20A? Just curious...

The MC-E is nice since it is possible to wire it in series and run the entire emitter at the current level of your choice. However, the P7 has four dies bonded directly to a copper slug (same concept as the P4). The MC-E is different (I still didn't get mine shipped yet to observe it more closely). The MC-E has a smaller thermal pad, meaning a smaller region to transfer the same 10W+ of heat than the P7. This is scary to me! Fortunately, the Cree did not skimp on bond wires with the MC-E...

-Tony

 

[Holzleim]

I'll do some high current testing with an MC-E soldered to massive copper in the next weeks.

The thermal resistance (junction-to-thermal-pad) is claimed to about 3K/W, I think in such a direct-solder setup I'm planing do do in my next multi-MC-E lamp one should be able to reach nearly the 3K/W from the dies right into the copper.
If this is the case one can easy calculate maximum sense full current to not overheat the dies themselves.
What I'm not sure about is the question what can happen with the phosphor???
IMSabbel mentioned calculation of 3-4W power loss WITHIN the phosphor; is this based on some conversion loss from absorbing photons of wavelength "blue" and emitting ones at wavelengths in the "green-to-red" range?
Does anyone know how hot the phosphor is allowed to get before excessive damage is caused?

 

[saabluster]

Does anyone know how hot the phosphor is allowed to get before excessive damage is caused?

Only one way to really find out.

 

[IMSabbel]

IMSabbel mentioned calculation of 3-4W power loss WITHIN the phosphor; is this based on some conversion loss from absorbing photons of wavelength "blue" and emitting ones at wavelengths in the "green-to-red" range?
Does anyone know how hot the phosphor is allowed to get before excessive damage is caused?

Yeah, the 3-4W would be stokes-losses: Somewhere, the energy has to go...
I asked that question before, and got as a reply that there _are_ very durable phosphors available (for use in some kinds of discharge lamps). But no real numbers, and somehow i doubt the ultra-efficient coatings for leds are also that tough.