I don’t know how the 351d compares to the XM-L2. It seems like maybe the specs show it as brighter. I put one in an SK68 thinking it’d be an extra bright little pocket thrower. Not much better than the stock (knockoff China at that) tiny LED that was in there. And to boot it burns 4.5 amps vs 1 amp with the stock led. So it’s a lot of juice wasted for just a little bigger beam and not much brighter.
Saving someone else the trouble just in case anyone else had this thought cross there mind.
I guess your sk68 has hollow pill?
Plus, pushing 15W through a ’351 on a thin pcb resting only on a little ledge would be cooking the crap out of the poor critter. Surprised it didn’t burn blue at that point.
So yeah, pent-up heat being so high, of course efficiency will be down in the toilet.
Limit it to 2A tops (I’d be happier with 1.4A) and it should be much better.
I leave dont leave it on for more than a few seconds (knowing heat sinking isn’t up for the challenge).
But yeah, even at direct drive 5 amps approx, it’s not worth it. Must be the lens and reflector don’t do much with what the led has to give.
Something must be wrong.
At cool start it should give you at least 4 times that lumens, maybe more than that.
I’d done something similar in an SK98 and a Convoy C8 years ago. I put an XM-L in each and the C8 was much brighter with the same battery. Both were direct drive.
I’ve had no noticeable output loss swapping 90CRI bin LH351Ds into lights with an XP-L HD or XM-L2. Something funky is happening with the particular build you have.
it’s 4.5 times more current, and you said it was not much brighter ?, this is strange, where would that energy go ?
Maybe some confusion is going on?, the OP meaning brightness as in how bright is the spot, and the responders meaning brightness as in total light output. The LH351D should produce a much wider hotspot (lots of light needed to produce that), but the hotspot is not necessarily brighter than the stock led with tiny die and narrow hotspot.
What kind of 14500 battery is OP using?
… unless its a high-end IMR battery I highly doubt it’s going to supply 4.5 amps.
Yes (djozz) wider and not much brighter.
And I do also think the reflector has something to do with it too similarly to my previously mentioned experiment with my SK98 and Convoy C8 each using an XM-L. Same LED and same battery, the C8 was brighter. Only difference was the reflector and lens.
If you want a more intense hotspot. You should go with a xpl-hi, a sst-20 or maybe even a osram white flat. In this case with a bad thermal path the xpl-hi would be the most robust choise.
at 4+ amps its instantly overheating.
keep playing and it will desolder itself.
even on dtp copper with the flimsy hollow pill its in trouble.
that tiny ledge wont cut it for more than about 1.5a.
I think djozz has it nailed.
If you take a small led making 500 lumens in a light and shine it on a wall 5 ft away and you get a hotspot of 6 inches.
Then you swap for a bigger led making 1500 lumens and shine that on a wall at the same 5 ft and you get a hotspot of 12 inches, which do you think is brighter.
A 6 inch circle has 28.27 sq inches in the circle that the light is divide into.
A 12 inch circle has 113.09 sq inches in the circle that the light is divide into.
You have 3 times as many lumens with the bigger led but you have almost 4 times as much area the light is divided into with the 12 inch hotspot.
The smaller led will look brighter and it is if measured but its putting out 1/3 of the lumens.
Its basically brightness per square inch which is lux. A throwers friend, more lux more distance.
clones sk68 cost very low but often are junk
A reflector redirects to the front close to all the side output of an led, crossing the lens and attaining a very high emitter light output extraction ratio.
An aspheric lens can only focus the led output which crosses it. When in flood, with the lens closest to the emitter, still a certain angle of emitter side output is lost which, despite output being less intense at high angles, this is more than compensated by its total area. I'll make use of a couple not so self-explanatory, but clear enough, images:
https://seos-project.eu/laser-rs/laser-rs-c03-s02-p02.html
What this means is that an SK68 or SK98 are only outputting ≈70% or little more if lucky of the total emitter lumens when in flood. Then, as you zoom and increase the lens distance to the emitter, the output diminishes even further, probably down to ≈30% or less of the total output when fully in spot or zoomed.
So, the emitter light output extraction ratio in a small lens zoom flashlight ranges from bearable down to @#$%. On the positive side of things, nothing beats the evenness of the blanket of light coming out of a zoomie as a flooder; this means they're the best for close range lighting. 
I’d say mules actually still do a better job of that, and the flat lens makes it easy to add some diffusion film to smooth out the angular tint shift.
LEP tech being combined with LEDs might finally make compact throw/flood combo lights that are actually very good possible in the future, I’m excited to see what innovation happens there.