Cree XP-G3, testing a R3 4000K 90+CRI emitter

Thank you for the fast reply.

Your sentiments capture, precisely, exactly what I had been hoping was the case,
i.e., you translated my hopes into printed matter most excellently.

I emailed Mark R. this past week to ask him which 5-bin tint they were but have not heard back yet.
Perhaps he can chime in here to clarify:

1) Whether they really are XP-G3 S3 efficiency bin, and not the commonly available R3 bin
2) Which 5-bin tint they are

Mark, you out there…?

Ohhh - minor/major issue - he's really bad at emailing. Might not respond at all, might take weeks.

MRsDNF (Steve) lives kind of/sort of close to the Cutter store, visits there on occasion, and can help out contacting them, and Steve is very helpful. He might not be engaged on this thread though, so I'd suggest pm'ing Steve to help in contacting Mark.

I was think'n about ordering a couple of those 7X boards - looks like copper DTP. Gotta check but that TPAD board with the 50 mm optic, or 40 mm with a lens, might be a good fit for a SRK, but you'd have to build up under the MCPCB with a pretty big spacer.

Ohh - hear is an email excerpt I got from Mark back In October. This was about an XHP50 originally:

Hi Tom

Sorry thought you may be aware of this, Steve may want to chime in and confirm as he has been here and purchased non published bins from us on a fairly regular basis. The problem for Cree is they will often yield higher than published bins but only in non commercial qtys, they only officially release the bin once they can supply everybody, occasionally works in reverse, just has Cree drop a led 2 flux bins as they are having a yield issue

Cheers

Mark

Okay, that makes sense.

I would feel better before ordering if I could confirm.

The triple xp-g3 4000k 90cri has been available for quite some time at MTN electronics. It seems that everyone is building their lights with Nichia 219c’s instead. The xp-g3 seem to have much higher output than their Nichia 219c 90 CRI counterparts.

Why is this LED not getting any love? Is the tint that bad compared to the 219c? Anybody build more lights using this LED and care to share their experience? Any beamshot comparisons with 219c’s?

I am thinking of putting a triple of this LED in a jaxman e2L… Any thoughts?

Thanks for all the info posted already.

I have a triple with this (bought from Cutter) and it is really good. It is not as popular as the Nichia because it has the yellowish corona around the hotspot that all new Cree’s have, and the tint is slightly greenish, that is why people prefer Nichia.
Both drawbacks are quite minor in a triple IMO, and arbitrary (personally I like the slight greenish hue), I can imagine that the extra output makes up for that.

I actually swapped from the 219C triple to a XP-G3 based triple a few months back in my EDC.

I now run a mixed tint setup with a 90cri 5700k, an 80cri 5000k & 3000k emitter setup. Works out to a bit under 5000k final tint with good CRI and good output. I get a little over 3000 lumens on turbo with it but I generally run it closer to 2500 lumens with a GA cell to keep the heat down.

I did have to switch from the 15007 to the 15011 optic in order to get rid of the tint shift but that works great. I now have a nice smooth tint and a pretty good beam as well.

Total throw is about the same as the 219c with the 15007 as well due to the extra lumens from the xp-g3.

I want to build a triple XP-L2 and see what they would do.

Thanks TA! This is really good info, the kind we've been looking for in the FW3A project. Maybe you mentioned it there already and I missed it or forgot (I do that a lot lately). I (or we) need to actually do a similar test though and do some comparative tests, hopefully with pictures, but just haven't had the time.

Also my personal preference would be this sort of EDC - super quad with a great tint. Along the lines of the FW3A, Astrolux S42, with something in the realm of 4000K-4500K, and the XP-G3 seems like it fits the bill, though the XP-L2 would be the power choice, but not sure about combo of power, tints, and CRI choices.

Did the 15011 cut down the tint shift or completely get rid of it?

I doubt I mentioned it in the FW3A project as I have no idea what you are talking about lol. I have not been on here much the last few months. Got a link?

I have been thinking about the perfect EDC light for some time now and I have a design I have been working on for awhile. All the EDC’s lights I have now just leave something small that bugs me.

I have thought about the triple vs quad setup but I don’t know about you but the S41 quad optic seems very inefficient to me. For example with the same LED’s I noticed I got about the same lumens OTF as my triple yet it was pulling over 20A. Plus the throw was worse and I just saw no reason to use that.

A mixed tint XP-G3 setup works great although finding 3x of a good tint XP-G3 could prove to be just as good. I had to work with what I had access to. I personally find that the sweet spot for tint for me and those around me is right around 4500k-5000k depending on which way the tint shifts (green or red).

The 15011 worked amazing, it completely removed all the tint shift from the edge of the beam unless you really look for it on a white wall. It also still have a noticeable hot spot and a good overall beam.

The XP-L2 would prove to be a better option for output obviously and is available in the same tints as the G3 IIRC. The real difference is the cost, the L2 costs ~3x more. Although I think that it should be capable of hitting ~4000 lumens in a single cell triple, at least as I have it in mind.

Ohh - the FW3A thread/group buy is here: https://budgetlightforum.com/t/-/45538

It's a custom 18650 EDC triple LED design done by Fritz15 from the German flashlight forum (TLF - http://www.taschenlampen-forum.de/). It's got a distinct look (I think) and somewhat unique feature with a tail e-switch.

It started because he was gonna do a small run, but the interest on TLF was overwhelming, so they decided so look around to see if it could be made in bigger #'s in China, reduced cost as well, and teamed with BLF to do it.

Fritz though is calling the shots, rightfully so, - this is how I understand, and agree with it.

Such good feedback, thank you so much!

Interesting idea, very compact and down to earth design. Not really what I want for an EDC myself but should be a nice light none the less, particularly for the price.

Curious why it has 2 tubes for the driver? I had a husky light at one point with a tail eswitch and it just had a normal battery tube, I see no reason why it would need more then that, you are simply changing where the driver is in the light. The power for the driver should be able to be handled by a simple bleeder resistor on the front side I would think? Like I said, I didn’t read the thread to see why it was done that way.

The biggest issue will be heat obviously. With virtually zero mass to absorb the heat it will only have mere seconds to run before overheating. It should really have thermal control but not sure how that would work with the tail mounted driver.

I have taken to using the S2 over the S2+ for triple builds as the thicker copper spacer gives significantly longer runtime before it gets hot. I think I can get about 90 seconds out of it before it is too hot with a GA cell. So turbo is actually usable for something other then showing off.

If it will be run with less then ~3A worth of power, I would prefer a single LED setup with a TIR optic. It allows for anything from a tight spot beam to a ultra floody beam by swapping the optic and you will get almost the same output.

I would HIGHLY recommend using the 15011 optic over the 15007 in that light. Particularly if it will be using XP-G3 LED’s. The tint shift with the 15007 is quite noticeable in that optic, I see the 15011 being much more widely accepted. Plus people can easily swap in a 15007 if they want.

Honestly I prefer the 15011 optic beam for the most part, except for the loss in throw obviously. It is not nearly as harsh and it is easier to see things at closer ranges. If you want throw, then a triple is not what you should be using anyways. The 15007 does look pretty good with the 219B or C though.

It looks like Miller is running the GB or is Fritz? I just scanned the first and last page and it looked like it was miller running things?

Overall it looks like a nice project from the little I saw of it.

Miller is running the BLF side of things, yes, but Fritz has, I think, the final word on the light itself. The way it was described is we are tagging along with the TLF group buy, but of course BLF will probably be the big # of sign ups when it's all said and done.

In the FW3A the driver is in it's normal position, so the double tubes are isolated, and the inner tube simply carries the switch signal. There's other ways to do this of course, but you still need some isolated way of getting the e-switch signal to the front.

Why not simply move the entire driver to the rear? That would seem a lot simpler to me and I see no downsides to this, just less cost and complexity.

The front would just be a simple contact board with a bleeder resistor to feed power to the driver. The driver would be a normal FET+x setup in the rear. This way the e-switch is on the driver itself or right next to it.

I am sure there is a reason this was not done, just not sure what it is.

Thank you for your posts on the tripple optics! Do you have any pictures on the light or any measurements? Currently, we are having a hard time convincing a potential manufacturer that three XP-G3s can indeed reach on output of 3000 Lumens…

Regarding the driver in the rear - there was a discussion about that. One of the reasons is the thermal step-down, which is quite important for this light, since it will overheat rather quickly. For my part, my mind is more at easy knowing that at some point the light will throttle down and not allow any higher modes rather than leaving it up to the user to not activate the Turbo-mode too much (as in “let’s see how hot it can get”). Not that anybody on here would do that but since the light will be sold afterwards as well I feel like this is a concern. Also, there is no space for the FET and the mcu on the switch-pcb. So there’d need to be two boards stacked on top of each other which would also make the construction more difficult. But the main thing is the thermal regulation I think.

Thermal regulation makes sense, I was thinking that it needs to have some good thermal regulation when I first saw the tiny amount of thermal mass. Personally I would sacrifice a few mm for a thicker shelf to add mass but thats just me.

You could run a thermal sensor to the top of the light but that would just leave you back at square one with needing 2 tubes.

As far as numbers and pictures. I don’t have any pictures, I have never been able to get good, reliable pictures that I feel comfortable sharing of any light. The settings on the camera rarely match what I see with my eyes and if I tweak the pics to make them match what I see then they are not really comparable. Plus I don’t have many good places to take beamshots in the first place. If you really want to see it though I could try getting one and see how it looks.

For numbers, those I have. A triple XP-G3 setup can make over 3000 lumens for sure. In fact my highest is ~3500 lumens IIRC but I had to de-shroud the optic in order to reach that (adds about 10% OTF lumens in my experience).

I also don’t generally try all that hard to maximize the output, only 20awg max wires (usually less in these small lights to reduce current), sir800 FET only if I have them laying around ect.

I just tossed my EDC Convoy S8 with the above triple setup on the sphere and got 2900 lumens with a 30Q cell. Remember that these are 80/90CRI emitters and not top bin.

I had a triple I built last year with 70cri G3’s with a de-shrouded bezel that made just shy of 3500 lumens IIRC. I sold it though so I can’t re-test it.

So 3000 lumens is easily doable, at least at turn on, with a triple XP-G3. With a VTC6 I bet my light would just touch 3000 as it sits. Now the 30 second numbers would drop significantly due to the heat but that is inevitable with so little thermal mass.

A 70cri 219C would make very similar numbers as well as another option.

Point out to the manufacture that the XP-G3 is capable of over 1500 lumens each (link in my sig or Djozz has a test as well), so the limiting factor is not the LED but the battery.

This post totally went past me, thank you very much for the detailed reply!
We considered to add some mass to the shelf, but every additional 2mm add around 2 grams of mass, which isn’t very much compared to the overall weight of the light. Cawi did some calculations without considering convection and assuming that the light heats up uniformly (of course that’s very far from the real life). The result with a 2mm thicker shelf (currently the thickness is 3mm) was around 48.5 seconds compared to the 45 seconds of the 3mm shelf, both at 30 Watts of heat. Also, I ran some thermal analysis and there wasn’t a big difference in heat distribution. I think the 3mm shelf does a good job at getting the heat away from the led and then the bottleneck is the small surface of the light.

Yeah, I think we will refer to your test and djozz’ test. But what you are saying sounds really good, so after all around at least 2500 Lumens should be quite realistic. I am also excitedly awaiting the parts I need to finish my light, I never had a light this size with only comparable output.

Again, thank you for the detailed reply, it is much appreciated. Also for your opinion on the 15011 vs 15007 optics :+1:

Yeah, aluminum doesn’t add a lot of mass for the size, I use copper spacers generally for the triples I actually use, and it does make a difference.

2500 lumens should be easily doable even with high CRI LED’s, very conservative number for sure. The G3’s are great in lights that can tame / eliminate the tint shift. Which the 10511 does (I got it wrong before, it is not a 15011). The cute-3 optic also works great with the G3 or L2 LED’s, gets rid of the tint shift and a very nice beam.

Anything that can be done to minimize the amount of overhang of the bezel over the optic helps with the OTF lumens as well. I saw about a 10% improvement by de-shrouding the optic as I said above.

Interestingly, it doesn’t make as much of a difference as one might think. While copper is much heavier than aluminium, aluminium has a much higher heat capacity per weight unit. After multiplying the heat capacity with the density you get that copper has around 50% more heat capacity per volume compared to aluminium. So a light that runs 30 seconds on high with aluminium will run 45 seconds if it’s made of copper.

But I totally agree with you, I don’t understand all the manufacturers hiding a significant part of the optic behind the bezel. And not just one light but all of them, the Jaxman E2l, the new Astrolux S42, the new Manker E14 ii and the list goes on… By de-shrouding the optic do you mean increasing the inner diameter of the bezel?

I think part of why I see the larger change in the heat up time is that the spacers do not contact the body of the light. So it has to move through the spacer, then the pill and then to the body. All I know is that I can get about 90 seconds on turbo with my EDC S8 triple before it is too hot to comfortably hold.

The optic shrouding is due to none of the lights being designed for the triple optic, they are just reusing other lights and stuff that optic in there.

On top of that there is a practical reason, the only way to hold the optic down is to have a lip over the edge unless you use a glass lens on top of it. Which you would need a good lens to not undo all the gains from the deshrouding.

In my case I took the time to carefully machine out some areas to match the optics and thus hold it in while still giving it full output. It made about a 10% differnce in output but also made it less waterproof.

OK, so I finally got around to building several lights including a triple E2L with XP-G3 4000K 90 CRI and a Skillhunt H03 with Nichia 219C 4000K 90CRI and thought I should share my feedback.

The E2L has the Carclo narrow frosted spot optics.
The Skillhunt has the stock TIR lens.

The Nichia 219C emits a very neutral light without any color shifting. Very clinically clean light with slight deviance towards the warm side of neutral.
The XP-G3, for the same color temperature, is noticeably more yellow and also warmer than the Nichia. There is some color shifting on the outer spill of the light (not an ideal comparison as optics are different).

The XP-G3 is cozier but appears to offset true colors compared to the Nichia. I don’t know how to measure color temperature but if the Nichia is a true 4000K, the XP-G3 feels more like 3500K.
As for using the XP-G3 to see actual things as opposed to white walls, without comparing the beam to any other light, its still pretty awesome. The saturation and contrast from the high CRI and warmer tint make everything look super crisp. The light was built as a cycling specific light with the H17F driver. The extra lumen output is appreciated over the 219C for this particular use. For an everyday EDC to use for up close lighting, I may consider the Nichia a better option for the more neutral hue representation of colors.

In comparison to my Wizard High CRI (about 3000K), the XP-G3 leans heavily towards the yellow, and the Wizard emits a more comfortable orange.

My next project will be an Astrolux S41 with mixed tint (2 x 219C 3500K and 2 x 219C 4000K)… very excited about that one!

Thanks again for all your help and feedback!