Nichia 144A series, an output test of a 6500K 70 CRI sample. Nov 7th: added a test of a 5700K R9050 version of the 144A


(picture of the led after testing, so some superficial damage can be seen)

This is one output test in a small series of tests of new Nichia’s, with led samples provided by BLF-member clemence. These leds were first analysed for colour spectrum by maukka, his thread about that: Nichia NV4L144/W144 AME/ARE CRI and output testing

I have very little time at the moment, for testing and for posting, so this post will be very brief, I may extend it when I have time again.

The output test is of a 6V version of a 6500K 70CRI E1200 output bin Nichia 144A led. I did the test on the custom made board by clemence (‘VirEnce’) which is DTP on the cathode side (the led has no central thermal pad), and I repeated the test with the same led reflowed on a custom non-DTP board that I have had made (see my sigline for a link), the solder pads widened by scraping off some solder mask. The VirEnce board was too big for the hole in my integrating sphere so I made it smaller for the test.

Here are the results, compared to older tests of mine of a Cree XHP50 led and Cree MT-G2 led, which are both also 6V leds. The bin of the XHP50 is unknown because it was one of the first sample leds that was sent to me by Hank of intl-outdoor at the time.

What can be seen (in short):
-the led has good output up to 6A, comparable to the XHP50 test led (but note that there will be better bins available of the XHP50 since this test led), but on both boards it can not handle the same high currents of the Cree leds.
-the cathod-DTP ‘VirEnce’ board performs clearly better at high currents than the non-DTP ’djozz’board, but under 4A (±25W) the difference is not spectacular.

Conclusion: the Nichia 144A does not beat the Cree leds in output, but they still produce a lot of light, and as these 144A leds probably come in some great tints, I can see them having a niche in flashlights that are not occupied by the 6V Cree leds. For modding they form a challenge because of the (for flashlight boards) non-common solder pad lay-out.

As said more testing to come later, but for the next two weeks I have zero hobby time.
Cheers, djozz
_

Edit, nov 7th 2016:

I wanted to test another 144A leds, this time a 90CRI one because the main attraction of this 144A series will be the high CRI versions. From the limited leds I got from clemence I chose a 6V 5700K R9050 led and test it on a ‘VirEnce’ DTP board, because a 6V led is more likely to be used in flashlights than a 12V led, and because the 5700K version was the only R9050 led in 6V present in the collection.
The output bin of this led is E1000, which should on average have about 17% less output than the 6500K 70CRI led tested above (which is E1200).
You can immediately see the difference of the dies of the two leds, the left high CRI 5700K, the right low CRI 6500K. The high CRI led appears to have a thicker phosfor layer, with a more ‘egg-yolky’ colour, typical of high CRI Nichia leds, while the 70CRI led has a more pale-greenish yellow colour.

After I reflowed the led on the test-board I discovered that I reflowed the wrong 5700K R9050 led, namely the 12V version. Should be no different from the 6V one except for the internal wiring of the 4 dies, but I wanted to test the 6V one, so off the board it went again. Before doing that I took this picture at very low current, to show the phosfor layer well:

I reflowed the correct 6V led, but that same picture on low current with this led looks strikingly different now:

It looks like the phosfor layer is very unevenly distributed in this particular led, I doubt that was on purpose. And it may not even be out of specs like this, the average tint and CRI may be perfectly correct, and so can the output be. But you do not want that in your aspheric flashlight, I checked with an aspheric flashlight lens and the hotspot looks like this:

I checked with a small OP reflector as well, those are usually quite forgiving. But no:

Ok, I did not have another specimen of this led, or any other 6V R9050 one, so I did the output test on this one, standard method as described above as well. I compare it with the low CRI 144A led from above.

What can be seen?
*as expected, the 90CRI version has considerably lower output than the 70CRI led from above, about 25%
*from the voltage and output performance you can not see that this led would be faulty. It does do clearly worse than the expected average difference between E1200 and E1000, but if you take the maxiumum output of the E1200 bin, and the minimum performance of the E1000 bin, at the Nichia test current of 1400mA, this difference is (only just) within specs.
*the maximum output is reached at almost the same current as the 70CRI variant, that current is slightly lower, voltage is a bit lower too, maybe all this because it runs a bit hotter, maybe it is just a voltage difference between the leds (well within specs).

Conclusion: I do not know if the tested led is a rarity, I hope so because you do not want a led like that in a flashlight, at least the other 5700K R9050 led has a proper looking phosfor layer. It performs well enough though, no surprises in output and voltage behaviour. I have had no chance to judge the tint and colour reproduction inside a flashlight, but maukka has already covered that.

It was fun to see and test such an odd led! :party:

Thanks for the measurements. I was awaiting this test to see if it could beat the cree.

Thanks for the testing djozz. Its a really good comparison of the differences between the boards as well. :+1:

Thanks for the comments! Some extra notes:
*clemence used a high temperature solder, so to refow the led to the new board I had to use a small blowtorch under the led instead of the well heat-controlled heatblock that I normally use. The led may have suffered somewhat before the second test on the ’djozz’board.
*at the end of the test on the ’djozz’board the led actually died at 7.4A, it shorted out internally to be precise.

Thanks for testing mate. Do you know the factory rated output of the Nichia? It would interesting to see the difference in performance from this 70 cri to a 90 cri 144A.

Here’s the datasheet:
http://www.nichia.co.jp/specification/products/led/NV4W144AM-E.pdf

Rated output at 1400mA
CRI70: 1290 lm
CRI80: 1180 lm
CRI90: 1050 lm

Thanks for the test djozz. The low CRI one looks like it has good efficiency.

Interesting results. Could be hard to control current with the really low Vf.

Oh, I didn’t see it! Thought you haven’t test it =))
Nice test, even for the brief result Djozz. As expected (though I calculated it against XHP50 using 30mm x 30mm).
Can’t wait for the detailed results later.

Yes, I think this one is more efficient too for real world applications. Let alone the colours and high quality CRI variants available

- Clemence -

From 0 - 5A 144A looks more efficient with more output than XHP50. Below 5A the efficiency of 144A is even more higher.
NV4W144AM 3250ish lumen at 32 watt (6,4ish V x 5A) = 101,6 lumen/watt
XHP50 3250ish lumen at 35 watt (6,9ish V x 5A) = 94,2 lumen/watt

For brute power application Cree’s XHP50 is still the King in 5x5mm die. But now we have more options for anything not so extreme. :slight_smile:

You can use the spare 144AM 6500K R70 I sent you for later test, unless you want to keep it. I really hate the blinding blueish light out of it. I got several blinding green dots in my eyes for almost 30 minutes, longer than the other 144A I tested. It’s the “blue spike” I guess

So is this 4 219C size dies?

I’m not sure until I remove the dome (will do that for you this weekend, I’m away now), but my feeling is that the 4 dies are smaller than 219 dies

Here’s a dedomed (and wrecked) 144A next to a dedomed (sliced) 219C, and it appears indeed 4 219-size dies.

I updated the OP with a new test of a Nichia 144A 5700K 90 CRI led.

Strange phosphor coating on that 90+ CRI version!

@ Djozz: Try to bring up the current to at least 10% below the rated current.
All Nichias CRI90+ especially the Ra9050 and 9080 have thicker phospor layers. Seems they need more photons to energized correctly. The 6V needs a minimum 1,4A while the 12V needs a minimum 0,7A. The spectral testing by Maukka confirmed this too.
They also need heavy diffuser or hybrid reflector/collimator or faceted colour mixing TIR optics for narrow beam applications. Otherwise you’ll get the sunny side up for breakfast.


Multi die LED always comes with problem eh?
With XHP50 and XHP70 we got a donut, now 144A give us a spiegelei. Perfect combo for a breakfast! :person_facepalming:

What happend? It died?

It did, at 7.4A . I wrote it in post #4 but I should add it to the OP as well. So I used the dead led to remove the dome in a crude way to have a look at the die size.

Ah, ok. It would be interesting to see how de-doming changes the beam and tint of the high-cri versions.