OSRAM CSLNM1.TG & CULNM1.TG 1mm², CSLPM1.TG & CULPM1.TG 2mm²

I can’t remember all of the correct reflector opening diameters, but this new listing from Richard might be of interest:

OSRAM Flat White Adapter / Spacer Ring

Anybody try one yet? He also has 1mm WF’s listed.

The tint is probably good, the older black flat was more reddish than greenish at high currents. The cri is obviously low (the only way to get max luminance), it should be stated in the datasheet. Probably around 65-70.

The xml smo c8 reflector is 7mm but all the c8’s with xpl hi came with larger 9mm holes. Im not sure about the L2…

That was way back then. Now the XM-L2 version also 9mm. I got a 3 piece batch few months ago with XP-L HI and 7mm reflector holes and different centering rings but that was one time only. All others 9mm. L2 has the same 9mm centering ring as C8.

Hey gang, I’m trying to catch up on this Osram CSLNM1.TG emitter by reading through the majority of this thread. I’ll admit, some of the info is beyond me.

I have a clear anodizing C8 host sitting around I’d like to try out a Flat White CSLNM1.TG emitter in. But I have a couple of questions.

  • MtnElectronics lists the Osram CSLNM1.TG FLAT LED as “Neutral thermal path.” I keep reading that some of these Osram emitter are mounted on MCPCB’s which need to be insulated from the flashlight body. Is this true with this setup? Or has that issue been fixed and I can mount it like any other emitter without shorting it out? Osram CSLNM1.TG FLAT LED on 20mm Noctigon DTP MCPCB
  • I would like to run this with a very well regulated driver. Djozz’s graph says 4.5A-5.5A is the sweet spot. Anyone have a US retailer that supplies that? Or is Led4Power going to be my best place to get one?
  • I may try this out in my Jaxman Z1 (Cometa) which needs a 22mm driver. What have you guys been using in this case?

Thanks for the help guys!

Black flat needs insulating. White flat is like most LEDs. No issue. Above driver configured with less 7135 will be good for Z1.

Edit: not sure if u can use retaining ring with that one so ymmv.

Thanks for the help Contactcr!

Thanks a lot for this data Enderman.

I tend to define efficient thrower LED as one that gets high intensity (cd/mm²) while retaining good efficiency (lm/W).
I asked myself a question: how does a collar affect efficiency?
Not efficiency in combination with a well matched aspheric; there the answer is easier. Efficiency of emitter-collar combo.

Collar reduces lm/W. It increases cd/mm².
Emitters tend to be very inefficient when driven to top currents. That’s especially true with Osram thrower LEDs - because they can be overdriven much harder than any other LED BLFers have tested.
If they are driven with lower current they are much more efficient.
The thought was: the top intensity that emitter achieves is a mild one for a collared one. Will the output loss caused by the collar offset by better efficincy of the LED itself?

Thanks to your data I was able to answer this question (note: I estimated lumen. My estimation ignores light that passes through collar and reflection off collar edge).

The answer is “probably no”, collared LED is not more efficient. At 0.96A your collared LED does about 56 lm/W and 227 cd/mm². Emitter alone can reach 247 cd/mm² with 87 lm/W.
At the peak emitter alone does 56 lm/W and 334 cd/mm² (awesome result! L4P measured 285 cd/mm² at 47 lm/W and djozz measured even less).
The closest point of collared LED is 45 lm/W and 387 cd/mm².

For the curious: at the peak a collared LED does 24 lm/W.

[quote=Agro]

Small Xenon short-arc bulbs like the one in the Maxabeam (75W) have about the same efficiency, but they have an even higher luminance. Of course, they have many other disadvantages…

Collars can only be used with lenses.
The angle of light that will hit the lens will not change whether you use a collar or not.
So essentially the lumens that the LED puts out is irrelevant.

What matters is the lumens that actually hit the lens and come out of the flahslight.
Everything else is lost.
By doubling the intensity of the LED, a collar also doubles the flux output from the lens, doubling the effective lm/W.

If the LED is doing ~50lm/W at 5.5A only 25% of that will be captured by a lens that collects 30 degrees.
That is ~12.5lm/W.
The collar will double the intensity as well as the collected luminous flux, giving ~25lm/W.

Just thinking out aloud here… if you had a small enough collar with a large enough parabolic reflector in a recoil configuration where the focal length the parabolic reflector was long enough to use a collar, would there be a benefit?

This is over-simplified, doesn't it? Typical emitter datasheet spatial distribution graphs reveal the light intensity is higher at the center and decreases gradually with increasing angle offset. This means emitters deliver most of its light out the front. Topmost 60° out of an XP-G3 are above 90% relative intensity, whereas at 120° cone borders we're already below 55%.

You’re basically replacing a lens with a recoil reflector, yes you can do that and get the same 2x benefit with the collar, you just need a recoil reflector that has NA0.5
The main reason to use recoil is to collect almost all the light of an LED into the beam though, and using a longer FL reflector with a collar does the opposite.

30 degrees half angle (60 full angle) is 25% of emitted light from a lambertian distribution.
This already takes into account the higher intensity at smaller angles.
That is why 30 degrees is 25% of light, and 60 degrees is 75%.

Just thinking in terms of practicality too. I understand about having the shorter FL, but then you’d need a very large reflector. If one set the FL so that the diameter of the reflector coincided with the 30 deg. half angle of the emitted light, then the RLT would “catch” the rest and give the most intense beam with the smallest possible reflector. Or not?

I was toying around with the exact opposite of a collar - a spherical reflector with a hole in the middle that encompassed the angles where “spill” would occur. Basically the angles that the light does not hit the reflector. I was actually doing this in an attempt to reduce glare in an outdoor spotlight build. But what if we could take that spill light and reflect it back somewhat onto the LED? My test setup was not the best - I was using an undomed BLF GT70, the spherical reflector was hand-made and polished, and the lens was Lexan with an aperture for adjusting the FL of the spherical reflector. I was able to get the spill to almost nothing, but did get some rings around the hot spot. There’s a bevel on the edge of the reflector I made (a drop-off from a large DIY RLT collar) that possibly caused this. Did not do a lux reading before and after because it was cold, and I was too lazy to set the GT70 on a stand. The hot spot did look a little more intense, but hard to tell definitively.

The other thing I was thinking of… there are quite a few “hybrid” reflectors that have TIR optics - either molded as a whole, or as an assembly. Wouldn’t there be a way to create a side emitter TIR that could work with a reflector, retrofit based on its geometry?

Using a collar with a longer focal length reflector would simply mean you can have half the front area and get the same lux.
That means that the diameter is ~30% less.
That’s not a lot.

If you’re going for highest lux, then you choose a diameter FIRST, then decide if you want short or long FL.
Having a long FL (in order to use a collar) means that the light assembly needs to be much bigger.
Think of the syniosbeam, but instead of being 4” thick it needs to be 12” thick.
That’s almost a 1ft cube, way bigger than having a short FL.

A collar is literally a hemispherical reflector with a hole in it.
I think you mean a concave spherical mirror without a hole in it.

The main reason people don’t do it is because the best case scenario is 1.11x intensity, since typical reflectors like the GT collect between 30 and 90 degrees, so ~75% of the light, which leaves only 25% to be recycled by the collar at 33% recycling efficiency (based on the normal collars which collect 75% of light and turn the remaining 25% into 50%).

So a lot of work for 11% increase is not worth it.

I’ll give that argument to Optiforms :smiley: Which is part of the “practicality” aspect. 12” vs. ~8.5” sounds like a lot to me…

Now I’m lost… You just mentioned earlier that ~50% of the light from a Lambertian emitter comes from the front 60 degrees. That leaves ~120 degrees to collect the other 50. In other words, while the surface of the parabolic reflector is collecting more light, that light is less intense. So, if we have a “thrower” flashlight with a beam divergence of around 4 degrees, then about 44 of the total light the LED puts out is going toward spill. Which also explains my theory for having an aperture in the spherical reflector - since 4 degrees of that “spill” light is actually going into the beam. So at 33% efficiency, that’s a 15% potential increase. For a cheap concave mirror, tube, and lens with a hole in the middle - probably about $3 in parts.

  1. you need to think about volume not just diameter.

2) yeah i made a typo, it’s 25% of lumens that are emitted from a 30 degree half angle cone (16 degrees full angle)
This has nothing to do with beam divergence.
A typical forward-facing reflector collects between 30 and 90 degrees, aka 75% of the light. That leaves only 25% of the light to be recycled.
As opposed to a 0.5NA lens which collects 25% of light, leaving 75% to be recycled.

I think it is possible to design a reflector with a particular shape suited to match a crowning aspheric lens beam converging wise. Of course it should.

:-)

There are many optics that work with LEDs. Some of them don’t work well with a collar. But I didn’t mean to fix this parameter. For example it makes sense to compare LED+TIR to LED+collar+aspheric.

ADDED:
Or one can compare out-of-focus zoomie performance.

For me LED+collar together make an emitter. That emitter has limited emission angle. And it has a power-performance curve. I was interested just in that curve.
Actually I didn’t end up drawing it because the numbers above turned out to be enough for me.