LED test / review - Fireflylite FFL351A (3700 K, 95 CRI) - the rosy allrounder

koef3 · 2024-05-13T19:08:56.919Z

Here is something that I determined as part of a quality control measure and creation of an algorithm for creating a new process to measure luminance.

This diagram shows the ratio of the FFL351A 3700 K 95 CRI tested here between luminous flux and brightness in a reflector at a defined operating current, from 0 to 7 A. The x-axis shows the current, the y-axis the ratio between luminous flux in lumens and brightness in lx. The labels are in German, but certainly still recognizable.
The red line is the trend for these values to get a smoothed dataset for easier comparison.

The brightness was measured in one meter, the LED sat on the large heat sink with the fan switched on, as usual from the tests.

The luxmeter can exhibit a slight non-linearity, whereby this curve looks different depending on the LED type (especially domed/domeless and the design of the LED chip). This effect also occurs with my other luxmeter, to a similar extent. It does indeed seem to be the case that the brightness does not increase one hundred percent linearly with the luminous flux, but is subject to fluctuations.

I don’t know exactly where this effect comes from. It is possible that the refractive index of the silicone changes with increasing current and thus increasing LED chip temperature Tj. The phosphor could also promote altered radiation behavior. In any case, this effect - as far as my data already allows - is less pronounced with LEDs without domes.

ControlTheController · 2024-05-13T20:05:42.420Z

Very interesting discovery, I always assumed lux/lm is constant across current levels but that doesn’t seem to be the case.

So even for the same flashlight, it might not be accurate to estimate relative output of different brightness levels using a luxmeter.

koef3 · 2024-05-13T20:14:21.226Z

This is true, but we are talking of only some percent of difference in luminance (and brightness). For me as LED tester it might be somewhat relevant but for flashlight enthusiasts who want to measure the brightness it should not really matter.

To get more reliable data I have to measure this with another lux meter and more LEDs. Since I am completely busy with updating all my tests to the most recent testing standards in the moment such a research is not really possible unfortunately…

QReciprocity42 · 2024-05-13T22:01:39.856Z

This so strange! The effect is not just not monotone, but also flips in convexity with higher drive currents.

If the effect is less pronounced with domeless LEDs it would seem reasonable to conclude that the effect comes from the dome, though it’s quite unclear to me the exact mechanism as the relationship is not monotone. I would be okay ignoring this effect in practice as the maximum variation is less than 2%, well below performance variations between LEDs of the same type.

NeutralFan · 2024-06-07T13:48:23.262Z

I see that Fireflylite has added a rosy 4000K option, although the CRI is less: FFL351A 3V 10W 3535 Emitter

Screenshot 2024-06-07 at 8.30.55 AM1172×1094 89.2 KB

koef3 · 2024-06-07T15:04:58.843Z

NeutralFan:

although the CRI is less: FFL351A 3V 10W 3535 Emitter

I don’t think the 3 less CRI are visible to the naked eye

Lumaine_Moore · 2024-06-12T08:36:33.081Z

I just received a P01 with 3700k and I absolutely love the tint on this LED! it looks less yellowish than my 4500k 519A lights.

koef3 · 2024-06-12T08:40:10.683Z

Are you able to make Opple measurement or something like this? These should be the newer batch of less rosy FFL351A.

Lumaine_Moore · 2024-06-12T08:45:08.177Z

unfortunately i dont have any device to take such measurements. I will try to take some comparison beamshots with a 4500k 519A with WB locked on a white wall

koef3 · 2024-08-07T20:23:44.776Z

Because I’m in the mood for teardowns right now: The shot of the FFL351A 3700 K 95 CRI.

The LED chips of FFL351A and FFL350RD are not identical. They differ in various respects, including the arrangement of the individual points:

The substrate also looks different for both LEDs; different manufacturers are therefore likely. Again it is very difficult to remove the phosphor layer from the FFL351A, it is very stiff and crumbles away if pressure is applied to.

ebastler · 2024-08-07T20:25:04.870Z

Interesting, considering the same footprint + similar (real) LES size I would have expected them to source the same chips for both. Curious.

koef3 · 2024-08-07T20:26:19.219Z

The actual size of these chips differs slightly, but not much. Around 4 mm² seems to be some sort of de facto-standard LES size.

But these LEDs are clearly different in some aspects. Weird, indeed.

Abarth_1200 · 2025-01-04T23:55:33.215Z

Ordering an emisar d3aa and I’ve been told pretty much the ff emitters are the way forward over the nichia 519 which I’m both used to and like. Although I just need a little nudge to confirm I’m making the right choice. Will the ffl351rd 3700k 96 cri be a good emitter choice for close to medium range edc type uses, note in this edc I would like an emphasise on throw with a not so much flood, my nichia 519a equipped D4K isn’t great for outdoors when you can only get those lumens 50ft in front of you.

QReciprocity42 · 2025-01-05T05:19:21.642Z

A D3AA is going to be pretty floody with almost any emitter config, simply how a triple optic works. If you want more throw, you should look into a single-emitter light.

Under maximum current for all emitters, the FFL350RD throws a bit more than dedomed 519A or FFL351A, but likely not very noticeable unless you do a side-to-side test. The high Vf and masked square LES make it significantly less efficient than the 351A/519A. However, if you plan to use med/low modes most of the time, the 350RD can crank out significantly more throw at the same drive power, just because of a smaller die.

Abarth_1200 · 2025-01-05T08:37:45.352Z

Ah even with a spot narrow optic you would recommend a single emitter reflector over triple for throw, maybe I can live it I have bigger lights I carry when I know I need distance. Maybe I should just stick with the 519 and have it dedomed, can’t remember if I read the beam isn’t as clean in optics with dedome, also is it reduced output?

jon_slider · 2025-01-05T13:29:19.938Z

Abarth_1200:

519 and have it dedomed

519a Dedomed in my D3AA with stock optic has a very clean beam.

d31920×1440 133 KB

It is more floody than my TS10. I have not tried the Spot optic, but they are easy enough to install if you want to test it.

Abarth_1200 · 2025-01-05T13:47:31.623Z

That one’s a perfect example just send it lol

QReciprocity42 · 2025-01-05T19:17:18.462Z

A “spot narrow optic” is quite the misnomer. It is the spottiest among all optics in the Carclo 105** series, but objectively is still very floody.

A dedomed 519A would be a pretty good solution. The 10511 might play slightly nicer with the dedomed 519As, or you could get some weakly frosted tape for diffusion (which is what I do). The number of lumens is reduced compared to pre-dedome, but that is really a faulty comparison because lower CCTs have fewer lumens to start with. For example, a dedomed 519A 5000K is around 3600K, and thus should be compared to a 519A 3500K for output. The reduced die size upon dedoming may also increase the optical efficiency of the TIR. In short, lumens loss should not be a worry for dedomed 519A.

QReciprocity42 · 2025-01-05T19:19:45.503Z

jon_slider:

I have not tried the Spot optic

Just for information/clarity: the optic in your light is the 10511, which produces a very, very different beam compared to the 10507, which I assume is what “spot optic” refers to. The 10507 plays better with domed emitters, while the 10511 is better for dedomed; there are fewer artifacts.

Manulovich · 2025-03-15T23:52:29.445Z

Will the ffl351a 2700k be tested? =)