I couldnât use hot air, was too slow and not hot enough. Ended up using soldering pen old-lumens style.
The first time I reflowed it, the LED pads didnât connect, so I tried again but forgot I rotated the PCB 180 degrees and accidentally soldered the LED backwards and blew it while testing xD
Luckily I bought 10 of them!
Next two LEDs I reflowed worked well.
I did two LEDs, since I only had two MCPCBs.
I was short on time, so the second LED (which I did a better reflow on) I only tested 3A-5A.
Here are the results:
As you can see, the second LED was reflowed a bit better and is getting higher output.
The X axis is amps, the Y axis is relative lux, so you just use that number for comparing. Ambient light was 7 on the y-axis.
As you can see, my large CPU heatsink with fan cooling and MX-4 thermal paste does not peak at 4.5A, or even 5A.
My power supply could not do over 5.2 but the output seems that it would keep going and peak at about 6-7A.
Very good results IMO, and driving it at 6A should give me even higher intensity than in DJozzâs tests since it wonât start dropping at 4.5A
.
REMEMBER I measured this is just for me, since my âflashlightsâ use extreme cooling. For realistic results in a normal flashlight donât pay attention to this graph
The max at higher amps can also be explained by a better led, by accident or not. My led (just 1) was tested already 1.5 years ago, Osram may have improved the thermal properties in the meantime without officially calling it a new version. I say that because there is indeed a new version out, the Black Flat S, but just in multiled form, with better thermal performance. They may just used that improvement for the single die version too without telling.
True, definitely the possibility of a better bin or revision.
What kind of cooling setup did you use in your tests?
My heatsink was at least 100W of cooling capacity, with a 120mm fan and copper plate and heat pipes.
MX-4 is also some of the best thermal paste that isnât liquid metal.
.
In the actual flashlight build I will be using liquid metal, it even more heat conductive than solder
I would totally buy a 7 or 8A buck driver from MTN electronics, if they had one seems like the Black Flat could take it.
Thanks Mr. Enderman,
Properly re flowed and pressed led(solder as thin as possible) will perform better than one that is re flowed without press and has thicker solder layer between led and mcpcb. That is imho and ime and with classic lux meter test of same types of led I did years ago.
I am really interested how would they perform in single cell FET lights.
Mr. Enderman any chance you could start up group buy for this led? Or any of you guys of course?
Yeah I did press the second LED down better, that could be it.
Either way it does seem like all the LEDs are the same bin like it says in the datasheet.
They will not be good for direct drive, unless you are ok with less current.
At 4.2v the LED only took 2.5A, and to get it to 5A it required 5v.
Direct drive with a single cell would make it run at ~2-2.5A which isnât very high.
I recommend a CC buck driver for people who want to drive this LED hard.
(at least this is what the power supply showed me, I didnt have time to check it with a voltmeter)
On second though, the 3ft of wire might have made a difference, I bet the voltage drop at the LED is actually less than what the power supply showed.
Maybe like 1v less? that would be more similar to DJozzâs resultsâŚ
Nice, I think my setup did provide better cooling than that since it didnât have a huge copper column that heat had to travel through.
What thermal paste do you usually use?
Yeah it was probably just the power supply.
Now that I know it wonât burn up I will do a test using my MTN max 6A driver and measure the voltage at the LED.
Lexel when we can expect it and how much it will cost?
I see MCU there but is it possible that you individually re flash it for costumer demands (I need only Turbo, mid, low without memory effect so it always starts on high.
The parts price will be around 1.3$ less than the other Narsil/Bistro drivers, as the FET is cheaper and no AMCs
The MCU can be Attiny 13a, 25 and 85 or other
Pin 3,5 and 6 are bridged to the single FET, so it is possible to run different 1 channel output firmwares
@Djozz: thank you for testing the Black Flat two years ago! Because of you my big thrower now does 1.7Mcd. We all might have never noticed this LED if you hadnât tried it.
Youâre welcome. But my own first test in a Brinyte B158 made me believe that in practice this led was not so great as the test suggested. Must have been an imperfect reflow. Good that you guys went on with this led.
In what distance did you measure the light? You need to be further away compared to the same light with XP-G2 because the LED is smaller.
There might also be another reason. Itâs possible that the lens of the Brinyte is not precise enough for such a small LED. You could check this by putting the light into full focus and looking straight at it from a few meters away. The entire lens needs to be yellow (or lit up).
If itâs not, then a fix would be to add a pre-collimator. It makes the LED seem twice as big which âmakes it easierâ for the lens.
I measured at 7 meter, but an aspheric flashlight measures well at shorter distance just as well.
The Brinyte plastic lens is one of the best I have seen, and much better than any glass lens that I have encountered in various flashlights. The die image is very crisp, if the lens was imperfect you would see the imperfect focus coming from the lens part that gives different focus or chromatic abberations superimposed over the sharp die image. But you donât see any of that.
Some recent testing with L4Pâs Oslon board mafe me realise how sensitive specifically the overdriven Black Flat is for imperfect heatsinking, I really think that it is likely that something like that happened at the time.