LED test / review - Cree XLamp XHP35 High Intensity C4 + E2 (5000 K) - Really high performance in XP footprint?

Koef3, so you were getting 2850 lumen at 25°C temps? Did you do short pulses to measure this?

If you were to measure the temps after 30 seconds, where do think it would drop to? Would it hit 85°C temps in 30 seconds?

My cooling rig is very big and the thermal resistance of it is very, very low especially since I optimized the whole cooling for lowest thermal resistance. So I got almost no drop in light flux due to higher Tj / Tsp.

After 30 seconds the light flux at highest current possible hasn't changed in a significant way (+- 0.5 percent maybe). This means also I have not to measure the light flux in short 'pulses'. However, in flashlights (especially these with less massive material in the head) it is almost impossible to get these really good values from my lab testing conditions. But this depends really on flashlight type and board used.

Are we at the point where we can have 2700 Lumen out the front flashlights from the factory?

No, not really. From my experience flashlights have never the same low thermal resistance from LED to case so the highest current possible is lower than in my tests.

I think real 2200 to 2400 (OTF) lumens out of XHP35 HI are the maximum in a mass-production standard factory flashlight. All above may shorten the lifespan of LED and/or electronics significantly...

I have just recognized that my Nitecore MH23 with an XHP35 HD shows a little less brighter spot or dark area right in the center of the big white spot. It can only be seen when doing a whitewall shot but it’s definitely visible. I checked the flashlight’s LED but all four die areas seem to work fine. The dark area only disappears when being closer than 1,5m to the wall. Any distance beyond 1,5m will reveal that there is a minor dark spot within the spot.

According to the first post of koef3 I assume this is absolutely normal, isn’t it? Unfortunately I cannot post any pictures of the whitewall shot as the camera sensor will immediately equalize the light intensity to the aperture.

That can also be from the reflector not being at the perfect distance from the die.

Found same issue with Nitecore light, seems the focus of the reflector is not optimal

Thank you Enderman & Lexel! Is there anything I can do about it or would you recommend to accept this “as is”? I must admit I have not recognized it when using the MH23 outside.

Just to rule out any own faults:
Can this dislocation between emitter and reflector happen when the flashlight suffers small shocks from falling (during transportation) or is that very unlike to happen? I’m talking about veeeery small shocks, so 4-6cm of drop height. :wink:

The only way to fix it would be to open the head, which I’ve heard is pretty hard on most flashlights from nitecore.
If you don’t notice it much during use, don’t bother.
The reflectors are very solid inside the light, and the thing that would cause a donut hole is a Z axis translation, so no it almost certainly isn’t from any drops.

Alright, thanks a lot! So, I don’t need to be worried about it. The dark spot can only be seen if you really look for it on the ceiling or wall. My MH23 only “dropped” 4-5cm from a bag into a plastic box while inside its holster. So, I guess this does not even count as a real drop. :smiley:

Raw data

Cree XHP35 HI C4

Amps lm Vf
0,20 354 11,72
0,40 623 12,22
0,60 890 12,62
0,80 1113 12,92
1,00 1311 13,17
1,20 1493 13,39
1,40 1686 13,61
1,60 1851 13,83
1,80 2000 14,02
2,00 2124 14,20
2,20 2240 14,38
2,40 2339 14,56
2,60 2413 14,75
2,80 2463 14,94
3,00 2496 15,14
3,20 2496 15,37
3,40 2463 15,51

Cree XHP35 HI E2

Amps lm Vf
0,20 360 11,47
0,40 707 12,03
0,60 976 12,42
0,80 1246 12,74
1,00 1487 13,03
1,20 1702 13,27
1,40 1909 13,49
1,60 2091 13,70
1,80 2273 13,91
2,00 2421 14,11
2,20 2545 14,29
2,40 2645 14,48
2,60 2727 14,67
2,80 2785 14,86
3,00 2826 15,08
3,20 2851 15,29
3,40 2843 15,52

If you create diagrams / comparisons with this data, I would ask you to publish these charts in this topic and also to specify the source of the data used.

I’ve been asking myself a question: what makes efficient thrower LED?
High cd/mm² together with high cd/mm²W?
In a way that’s a good metric. If you want a lot of throw in a thermally limited host, that’s what you need. But a smaller emitter always wins. When your thermal limits are more relaxed that may not be so good….

I wanted to abstract away LED size.
What is “efficient LED”? Here I say it’s one with high efficacy.
What is a “thrower LED”? One with high surface luminosity.

What I call efficient thrower LED? One that can achieve high surface luminosity without becoming inefficient.

So I charted lm/W against cd/mm² using koef3 data. That’s what I got:


  • It seems size abstraction works like expected - compare Blackie HWQP with Ostar Q8WP - their behaviour at high intensity is about the same.
  • XHP35 HI really looks very nice here as well….until it reaches (probably thermal) limits.
  • Looks like if someone reinforced SST-40 bond wires it could go for longer.

Note that alone this chart says nothing about “what LED should I use in host X”. Size is a critical factor here and I explicitly ignore it. But I think it’s an interesting secondary metric.

Domed emitters are more efficient than flat ones?
It’s not strictly false, but from now on I will call it a misconception - because that’s not so simple and often straight incorrect.
If you take a LED in domed and dedomed variant, the former is more efficient, regardless if it was domeless from factory or dedomed later.
But unless you want to have your LEDs severely underdriven, a flat LED with larger die will offer same throwiness (i.e. cd/lm) with better efficacy. Obviously there are often market related reasons why a domed emitter might be preferred. F.e. there is no larger one in the footprint and price point that you’re targetting. But in general I think that (outside of lanterns and other lights targetting extreme efficiency) flat LEDs are just better.

And thank you koef3 for enabling me to draw these charts as well as these conclusion.
Your intensity measurements are what makes your tests special.

Nice chart, thanks. :)

Very interesting. The XHP35 HI reaches thermal limit, indeed. Tj is reached very quickly, because the Vf is very very high (3,400 mA with 2,843 lm at 15,52 V!) which lowers the efficiency significantly. The small thermal pad of the XP footprint is likely to negatively impact heat dissipation.

I added more LEDs to the chart above as well as a few more words.

The binning is not surprising. Intel used to do this all the time with tbeir processors. This is because they have to sell different speeds at different price points. So they would rate one processor slower than another even though they were the exact same chip. Of course once overclickers figured this out Intel started to lock speeds on-chip…

Does anyone have a meaningful experience using PWM with overdrive amps with the Cree XHP35B. What are optimal settings to get a significantly increased LUX while not jeopardizing shelf-life of the LED. What frequency, amps , volts and duty cycle. I have read the publication by Cree on overdriving ( https://www.cree.com/led-components/media/documents/XLampPulsedCurrent.pdf) . I am looking for concrete recommendations. thanks

When someone like ToyKeeper designs Anduril she always tries to use a frequency of around 16,000 Hertz or higher. This should eliminate any visible pwm as well as any noise the driver might make.
Amperage wise, that led can do 2.5A. Some drivers push it to 2.8A, but you don’t get much more output for all the extra heat and stress.

Thanks Jason, and duty cycle ?

The duty cycle is what determines brightness. So Turbo would be 100% duty cycle. As you decrease the duty cycle the brightness goes down.

How would the SST20 (and SBT90.2) LEDs fit in the table with the “LED / Voltage Class / mAmps / Cd/mm2”