Nichia 219C, testing a 5000K 83CRI emitter, comparing with a XP-G2 S4 2B and other leds

So… next we need high CRI monitors to view beamshots on! :bigsmile:

Well, actually that is far from enough: you need camera's in which each pixel records the full spectral distribution of the incoming light, and display monitors that in each pixel not just emits RGB, but full spectrum as well. Not until then you are close to a solution :tired:

Now Nichia needs to scale up the 219 to make an emitter to complete with the XM-L2/XP-L.

Tester's horror!

Last night I was measuring things in my integrating sphere and I had the uncomfortable feeling that the reading of my luxmeter was a bit high, like a few percent, not much. And during the testing suddenly the readings were way off! It appeared that the battery was empty, it was actually very very weak. The meter apparently can run the battery very low before failing, but does not give any indication in the display, which was quite unexpected for a high-end luxmeter. And then the really horrifying thing struck me: how long has it been reading too high?

The last serious test I did was the Nichia 219C, so this one had to be done again :tired: . Luckily I had ordered two more of these, along with two for another BLF-member, so I did the test again today. Here's the results of the second led together with the first results from the OP.

Ok, I got away with it, the second led does not test significantly different from the first one, so the story of the OP still stands. As you can see the second led has a slightly higher Vf (~0.03V), and (only at the higher currents) a slightly lower output. So the efficiency is a tiny bit less. The output difference above 5A could be explained by a slightly worse heatsinking, inside the led, or perhaps it is reflowed a bit worse.

I does look like the luxmeter was working fine during the test in the OP.

This exact LED according to the datasheet has a tolerance of 6.25% So the same exact LED under the same voltage bin can have a significant difference without thinking of reflow or micro-metric errors.

10 amps! Your not doing any testing on me. Thanks for the update djozz and sense you must be scratching for something to do?

@ djozz: Oh man, I feel a bit stupid now, because I’ve stumbled upon that test you did some time ago, but without realizing that it was a NCS… 219B. Even that one looks pretty interesting even though it’s a B. They have NCS… 219B-V1 and I regret that I haven’t ordered some of them last time I ordered some LEDs from Nichia via Lumitronix. Anyways, it’s a pity that there isn’t a NCS… 219C, because if there was, it would probably outthrow every other LED we’re using in our builds…

You may be right about photographing tints of LEDs and their color reproduction, no, sure you’re right about that. You can’t show how they look to your eye in real life in a photo. But I think, you can show at least a little difference to lower CRI LEDs. I’m going to try it out, not going to be of great scientifical value, but fun to me. I have NVS…219BT-V1 in sw40, sw50 and sw57 and the first thing I want to do with them is building a single P60-drop-in with each tint, just to find out what I like best.

Thanks djozz, very good work as other ones you have done in the past. I am not familiar with the issues of dedoming Nichias but isn’t the gasoline method safe and reliable with Nichias? Why did you try the hot dedoming instead of the gasoline method?

It is a bit of stubbornness, I had one bad result in the past with the gasoline method and hot dedoming always works fast and reliable for me (with Cree emitters). And there's only so many emitters that I want to sacrifice, so I leave it to others to find out about this. My fear is because Nichia's have stiffer domes gasoline dedoming does not work out so well as with Cree leds.

Djozz, i have a favour to ask ( for us all :wink: ) Can you pleace test Nichia vs XPG2 with a smaller heatsink, or straight in a C8 body
Those would be real life flashlight conditions, it would be interesting to see the lumen output when temps are 85+ deg C

Sorry, I'm done testing for a few weeks, from tomorrow on I will be away with the family camping on Terschelling :-) . There will be enough opportunity to keep rattling on on the forum, but nothing practical can be done.

Thank you for this test!

KKW sent me the link to this graph yesterday and in turn, I became interested in Nichia LEDs for the first time in ages after reading it. So I went looking for some lights I have that were not worth doing anything with, and found that I do have 219B LEDs on hand. Both “NCS” and “NVS” versions, which is like XPE and XPG. Not the C version, no. But something to fiddle with.

Here’s a few things I noticed:

The 219B uses a first-layer semi-conductor that is somewhat sunk into the apparent die face. It exists on what looks like a full circular trace of copper or gold plated metal—no bond wires anywhere I could see. You can scrape off the entire yellow phosphor layer, taking no precautions, as if trying to kill it. When I finished doing this, my NVS lights up blue just like the first layer is supposed to emit blue. It appears the die phosphor is added, then a white solid glue/silicone is added after that. Which creates the look like the first layer is buried in the ceramic. This construction technique must have something to do with the very large amperage overage they can take without popping. Try scraping a Cree down to the semiconductor and see if it still turns on! 0:)

The B version uses a pattern of quantum wells that are in-line. There are 4 rows of 4 dots for 16 dots visible on the NVS 219B, and just 2x2 on the NCS version.

What I noticed is that the C version you tested here is very similar in look to the Cree semi-conductor, they are staggered rows of quantum wells under the phosphor, not straight rows+columns. This indicates the 219C may be using the same semi-conductor wafer material as found on Cree’s E2/G2/L2 line, they maybe just get power to it very differently.

All in all it’s definitely worth trying, though the de-dome will go much different I feel. I’m about to find out what happens after they soak for a day.

So where, how, and how much for these LED’s?

Henry, you know where to get them, stop playing games with us. :stuck_out_tongue: :beer:

Here’s some of what I was thinking about, when encompassing the “high” forward voltage changes we are seeing over time, and just general LEDs we are seeing as time moves ahead. It’s important to look at another industry to understand where I’m leading.

If you think about the core technology driving the common white phosphor-LED, most know what it is. I think just a grade of silicon semiconductor, just like Intel and AMD use to do processing with computer chips—with LED silicon arranged to do a much different type of work when made. If you think about the core “product” in these LEDs that can make them potent for throwing, then, it’s really not the phosphor, the bond wires, or the domes really; the “nuts and bolts” are the same between most of the MFGs. Manufacturers are capable from what we’ve seen, of making near identical components like those, and arranged how they want them. I would ask what it is that they (LED makers) cannot completely control, then, and again I think of the semiconductor.

Silicon semiconductor wafers are usually something a large company buys from another large company who specializes in the production of intricate, advanced semiconductor growth through advanced methods of stereo lithography types of production. The LED maker implements that final chip somehow into their product as they see fit, to do its job as they see fit, which includes handling heat dispersal methods. The silicon itself has much different standards than the rest of the LED, but likely similar in many.

It would not surprise me then, if some LED makers used the same exact silicon wafer material in their LEDs.

This is also why I brought up the ‘quantum wells’ looking similar, or the “holes” you see in an LED die surface now instead of line traces—between Cree and Nichia.

Have you ever heard of Moore’s Law? Take a look at computer processors of recent years. “(Moore’s Law)”:Moore's law - Wikipedia It’s not an actual law, but it, for the most part held true for a very long time as a ratio to product.

Then silicon makers started meeting the end of a dark tunnel; they were getting close to exceeding the trace limit width, defined in our physical world by the size of a few atoms. Talk of this started happening as far back as 2010. We cannot break this law, fundamental to nature, or make something smaller than an atom. Computer processors adapted, instead of increasing transistor count by making the chips smaller with more micro transistors and traces, they condensed more CPU chips into one piece of silicon, creating larger, multi-core processors, that weren’t actually smaller inside; the overall package and footprint remained the same. It was like taking a Prius car, and combining 4 of their engines, but still inside one Prius. :wink: The “engine technology” is there which is efficient and now, and power still grows. Oh well, Moore knew his ‘law’ had to have an end someday. :wink:

Someone mentioned on BLF the fact that chips have gotten bigger, brighter, but for this many years to go by, surface intensity has not increased exponentially above the old XR-Es we would likely have expected back then while predicting forward.

If you feel this same sort of parallelism between silicon makers, it tempts me to believe the LED world is running down that same tunnel. They can get the silicon to do a little better, and a little better some more, but to get the lumens they need, they went to mutli-chip use instead of driving up the intensity, regulated by the silicon semiconductor inside. I believe we will see LEDs stop at their current sizes, and possibly start to grow outward. This is even showing by Cree claiming new intensity, by simply applying little dome material. If they could, the dome would stay, and we would see those big intensity gains. The market isn’t calling for it. Efficient chips, with high L/W are needed.

My point with all this is not a great one, I just think it might offer some explanation as to why we are seeing things that make us flashlight users frown at times, like high Vf.

So unless everyone out there forgets about flood right now, starts buying aspherics, and showing what a sucker they are for big THROW, Cree simply isn’t going to pay any attention to us. Ha, if only… :bigsmile:

Whew. That’s a lot of words, MEM.

Here’s one picture:

Recent news from Intel shows Moore’s law to be stumbling and staggering a little.

On the other hand, Haitz’s law, which applies to LEDs, seems to be healthy.

Note the logarithmic scale on the vertical axis.

Great test and results djozz, great to see how far these new gen emitters can be driven, the new nichia certainly looks much improved and the xpg2 out put is impressive, thanks for posting, much appreciated.

I’ve been in contact with a company to buy 219C’s. I’ll probably be ordering them tomorrow, with extras to sell.

LEDs are not built out of traditional silicon. They use things like indium, gallium, nitride, with a sprinkle or two of stuff like phosphorus. Some LEDs use a silicon carbide base, others use a sapphire base or some other exotic material. There is a LOT of work going on to find a good way to build LEDs on cheap silicon substrates.

All of these materials are a lot harder to work with than pure silicon. The bare material wafers used to build LEDs are smaller than what is used in IC manufacturing. The performance/characteristics of the device are very sensitive to minute thingies in the atomic structure of the device… much more so than a silicon IC. This makes the manufacturing processes harder to control and harder to get consistent devices.

I thought I'd read somewhere here that they weren't going to be manufactured.

DBcustom had spoken to a Nichia representative who told him that the 90+ versions were not going to be produced, other CRI's like obviousky the 83CRI one of the OP, are (somewhere) on the market. Apart from Lumitronics, they are at Ledrise now (also Germany, same types as Lumitronics has).