XHP70.2 P2 4000k Output test by Texas_Ace - it's over 9000!! lumens and still going strong!

Ok, I finally managed to get my hands on an XHP70.2 thanks to Lumintop.

So I put it on the sphere and was seriously impressed.

I ran this test at 12V so that I could test it fully since my power supply is limited to 15A. I have converted the numbers to the more common 6V in the spreadsheet.

If you would like to see the performance at 12v, simply double the voltage and half the current. So 10A at 6V becomes 5A at 12V (which is what the BLF GT is capable of doing on the stock driver).

This LED is so powerful it overloaded my 95W CPU heatsink. The numbers over about ~120w would be improved with better heatsinking I am sure, although not sure how much improvement.

With better heat sinking I would expect it to preform very much like my XP-L2 test here (since it is 4x XP-L2 dies on a single LED). So the lumen output should keep rising slowly until ~20A before leveling off instead of maxing at around 18A.

The cold start numbers are after letting the heat sink cool down and taking a reading at 5s, it is a good indication of what to expect in the first few second in a light. 30s numbers will be closer to the normal hot test numbers unless you light is massive.

Even after pumping over 180W into this LED when retested at 2A it put out the same lumens as it did when I started, very impressive!

Far as the beam goes, I have only tried it in the GT reflector so far and it is much better then the xhp70 but still does have a darker spot in the middle but it is much improved.

Now for the numbers, It peaked at just under 9k lumens, with a larger heat sink I am guessing it would have made it a bit over 9k before leveling off.

on a cold start at ~24A it did over 10,000 lumens at 5s!

I did a test of a xhp70.2 M4 80CRI for the GT, I am quoting that below.

1 Thank

Whoa! Thatā€™s a monster :+1:

Nice! So your heatpath is quite a lot better than in my emittertest-setup. But not as good as the setup of Kƶf3 on TLF, who also tested a P2 bin, that maxed out above 20A. Except for the highest currents your results match his fairly well btw :slight_smile:
( [LED test, english] Cree XLamp XHP70.2 P2 40E (ā‰ˆ 4000 K CCT) | Taschenlampen Forum )

I was actually thinking about that, donā€™t you have a long piece of copper to stick the LED down into your sphere? That could actually be a bottleneck with this much power. Everything says your heatsink should be better then mine.

I actually tried to use a copper heatsink when I first built my setup, I broke 3 or 4 drill bits trying to tap it for the screws and finally gave up and just went with the best all aluminum heat sink I had. I knew it was a downgrade but at least it worked.

So I am not real surprised that this overloaded it, it was only designed to handle a 95W CPU afterall.

Kƶf3ā€™s results are exactly what I would expect with a better heatsink on the high end.

Hmmm, if I could get my hands on a Brass slug, it is possible I could solder that onto the copper heatsink and then drill and tap the slugā€¦. The question is where could I get a slug for less then the cost of just buying something better for the job.

Awesome numbers!

I hope you will share some results of your testing in the BLF GT. Iā€™m interested in some candela numbers domed and dedomed in a flashlight.

I posted the results so far in the GT thread and in my Giggles review thread. All results are domed so far. I am debating if I want to slice the dome or not, I really like the tint as it is now and it is a pricey LED lol.

I will mot likely slice the dome though and see what it does after I see the results of the latest focusing.

From the GT thread.

Measured throw in my less then ideal Texas conditions are about 1250m of throw at 5A and a bit over 1300m at 6A with the dome still intact.

Iā€™ve never seen your sphere or test setup. Have you ever posted any pics of it?

Excellent testing TA. Thanks for the effort. What do you think the ideal heatsink for testing would be like?

Thatā€™s a whole lotta output!

How long do you let the emitter stabilize between measurements, 30s?

Looks like this LED overclocks pretty well :slight_smile:
I wonder how would it do on LN cooling :slight_smile:

Thanks for the great test Texas_Ace.
Itā€™s really impressive that an LED can handle so many watts and still make it out intact.

Once the price of the XHP70.2 breaks the $8 barrier, budget-oriented brands will start going crazy with their new flashlights.

Impressive results, thanks for the time invested!

I posted a thread awhile back on it here: Texas_Ace integrating PVC sphere with no math involved

Nothing super fancy but has proven itself reliable and consistent, which is the most important part.

A big one lol.

Ideal is a relative term. A better CPU heat sink would do the trick nicely, something without heat pipes and good enough for at least 150W, 200W would be better. The problem is by that point you are usually dealing with heat pipes and that makes drilling and tapping the mounting holes almost impossible.

As a compromise the copper core stock intel heat sinks would work fairly well but man, they are hard to drill and tap. I already tried to do that to one I had.

Water cooling?

:smiley:

Indeed, I was very surprised when the output just kept going up and up and up. It didnā€™t even drop under 100lm/watt until over 7000 lumens!

I donā€™t really measure the time between measurements, just do them at a leisurely pace. Most likely around 30 seconds for each one.

Over 10,000 lumens for sure. It would be really impressive if not totally impractical lol.

Is it a full CPU cooler setup with a fan? How about putting the heatsink in a pan of ice cold water? Be careful what with the amps and water not mixing though voltage is low enough. Alternately inverted compressed air can?

This is a good idea, but Matt is not very knowledgeable in the subject. No offense to him. A simple copper water block, a high flow pump using 3/8ā€ lines and a bucket of water should easily handle the heat. The question though is whether you just want to keep the emitter cool or to better simulate real world conditions of a flashlight heating up?