Yaesumofo,
your numbers somehow make sense and then not so too
That a "J" Vf draws more current at the same voltage from a bench supply than a "K" makes a lot of sense - by design - this is what the "Vf" actually is telling us. For the same current draw it will need a lower voltage.
I´m a little confused with 4 voltage numbers and three current draw figures though, typically the current draw should increase with voltage, some of your number look different as if a higher voltage gets you less current ?
My experience with measuring quite some stars lately is that within each bin the voltage can and will vary, sometimes even outside the expectecd range, while this seems to have become rather rare with the new binninbg system.
LEDs with the new binning system overall seem to be more "EQUAL" than before I was told from people receiving the newer ones and had more experience with the old ones too. My experience is that out of blocks of the typical 8 attached stars you will see a tendency either to the lower, middle or higher end of the specifix Vf range, while still within those blocks you´ll see differences.
Looking at the voltage/current curve as documented by Lumiled even a 0,1V Vf difference can and will result in significant current draw differences of 100ma or more - so a LED from the lower end of a Vf range can draw 200-300ma more than one off the higher end of the same Vf bin code at the same Vin, from the lower end of one Vf to the upper end of the adjacent bin code (like J and K) this could be a difference well beyond 500ma at the same voltage - and this seems to be exactly what you are seeing.
I think this is the reason most people would prefer current regulated circuits which are equaling out this difference against voltage controlled or direct drive setups. While you will always get different current draws and subsequent different brightness (not taking brightness bins and lumens/watts into account right now, just raw watts vs Vf) with voltage regulated and direct drive setups - but you will have similar power consumption but differing runtimes when using current controlled circuit setups then. The only flashlights sofar which does even take the final brightness into account are the ARC4 and the upcoming EDCExtrem and EDCUltimate which the ARC4 designer is bringing to market soon.
As for voltage controlled and especially direct driven applications it is very advisable to check each and every LED for the actual current consumption to have no risk of overdriving it but also a good enough driving power for good brightness. This is quite a challenge without a current controlled circuit as seen lately with failing flashlights from CPF modders as well as commercial companies starting to use Luxeon LEDs - especially with direct driven or only resistored designs - and as some of the more experienced hacks on this board would say - its the nature of the beast and you have to live with it.
I hope I could be of some help
Klaus