NBD1200 flashlight - XBY 47mm driver (Buck+FET), 1x 46950, 1x NBD1200 144-chip (!) LED - 66000 lm from single LED!

Hello!

I’d like to show a rather unusual light here: probably the most powerful single-emitter lamp in the world.

Some time ago, an NBD1200 LED was developed and manufactured in small quantities. The number 1200 indicates the target power output in watts. I do not know the manufacturer of the LED. This flashlight was developed specifically for this LED as a kind of “technology demonstrator”; again, I do not know the exact manufacturer.

I am referring to the flashlight here as the “NBD1200” because there is no specific model number.

I obtained the flashlight through a user in Hong Kong; the price, including shipping and customs duties, was around €190. The required 46950 battery is not included.

As far as I know, there are only about 100 units of this light; the production numbers for the LEDs are also very low due to costs and especialy the yield.

The quality of the host is acceptable to good. There are minor imperfections in the anodizing, but these are only noticeable upon closer inspection. There are no sharp edges.

There is no fan or handle. Overall, this light looks like a significantly enlarged version of a small, compact light, though it doesn’t completely lose its proportions.

The side switch is easy to operate, though it clicks very loudly, which isn’t necessarily a disadvantage.

(from left to right: NBD1200, Convoy C8, Wuben X1 Pro, Convoy S2+, Emisar D4V2, Emisar D3AA)

Compared to other models, this lamp is large and heavy.

The NBD1200 light weighs 1,052 g empty and a hefty 1,466 g with the battery. It is 233 mm long and has a diameter of 89 mm at the head and 54 mm at the battery tube. The reflector has a diameter of 75 mm.

It uses an XBY 47 mm buck+FET driver. This driver was developed by enthusiasts in China and is specifically designed for extremely high-power LEDs and emitter setups.
It features a 20 W charging function via USB-C, which uses PD 12 V.

Additionally, a power bank function is integrated, which must be explicitly enabled.

The latest revision 4.0x is equipped with a 35 A Buck and 100 A+ FET (Rev 3.x 30 A Buck). The custom UI is kept quite simple.

  • 1C: Start at last saved brightness level (Buck; Turbo is not saved)
  • 2C: Turbo FET
  • 3C: Switch from stepped to smooth ramping, then operate at the last saved brightness level
  • 4C: Strobe
  • 5C: Tactical mode (momentary at the last saved level)
  • 6C: Battery check (in volts)
  • 7C: Turn switch illumination on/off
  • 8C: Turn switch illumination on/off when the light is on
  • 9C: Power bank mode on/off
  • 15C: Temperature control on/off

The ramping is very direct and fast, which feels surprisingly good during operation.

The LED is a custom NBD1200. It is equipped with 144 (!) 55-mil LED chips in a 12x12 array.

With a package size of 24x24 mm and a LES of 265 mm² (!), it is the largest LED of conventional design that I am aware of. Only COBs are larger, but they have a completely different design, which results in poorer heat dissipation and virtually no potential for overcurrent.

The emitter is mounted on a 60 mm MCPCB, which was also specially designed for this LED.

The performance is impressive.

Testing was conducted using a “Varicore” 46950 battery, which was charged inside the flashlight.

Immediately after switching on:

66,000 lm at 331 kLux (FET Turbo).
13,750 lm at 69 kLux (35 A Buck).

Lowest brightness level: 308 lm at 1,550 lx.

The maximum discharge current in FET-Turbo mode is likely between 170 and 220 A (!), which should translate to somewhere between 500 and 750 W of power. It’s quite possible that a different battery or a different MOSFET with a lower Rds(on) could boost performance even further.

However, the latter is not really recommended, as the flashlight heats up extremely quickly as it is. With a cold start at an outside temperature of 10 °C and in FET-Turbo mode, it takes less than 20 seconds for the flashlight head to become hot and too hot to touch. At 35 A in Buck mode, however, it takes significantly longer - more than 3 minutes.

The color temperature is acceptable; at the lowest brightness level, the DUV is 0.004 and 6500 K. At the highest level in Buck mode, it is DUV 0.002 and 6620 K. In FET-Turbo mode, it shifts to pure cool white with a slightly negative DUV (-0.002) at 7000 K, which looks very good. The spectrum corresponds to a typical cool-white low CRI LED with a pronounced cyan hole.

FET turbo, 300 meters to the curve:

FET turbo:

Buck turbo (35 A):

lowest brightness setting:

FET turbo:

The beam is surprisingly usable. Unlike many ultra-high-power lights with a floody beam, where a lot of light lands directly in front of the user and causes self-glare, this beam can be described as an all-rounder. The reflector is relatively deep and produces a relatively small spill.
The spot is clearly defined, as is the corona. The slight ring formation is not distracting and is only visible on smooth surfaces.

There is a slight donut hole. This is noticeable on smooth surfaces, but not outdoors.

The perceived brightness is impressive. Thanks to the all-rounder beam, the flashlight is also well-suited for longer distances, and the self-glare remains reasonably contained even in hazy conditions.

In summary, I can only say: I like it! :smiley:

Not only is the uniqueness factor pronounced, but the flashlight is - aside from the weight - surprisingly useful outdoors. With 66,000 lumens, it’s arguably the most powerful single-cell, single-LED flashlight. The beam is its greatest strength, as it clearly resembles a classic flashlight with a standard reflector.

Add to that a simple UI, the charging function, and the good color temperature, especially in higher brightness modes.

Thanks for reading.

18 Thanks

Wow, that’s 3.23 pounds!
That’s one very powerful single-emitter flashlight.
Thanks for the review! :grin:

Awesome! Should be much more than 66k lm however. I pinged you in discord

Pretty sure the battery is the weakest point here. After some tests in ceiling bounce I was able to get 85000 lm for a fraction of a second (first value schown on luxmeter), but it dropped almost immediately to 62000 lm (battery slightly discharged because of the beamshots). In my sphere it is the same behavior. Don’t think my equipment produces error.

I don’t know where I could get a high quality A-grade original EVE 46950 cell. “Varicore” is just a reseller, who knows where they get these cells from…

3 Thanks

XBY said there are multiple grades of the 4695e from <0.9mΩ to 1.5mΩ. I looked all over aliexpress and taobao and I think most if not all of them were the 1.5mΩ.

There’s the 46120, which got 116k lm on tests. But it also seems maybe this 96k and 116k lm is instantaneous peak, which I expected, but I didn’t think it would drop so fast. No problem really, 300a fet is a lot to ask of course. Still an awesome light!

Will you be testing the emitter by itself?

1 Thank

Ladies and Gentlemen, the new king of Arctic flashlight is here!

1 Thank

This could explain the extremely fast drop in light flux… I could test this with fully charged battery though. Yesterday the battery was slightly discharged (runtime under 1 min on FET turbo)

Are there any reliable sources of A-grade 46120 which ship to europe? Also I would need the right battery tube of course…

Don’t think so. So far I know there is no bare emitter available (maybe due to low quantity, it seems all of them are used for this flashlight) and also I am not able to put a 60 mm (!) board onto my heatsink :melting_face:

If you think a 60mm mcpcb is problematic, wait until you see how much current it can take lol. Sfy55 can take ~150a and you only went to 60.

I don’t know a source for 46120 other than forwarding one from China

I have no plans to buy a power supply with more than 60 Amps.

Btw, I was told that this XBY driver has (or had?) serious parasitic drain issue. There were 5 mA of parasitic drain mentioned, which would completely discharge a fully charged 46950 in about 9 months.

On my rev 4.0x I measured 0.8 mA parasitic drain, which is way better than the 5 mA before. seems this issue was partially fixed. After putting a battery in, for the first time (as long the side switch flashes up) it has 52 mA drain. With 0.8 mA it would take four and half years to completely drain a fully charged 46950.

Impressive light.

Varicore looks to be rebranded EVE cell. Keep in mind there are two versions on the market. The E and P (as in Power). These cells we’re able to buy are almost certainly the E version. P is capable of much higher discharge current.

Amazing. I want one :pleading_face:

The Varicore cell was rated in item description at 320 A (10C)… but yeah, I suspect I got a 4695E too…

You could buy a Lumintop Aurora for around 35 euros with a Lumintop 46950 battery inside

There is only 46800 battery inside

No. Look on AliExpress

He encontrado esto en AliExpress:
Linterna LED EDC, iluminación COB tipo C 46950 Li-ion, lámpara recargable de 32000mAh, puerto USB A, linterna para acampar al aire libre, antorcha Aurora
https://a.aliexpress.com/_EyDBZbK

Ok, google search does not show this special (?) variant at all. Costs 49 euro anyway and it also only has this cell with 32 Ah (so likely 4695E or another “low power” cell) and for this price I can get two Varicore/EVE 4695E, so this is not interesting for me at all.

I am more interested on a source for 4695P cells.

I have it in my Lumintop Mach 5.0, but I don’t know if it’s a high-power battery. What is certain is that with discounts, coupons, and so on, it’s a pretty economical way to buy a 46950 battery. And there’s nothing special about it; Lumintop uses the 46950 in several of their flashlight models. In fact, it replaced the 46800 some time ago

Yet it’s cheaper than Littokala 4695E :wink:

I took a closer look in the Varicore item description, and now they mention only 4C max discharge (128 A max.), which differs heavily from the stated 10C (320 A) in the item listing.

It would also mean that this cell in this light would regularly be overloaded when using the FET Turbo and operated outside the specified specifications… not good.

Wonder what cell was used in the 96000 lm test by the manufacturer.

1 Thank

I found this, but I doubt one could buy a single cell from them