Cree XHP70 up to 4022 Lumens and XHP50 up to 2546 lumens - Multi-die leds.

How did you set it? What star was it on? Forward voltage is high at high power, so 2 cells should not be enough to overdrive it. I suspect you don’t have it on a direct thermal path star and it overheated. Use a SinkPAD or a Noctigon, a MAXToch star will work well too. A pair of PF type cell will almost certainly not be too much, not sure about the Efest 35A or Sony C5 level of cell with this one. The XHP70 handled it find though.

The only time I’ve seen an emitter go blue within seconds was under high amperage and with an aluminum star that did not have direct thermal path. Answer that question and I bet you have solved your problem.

I’ve got 3 XHP50’s on Noctigon stars pushing 9522 lumens out the front from 8 Efest IMR18350 cells. No blue light.

Thanks for the advice. It was on an aluminium star. I guess that may be the problem. The changing of colours happened within 2 sec and the heatsink is barely warm enough that’s why I wasn’t quite sure if it’s the alu star. Will try it on a noctigon and see.

Holy cow! That’s gotta be like an aircraft landing light!

The XHP’s are almost here and nobody has the Iris in stock. :_(

Have you tried Arrow Electronics? They usually have a large inventory and are quick to ship, good or even excellent pricing as well.

They say they have 3 of the CA10908 that is supposedly designed for an Osram, bet it’ll work for the XHP as well.

This is what I wanted. Or is the one you suggested a better reflector for the XHP70?
https://parts.arrow.com/item/detail/ledil-oy/cn11770-iris#2nng

CN11770_IRIS

I tested the XHP70 again last night, now mounted on a XHP70 Sinkpad, with the romantically wandering traces avoided by scratching some masking away, does not look pretty but it works alright. What is most important is that the thermal path of the middle pad has improved compared to my makeshift board using a XM-L Sinkpad and copper bits.

It is the exact same led as the first test and I plot the results together the makeshift board results because that is pretty interesting: at high currents this led really wants a good thermal path:

As you can see, the led has suffered somewhat from the 15A treatment of the first test: the output is a tad lower and the voltage a bit higher. But over 6.6A that effect is taken over by the better thermal path of the XHP70 board. It maxes out now at 11-12A.

Still my numbers are very low compared to Dale's numbers, so after cooling down I did some quick tests that were done more like how Dale measures with his flashlights: still the board was on the copper mount but there was no active cooling (fan out), and I went straight to the target current (I tested 8A and 12A), then looked for 1 minute every 10 seconds at the light output:

8A 0 seconds 5340
8A 10 seconds 5280
8A 20 seconds 5230
8A 30 seconds 5190
8A 40 seconds 5140
8A 50 seconds 5100
8A 60 seconds

5050

12A 0 seconds 6631
12A 10 seconds 6258
12A 20 seconds 6164
12A 30 seconds 6024
12A 40 seconds 5838
12A 50 seconds 5744
12A 60 seconds 5557

That helps! apparently at very high currents there is a noticable difference between measuring at 30 seconds and measuring at 'steady state' like I do in my tests. These numbers look a bit more like Dale's, and any difference leftover I am willing to explain by different calibrations or differences within a led bin , or he used a different bin. And my led was already tortured, as can be seen in the graph above.

To get the maximum out of the led I turned the current from zero to 15A in a moment and for a nanosecond I measured over 7000 lumen.

My numbers were from an actual light build, reflector, lens, batteries, real world. So you’ve proven that the losses compute and the emitter is capable of some very serious output. Thanks Jos! :bigsmile:

I was beginning to wonder if that test of mine had been a fluke or something. :wink:

Jubeldum, you might give SinkPAD a shout out as they are a licensed Ledil dealer and have the Ledil District Manager in house. :wink:

Send an email inquiring as to availability of that part number to abdul.aslami@ledil.com

15A power supply! That's pretty sweet. I have had mine go up over 10A before, but it's only rated for 10A so I'm not sure if something will break or go nuts if I push it that hard. I have used a buck converter to push the MT-G2s over 13A with the power supply at a higher voltage, but not direct drive.

Looks like after 9A-10A you're not getting much more output.

My XHP70 leds came today! LEDDNA sent me 10 of the 5000k XHP70 N4 3A leds. I already have the mcpcbs and the patients waiting on surgery, so it's going to be a long week-end. Billy did a great job of getting them out and they were packaged well and arrived quickly.

The four up is a light that used to be 4xP7 leds. Now it will be 4xXHP70 leds in series, with one of George's HBFlex drivers and a 6D Maglite with 5x26650 cells in it. 3 amps per led.

The other light is the SP03 and it will have 3xXHP70 leds at nine amps each. Well, maybe... I think it will have a custom FET driver in it, done by RMM. Not positive yet, maybe a 7135 driver.

That leaves me 3 leds for some Maglite mods.

I use a simple but very decent lineair power supply that goes up to 20A :-) , but only at 16V.

Thanks for that testing djozz, i have been wondering why Dale numbers is often higher, but the differences between your usual testing in steady state slowly ramping up the output and in flashlight blast at full power directly like he does it explains a lot.

O-L, do you plan do diffuse the reflectors to help eliminate the donut hole? Wondering if you’d use your patented clear overspray technique. Been thinking about that…

LEDDNA is pretty quick, and they had bins not available elsewhere. Rock on!

I don't know yet, if I have to, I will. I plan on doing some testing with a few reflectors I have laying around and see what happens. I plan on doing one led raised off the star surface, with copper, so I can try some different heights of reflectors, to see what effect the it has and see if there is any way to minimize the X mark. I will post the results, with photos.

Thanks Djozz, very useful graph as always! Seems like 10A is a pretty decent spot to run these LEDs at.

Would it be possible to also measure the heatsink temperature in your usual test? As it seems to have such a large effect on the output it would be interesting to see how the heatsink temp ramps up in relation to the the current ramp.

I don’t need 90 of them. I only need one. :expressionless:

“Thank you for your quote request, unfortunately I don’t have immediate stock.

Please do note, the factory MOQ is 90.

If you would like me to send in for a quote, I will need the following:

your company name and address
end customer name and location
project name
annual usage

Best Regards,
Hang Tra

Hang Tra
Arrow eCommerce Support”

10861

173 in stock, ready to ship. One at a time or however you like. :slight_smile:

Can anybody guess what 3 XHP70’s do on a A17DD-S08 driver with a high end Toshiba MOSFET?

SupFire M6? Almost dead cells even? (8 Efest IMR18350 Purple)

:bigsmile: