The steady state depends on what temperature you set it to. It can be anything from room temperature to 70 C.
The sensors vary by as much as 12 C though, so even if left at the default value it will stabilize at very different brightnesses for different lights. So, if you want a specific brightness or a specific warmth, you’ll have to set the limit manually.
Is there a product characterization table for 219C, listing the lm/W data at different temperature? Like the one provided by Cree @ http://pct.cree.com/dt/index.html
Here is the specsheet . It does not give lm/W for different temperatures, but it does so for voltage and output so you should be able to figure things out.
Thanks, based on the inaccurate data readings from the graphs, some wild extrapolation, and rough calculation, it seems 219C’s efficacy is about 10% lower than a xpg2 s2, at 60 degree and 1.5A. That means the steady state output under temperature regulation should not be significantly different.
Hopefully the 5 hot chili ratings for 219C is an exaggeration, because I just ordered one Big Smile
Thanks TK 
, I understand its difficult to config the temperature for two lights. I guess I will have to put some faith in 219C…
Take a look here:
At the same current, the 219C R8000 is slightly more efficient an XP-G2 S4 in terms of lumens / watt.
However, because of the much lower forward voltage, the 219C is going to pull a lot more current from the FET, which reduces the efficiency since lower forward voltage = higher current = lower efficiency.
If we put in some numbers: let’s say:
Running that through the spreadsheet, we get:
Thus the XP-G2 version should be roughly 25% more efficient per watt than the 219C version. That is to say, at some “steady state” wattage, the XP-G2 version will put out about 25% more lumens than the 219C, which would appear roughly 12% brighter.
Wait, if 219C is more efficient than S4 at same power level, and offers better cri, then why does xp-g2 option exist? Just to get a little more throw?
After reading your analysis, I am more certain 219C is the right choice :laughing:
The 219C is more efficient at a specific current level, not power level. Efficiency drops off with higher currents, and since the 219C is less bright than the XP-G2 at any given current, we need to drive the 219C at a higher current to get the same number of lumens as the XP-G2.
For example, at 4 amps, the XP-G2 produces 1200 lumens at an efficiency of 76.7 lm / W.
At the same current (4 amps), the 219C produces 1083 lumens at an efficiency of 80.6 lm / W.
To get the same number of lumens out of the 219C as the XP-G2, we’d have to run 5 amps through the 219C, resulting in 1205 lumens at an efficiency of only 70.3 lm / W.
So at a given brightness, the 219C is actually slightly less efficient than the XP-G2.
However, this isn’t the end of the story; we still need to take the FET and PWM into account. The Emisar D4 is not current regulated, but rather uses PWM, which means the light rapidly switches between on and off. For example, the output at 50%, the light will simply be on at its maximum mode for 1 ms, then off for 1 ms (it’s not quite the simple, but for illustration, let’s just say that’s how it works). This means that the light will essentially always have the efficiency of its maximum mode. Since the 219C has a lower forward voltage, the current going through the LEDs will be significantly higher at maximum than the current going through the XP-G2 (4.92 amps vs 3.11 amps), which further reduces the efficiency.
TL;DR: 219C is slightly more efficient than XP-G2 at same current, slightly less efficient at same brightness, and significantly less efficient at the same brightness when using PWM + FET driver.
Isn’t efficiency for an LED defined as how many lumens per watt you get? And based on the emitter test graphs it looks like in all cases XPG2 produces more lumens than 219C when run at the same current. So isn’t XPG2 more efficient?
The only thing 219C has going for it is its lower vF so it can pull more amps. But vF isn’t the same thing as efficiency is it?
Not quite. Watts are a measure of power, not current, and power = current x voltage.
The lower forward voltage of the 219C helps it out a lot.
The 219C is slightly less efficient at lower currents (< 3? amps), and slightly more efficient at higher currents (> 3? amps), but they’re essentially comparable in terms of efficiency at any given current.
Thank you for the detailed explanation, it is very clear now. :student:
Curiously, can we just throw money into buying FETs with higher switching frequency and capacitor with high density, such that we can smooth out the driver current and get a higher efficiency for 219C?
Me myself not being into electronics at all, earlier posts about this calculated that a functioning circuit in which the block current coming out of a FET is smoothed out requires a much too large capacitor that would never fit in even a medium sized flashlight.
Maybe future designs can expose the circuit through contacts on the host (like h2r), and sell super cap as an accessory. Plug in the efficiency cartridge and unleash the true power of you flashlights!
Would you really want to plug in a cartridge that’s probably larger than the flashlight just to get slightly increased runtime on intermediate modes?
Got to save the earth by using energy efficient lights. 
… and extrapolating a little from that, we can calculate the approximate amount of heat:
Perhaps I should modify the XP-G2’s rating to 3 peppers instead of 2, since it probably makes about 3/5ths as much heat.
(note: 300 lm/W is the theoretical maximum for a white light)
I see all that, know it’s from the data sheets, and think to myself that somebody needs to put the paperwork away and start building some lights. I don’t have ANY XP-G2 multi emitter lights that come anywhere close to performing like the Nichia 219C multi emitter lights.
Led4power’s drivers do something like this, running a FET between its “open” and “closed” states to achieve a form of current control. It works, but it is unstable at lower levels and it requires the driver to be heat-sinked because extra voltage burns off at the FET chip instead of at the LEDs. This limits the amount of power it can handle, and makes the flashlight significantly harder to build due to the extra heat-sinking requirements.
A more efficient approach is to use a boost or buck driver. However, it requires even more hardware on the driver, and some of that hardware can be pretty big. And it’s extra-complicated when the battery voltage overlaps with the emitter voltage, because it won’t always be appropriate to boost or buck the voltage. So it’s mostly seen with multi-cell lights or higher-voltage LEDs. Or with premium brands like Zebralight, with complicated drivers, but even ZL has mostly switched to higher-voltage LEDs lately to avoid the voltage-overlap issues.
So… the idea does exist in practice, but it has a different set of problems.
The math actually worked out in this case, though. I can’t say whether it’s actually correct or if the numbers just worked out because they were defined in a way which was reversible, but when I multiplied the “Watts of light” numbers by the maximum of 300 lm/W, I got the expected 3300 and 3800 lumen numbers I measured in my light box.
So either the numbers are pretty close to reality, or they’re just clever like those silly tricks where someone can tell you your birthday based on some seemingly random numbers. (spoiler: it’s the latter… but they might be close to reality too)
F$#% are you saying that the damn thing is (54.7/67.37) 82% heat and 18% light! wow i didn’t realize it was that bad!