LH351D slicing measurements : output, tint, tint shift, beam profile, intensity.

After putting some sliced LH351Ds in a light I was curious of the effect on the beam, and especially depending on slicing height, I sliced them the closest to the phosphor layer, as I heard people do, but didn’t really know if that was the best, so here we go.

Slicing height = thickness of the washer placed on the MCPCB to guide the razor blade. 0.85mm is as close as possible to the phosphor layer.
Pictures can be clicked on to open them bigger, on mobile if quality is disastrous, request desktop version.

Tint vs slicing height of 3 samples at 100mA, 500mA 1A and 2A. Lowest current is always lowest CCT :

0.85mm doesn’t bring tint (duv) improvements but continues to lower the CCT, it allows to make a very warm beam out of a 2700K LED. Lowest CCT 3535 high CRI LED available I guess ?

Tint graph of tint vs angle at 100mA:
(A 5000K F2R2S2 LED but not the same sample as above)

CCT vs angle at 100mA :

Duv vs angle at 100mA :

At 1mm tint shift is virtually inexistant on a white wall.

Relative intensity vs angle at 100mA :

+70% intensity, +30% throw distance at 1mm
At 0.85mm intensity decreases despite the beam being slightly narrower.

Relative luminous flux :
(A 5000K F2R2S2 LED but not the same sample as above)

1.26mm (not the same washer of 1.20mm as previously) :

0.1A 89.2%
0.5A 89.5%
1A 89.6%
2A 89.5%
3A 89.3%

1.00mm :

0.1A 84.5%
0.5A 84.6%
1A 84.2%
2A 84.4%
3A 82.5%

0.85mm :

0.1A 80.8%
0.5A 81.1%
1A 80.9%
2A 80.5%
3A 79.6%

At 1mm the output cost per duv reduction is similar to a Zircon 804. I like the former better for the other improvements though.

There is a bit of a drop at higher current with 1mm and 0.85mm, I should have tested at even higher current to confirm that.

Color rendition :

I think I forgot to save the spectral data :person_facepalming:

with a 2700K from sofirn, original :

0.1A Ambient = 1726.1 Lux, CCT = 2701K (Duv 0.0013) Color Rendering Index (Ra) = 95.9 [ R9 = 74.5 ] IES TM-30-15 Rf = 94.10 Rg = 99.50

0.5A Ambient = 6778.6 Lux, CCT = 2714K (Duv 0.0010) Color Rendering Index (Ra) = 95.5 [ R9 = 73.0 ] IES TM-30-15 Rf = 94.00 Rg = 99.99

1A Ambient = 12251.9 Lux, CCT = 2722K (Duv 0.0007) Color Rendering Index (Ra) = 95.1 [ R9 = 71.5 ] IES TM-30-15 Rf = 93.61 Rg = 100.34

2A Ambient = 21058.3 Lux, CCT = 2720K (Duv –0.0003) Color Rendering Index (Ra) = 94.5 [ R9 = 68.4 ] IES TM-30-15 Rf = 92.71 Rg = 100.80

Sliced at 1mm :

0.1A Ambient = 2298.4 Lux, CCT = 2435K (Duv –0.0013) Color Rendering Index (Ra) = 96.0 [ R9 = 74.0 ] IES TM-30-15 Rf = 93.27 Rg = 101.86

0.5A Ambient = 9113.0 Lux, CCT = 2448K (Duv –0.0016) Color Rendering Index (Ra) = 95.5 [ R9 = 72.0 ] IES TM-30-15 Rf = 92.54 Rg = 102.25

1A Ambient = 16433.9 Lux, CCT = 2459K (Duv –0.0019) Color Rendering Index (Ra) = 95.1 [ R9 = 70.2 ] IES TM-30-15 Rf = 91.87 Rg = 102.53

2A Ambient = 28013.9 Lux, CCT = 2464K (Duv –0.0027) Color Rendering Index (Ra) = 94.3 [ R9 = 66.6 ] IES TM-30-15 Rf = 90.77 Rg = 102.79

It ups color saturation a bit (Rg).

2 Thanks

Nice dataset, lots to see happening :slight_smile:

What is your setup to measure luminus flux?

I should probably have mentionned that, it’s just a rectangular white box with two holes and a baffle, I’ve yet to make a proper sphere, due to its shape it probably doesn’t give very consistant results with different beams, but I assume the beam of a bare LED doesn’t change much when sliced, and so in this case should give decent results.

Thanks, sounds adequate :slight_smile:
Thanks again for the data, that is a lot of work. It is especially good to know that slicing right on top of the phosfor is not the best height.

Thank you for performing such a detailed set of measurements, and visualizing them to make them easier to understand for the rest of us.

Great tests here. You are now the official LH351D ambassador. I only wished to have seen the 2nd to last test Relative intensity vs angle with Zircon installed so I can re-assure myself to never buy a filter :stuck_out_tongue:

Thanks for the appreciation, it did take a bit of time, as well as for the formating, it makes me even more appreciative of maukka’s reviews with all those graphs.

Ah yes I though of doing that, but uh, more tests, more data :weary: , and now I can’t test with the same sample cause they’re sliced lol, that’s why some tests have different samples, because at first I didn’t plan about testing everything.
Though I suppose the effect of the filters should apply rather evenly on the beam independently of angle (except for extremes maybe), tint shift vs angle unchanged, and output and intensity reduced. Maukka measured the output reduction here.

This might the be-all and end-all of LH351D dedoming post.

There’s a lot of information and its all presented clearly and is easy to understand.

I actually saved the spectral data from the 2700K one, not the other one with all slicing height though, TM-30 reports :

Original : 1mm :
2700K+Zircon 804 : 2700K+Zircon 803 :

Not much change with slicing, Zircon filters improve the reds slightly.

Thanks for doing this, very useful.

Thanks a lot for this information, thefreeman. In the past I read some people argued with slicing versus chemical dedoming and bla bla bla. It works anyway (and so, my 3000K looking 1mm sliced CRI90 3500K XHP50A certainly makes my SK98 :-) throw a bit further). I still have a bunch of unused CRI90 4000K LH351Ds from AEDe's group buys, so nice to know.

Updated with 4000K from Sofirn. Curious how the tint shift vs current is more along the CCT lines whereas with 5000K it’s along the BBL.

Wow. Amazing dataset! Awesome work and thank you so much!!

The reduced angular tint shift is the best outcome of slicing imo - for high CRI emitters at least.

I wonder how much surface finish of the silicone matters? I’ve noticed differences in the past after polishing sliced domes with wet stones, but can’t quantify anything

My Wurkkos FC11 LH351D 5000k 90CRI had an atrocious green tint. I sliced and sanded the dome, but I did not take all that much off as I only had a thicker washer at the time (probably around your 1.5mm range IIRC.) Results per my eyeballs as I have no other way to test: a bit warmer, a TINY bit less green, and a bit tighter spot. I was expecting more of a difference in DUV, your measurements support my eyeball “data” in this particular example. Thank you for this information, I may try to track down a thinner washer for a deeper slice. However, I also added a Zircon filter, so may not need it. Still not quite the tint I want, but passable.

The FC11 LED is honestly the worst LH351D i’ve seen so don’t feel bad and in fact it will be worse than these measurements in the stock reflector because the above tests were with a beaded TIR.

Here is my stock FC11 and then sliced on high measure:

I’ve tried to polish after slicing before, but It didn’t seem to polish anything, felt like triyng to polish rubber or something, maybe I didn’t use the right grit size stone.

A good razor blade seems to make a very clean surface though.

Mine was arround 0.0085 on turbo at turn on, that’s the same LED on the graph, so yes the spot has a lower duv with the beaded TIR.

edit : I notice I forgot to put 0.85mm on the second graph so I added it.

Updated the first graph with 2700K sliced to 0.85mm to show how warm it can get.

Added 5700K V2Q1S2 from Convoy.

Very useful info!! Thanks so much for compiling all the data for us! Seems like shaving reduces output the same as using minus green filter and achieves the same lowering of DUV. Only difference I guess is shaving results in a more uniform beam and improves throw.

I have a sliced 2700k in a Tool AA. After slicing the emitter, I gained some strange artifacts in the hotspot, but after some sanding they disappeared. I’m not sure if this was because I didn’t do a great job on the initial slice, or because this light has a smooth reflector. I have another sliced Sammy in an OP reflector and it looked great without any sanding.