LED test / review - Cree XLamp CXA1304 COB (B4 color kit E1 / 6500 K) - Are COBs suitable for flashlight use? Maybe...


LED test / review

EN

Cree XLamp CXA1304 (9 V variant)

CXA1304-0000-000C00B40E1

In the meanwhile the market for LEDs is big – very big. There are many types and designs available, and not all LEDs are looking like that in our flashlights. Another big market are changeable LED bulbs and special high power modules for interior / exterior / large area lighting, downlights, and for this usecase there are COB LEDs (Chip On Board) used often.

COBs are not the same like the ‚classical‘ Lateral / Flip Chip LEDs we use in flashlights. These emitters must be soldered on a MCPCB / LED board, which is in turn mounted on a heatsink to ensures efficient heat dissipation. COBs doesn’t need a MCPCB, the LED dies are soldered directly on the substrate which is mounted on the heatsink directly.

Sounds like fun? Maybe! This test I had planned for some time and I always found that interesting to think about alternative LED (designs / technologies) in flashlights. So now I do not only want to test this LED, but also demonstrate the design and LED chip configuration and shows you how good such a COB LED performs in a flashlight optic.

And yeah, the classic overcurrent test is also included here. :)

Technical data

Tj 85 °C / If 400 mA

Order code: CXA1304-0000-000C00B40E1

Type: COB
Binning: 410 lm (457 lm @ 25 °C Tj)
Rated voltage: typ. 9 V (max. 10.5 V @ 25 °C)
Max. forward current: 1,000 mA
Viewing angle: typ. 115°

Color temperature: typ. 6500 K (color kit E1)
Thermal resistance: not specified in datasheet
Junction temp.: not specified in datasheet, Cree recommends a maximum LES temperature of 135 °C

I purchased this emitter at Mouser a few months ago.

You can find the official datasheet here (Cree website, pdf)

First appearance and LED chip

A COB does not look like the classic LED. The white substrate is square shaped and measures 13.35 x 13.35 mm (0.526 x 0.526 in). The so called Light emitting surface (LES) of the CXA1304 is 6.00 mm (0.236 in) in diamater, and the outer diameter of LES (white border) is 7.50 mm in diameter.

The marking ‚1304C‘ doesn’t exist on all CXA1304, though. Some of my other samples are marked, other not - like those in my teaser picture. The solder point left to the marking is needed for thermal sensor to measure Tsp (temperature solder point).

In general, the design and appearance of COBs are relatively equal to each other.

The whole LES is covered with yellow phsophor, unless well known LEDs like XP-G2 or XM-L.

Even without power the LED up the total of 12 single LED dies are easy recognizable.

Between the single chips the area lights up in yellow color which not does not represent the color temperature specified by manufacturer – a feature already known from newer Cree XLamp LEDs like XHP70 / XP-L2.

One of the LED dies is 1.00 x 0.65 mm (0.0394 x 0.0256 in) in size, so one chip is 0.65 mm² (0,0010 sq in) each, which amounts a total of 7.86 mm² (0.0122 sq in).

Behavior in optics

For this test I choose the reflector of a Nitecore SRT7 (first pic) and an unknown low-budget 1 x AA flashlight (second pic).

The single dies can be easily discerned, even in the reflector. At this point I expected a weird light pattern, and – yes, that was it.

The light color is not uniform, yellow and blue areas are visible and are very annoying if the light would be moved also on uneven surfaces. Also this problem is known from XP-L2 and other LEDs, too but these doesn’t show such a clear behavior.

To fix this issue, the use of self-adhesive film (d-c-fix or similar) or frosted lens is basically inevitable. To demonstrate this I used a piece of d-c-fix Milky static which was mounted in front of the reflector:

After this the beam quality and color consistency is greatly improved.

The yellow and blue spots disapperead, the whole light is mixed and the tint get close to the stated 6500 K. Unfortunately (for thrower fans) the beam ist extremely floody, and there is no throw at all.

All reflectors I testes were SMO. But the using of OP reflectors makes no significant difference, the ugly beam pattern is still there but at least something weakened. For some users it could might be ok.

Power and overcurrent capabilities

25 °C Tsp, unless noted otherwise

I mounted the COB package on the heatsink directly.

Due to missing 9 - 12 V LED tests I can compare this COB with the XHP35 HI (highest bin E2) only.

Within official parameters

  • At 1,000 mA (max rated current): 1,024.0 lm @ 10.77 V
  • Power at rated maximum: 10.77 W
  • Efficiency at 1,000 mA: 95.1 lm/W
  • At 400 mA (binning current, but 25 °C Tsp): 538.0 lm @ 9.43 V

Overcurrent:

  • Maximum at 1,750 mA (1,289.3 lm @ 12.23 V)
  • Power at maximum 21.40 W
  • Sweet spot at 1,200 mA (1134.7 lm @ 11.18 V)
  • Power at sweet spot 14.36 W
  • Efficiency at max 60.2 lm/W, in sweet spot 79.0 lm/W

Interesting facts

  • In power profile, this COB performs like a XM-L2 / Osram Oslon Square. The power dissipation is almost the same as mentioned emitters. The die size is almost the same.
  • The XHP35 HI has much more power! It almost reaches their 3,000 lumens at max 3.00 Amperes, at 15.14 Volts.
  • The possible gain of the CXA1304 in light flux is negligable overall. The light flux just rises by 265.3 lm (approx. 20 percent). The reason for this could be the (low / mid power) dies which aren’t suitable for high power LEDs.
  • At maximum current, one of the twelve dies delivers around 107 lumens of light.
  • I do not recommend to drive this COB above the rated maximum current.
  • The Vf of the CXA1304 reaches it’s 12 Volts. In some special cases it might be possible to drive this COB with (boost) driver to use with 12 V LEDs but also 9 Volts.

Now it gots weird.

The official Cree binning B4 for the CXA1304 states a light flux of 410 lm at 85 °C Tsp. For 25 °C measurements, this value is approx. 13 percent higher. In my test however, I got 538.0 lumens (!) at 25 °C Tsp. This equals to 476 lumens at binning conditions and this means also that the bin of my samples is also higher than B4, but C4!

So the thread title should actually be ‚CXA1304 C4‘ and not B4. ;)

I double-checked this result with some other COBs of same series, but other color kit and bin, and these LEDs follow the stated manufacturers specs. LEDs other manufacturers have also expectable results.

Maybe I got higher binned emitters erroneously, because of a mistake by packaging at Mouser? My test procecdure is reliable and proven and the results are congruent, in almost 40 measurements with the most different emitters. Some time ago with the XHP70.2 (English test in TLF here) I experienced the same phenomena.

Conclusion

I started this test with the question that COBs are usable in flashlights at all. The answer is: in principle yes, but… - with some very big compromises.

First, the package size is a problem. The most flashlights expects round MCPCBs, so the COB case has to be modified to fit in a flashlight correctly. A really important and problematic thing are the missing (custom) flashlight driver to power such an LED. Maybe the use of driver for 12 V LEDs (XHP35) is possible for this LED, but I hasn’t tested this yet.

The overcurrent possiblities of the CXA1304 are limited. The increase in light flux was only around 20 percent, despite my very good heat dissipation of my setup.

On the other side there are some very interesting COBs available with very unusual spectral bandwidths, like the Lumileds COB with Fresh Focus technology (specialized for lighting up certain food like fish, meat or bread, unfortunately the Vf is very high). Of the CXA1304 there is also a CRI min. 93 variant available, with nominal CCT 5000 K. With this LEDs in principle it’s possible to build portable lights with ultra high power AND light quality.

Use COB in flashlights?

Pro:

  • Very special spectral characteristics possible
  • Very high light quality at big amount of light possible
  • With some COB series extremely high light flux possible
  • Nice floody beam patterns possible if using certain optics

Con

  • Size not suitable for flashlight
  • in most cases changes in flashlight construction necessary
  • In most cases very high Vf (more than 28 V are widely common)
  • Not suitable to use in overdriven state (Hint: not sure, because I have only the CXA1304 tested!)
  • Special electronics needed

I hope you enjoyed this 'test' and explainations. :)

Greetings, koef3

Mistakes and suggestions are best sent via pm.

This might be quite nice with an aspheric lens right atop the LED for some awesome flood…

Few years ago, I bought an LED bulb just to crack it open and get one of these. Not much experimentation, but it looked quite nice. Wasn’t too much of a fan of WW back then, but am now, so might play with it again if I can find it.

What kind of boost converter (preferably 17mm) might drive this?

The COB can be made more compact by cutting it out in a circle instead of having it be a square.
Simply scratch some of the white coating from the top to make a new place to solder the wires.

The cool thing is that if you get a really powerful and bright COB you can underdrive it a lot and still get usable output at an incredibly high efficiency.
For example, a 30000lm COB running at 300-3000 lumens would be way more efficient than a XHP50.2/70.2 producing 3000lm.
Finding the driver is probably the biggest issue for use in a compact flashlight, since the LEDs are usually 24-48v.

The size of such a light flux monster is a problem too, because the LES can easily more than 25 mm in diameter.
You can look at Citizen, they have 400 W+ COBs in the product lineup. But these need a VERY high Vf (over 72 V or so) so the development of a reliable and powerful electronic is very difficult.
But this might be a problem for all flashlights, because nearly all COB need more than 20 volts Vf.

At this point i’m not sure that underdriving is a really useful usecase for such big light monsters…

Yeah but CRX managed to fit an mt-g2 into a light this size, so I’m sure a dedicated modder will be able to do the same in a 30-40mm light with a large COB :slight_smile:

The question is - it is possible to develop a driver board which is small enough to fit in such a small flashlight? (Without losing much power)
A thing for our electronic masters here. :slight_smile:

Shouldn’t be that difficult to make something like this but in a circular shape:

The advantage will be much lower (if any) if you use multiple XHP70.2 to fill the space that a COB would use.

Not only would that look bad, since the dies would have to be very far apart to fit the MCPCBs, but it would also take a lot more space because you can’t just stick a bunch of XHP70.2s on a piece of metal, they require an MCPCB.
With a COB you solder wires directly to it, no need for any PCB.
A lot more compact and no need to reflow.

I meant custom a MCPCB similar to

I don’t think it will take more space than a 30 klum COB….

Even though an XHP70.2 can do 10k lumens, that is at much higher currents than stock at much lower efficiencies.
If you compare some number of XHP70.2s at 100lm/w to a 30klm COB at 100lm/w, you would need at least 10 XHP70.2s to get the same lumen output.
Same thing applies if you want to get closer to 200lm/W.
So 4 XHP70.2s on that quad MCPCB are probably more comparable to a smaller COB, maybe like 10k lumens or something around there.
Making a custom MCPCB that has 9 XHP70.2s is possible, however I would still go with a COB instead for a more uniform emitting surface.

Thanks for the test koef3!

There has been discussion on BLF on using COB’s in flashlights (there is a thread about it) but in my opinion it is really hard to find reasoning for it, even not the tint, the leds that are suitable in flashlights are nowadays available in some great tints.

The only COB that I would like to see in a flashlight is the new 95 CRI 100W 9mm diameter LES COB from Yuji leds. But for that one (apart from the rather high price) there is the mentioned problem finding a suitable compact driver that can be fed with a limited amount of batteries.

There are several COB TIR optics available though, some as narrow as 10° beam, from around 35 to 75mm.
With the larger TIRs the light has to travel a long distance through the PMMA or acrylic TIR material though…

Is there data available on light loss/mm for acrylic? My understanding thusfar is that the light loss in TIR’s is almost exclusively due to surface reflections.

That’s what I was thinking too.
There is only some losses when the light enter and exits the TIR optic and inside of it when ‘internal reflection’ happens.
(something about this I believe: Refractive index - Wikipedia)

Afaik, light can go trough a material without dispersing/ losses. It will lose some photons when the photons hit ‘something’ and it bounces off.

For example:
Frosted glass: “Frosted glass is produced by the sandblasting or acid etching of clear sheet glass. This creates a pitted surface on one side of the glass pane and has the effect of rendering the glass translucent by scattering the light which passes through” (Frosted glass - Wikipedia)
Milk glass: “Milk glass contains dispersion of particles with refractive index significantly different from the glass matrix, which scatter light by the Tyndall scattering mechanism” (Milk glass - Wikipedia)

So, if there are some losses inside of the TIR optic due to the length of the path the lights has to take, it would mean that the clarity is not perfect.
A 100% perfect clarity is probably not possible (or is it?), so there will be indeed some small losses but that should only be very small.

Please tell me if I’m wrong, always interested in science.

Optical fibers can sometimes be made of acrylic instead of glass, and the light can travel many kilometres.
The losses come from the interface between the optic and air, where not all light will enter/exit and instead be reflected.

Anyway, if you care about throw you should never use a COB.
Even with a “narrow” TIR it will still look like just a big flood.
IMO a COB would be great for up close work with a nice spill and no hotspot, and very long runtime at high efficiency when underdriven.

Obviously no acrylic or PMMA optic is 100% clear.
It’s always worse than clear air.
Usually it is specified (or claimed) to be around 95% light transmittance.
But this is probably measured over a given distance, like perhaps 1 cm (10 mm).
But when you have a 75mm diameter budget TIR, the light has to travel some 50 mm or more through the PMMA or acrylic medium.
I guesstimate there will be at least 20% loss added to the losses of the refractions and reflections etcetera, with a large budget TIR.

^ The 95% transmittance of a TIR includes losses from surface reflections, and those may well be responsable for all relevant light loss, loss from absorbance is probably minute.

Well, that would be good news then. :+1:
I thought those numbers were referring to the light transmittance of the used PMMA and / or acrylic.

By the way, a while back i ordered 2 of these:

https://www.aliexpress.com/item/High-quality-COB-LED-lens-diameter-75MM-10-degrees-24-degrees-36-degrees-Multi-Plaid-Integrated/32421631144.html

and 2 of these:

https://www.aliexpress.com/item/COB-LED-lens-diameter-69MM-15-degrees-Multi-Plaid-Condenser-lens-Integrated-light-source-lens/32661928806.html

but i have a feeling the larger ones eat more light than the less large ones…

Both of them claim “Lens transmittance: 85% to 93%” by the way.