Sometimes good enough for one is not good enough for another.
If it makes an improvement it's a judgement call as to if the amount of improvement is worth the extra cost/effort; but how much are you willing to spend for zero improvement? This is not a matter of 'good enough' vs. squeezing out the last 10% of performance... there is nothing left on the table to pick up. It's not a matter of that last bit of performance not being worth the cost.
If 94 octane gas is good for a high compression turbocharged car then it should also be good for a 90hp 1.5L economy car too, right? That's the kind of faulty analysis of the data I'm talking about.
Once you have a MCPCB with no dielectric layer, and flat surfaces on both the MCPCB & pill, and some type of paste in the middle, it will not be any better in any way if you go beyond that. The boost to whatever you're imagining just is not there to be had! Doesn't exist!
Miracle paste vs. cheap paste, brass pill vs. aluminum pill vs. copper pill, direct-thermal MCPCB vs. emitter mounted direct to a pillar, none of those things will give any measurable change in the light output. For whatever reason you don't believe me, but this is not just my opinion, it's what has been proven in real-world testing of actual parts. I really suggest you gather up some of these parts and do your own testing, since you're not getting this for whatever reason. Please do test it, in no way am I trying to stop you from discovering anything new. It's already been checked and found there's nothing there once you go beyond a certain level of prep work & parts used.
The margin available is completely untapped in these forums.
Yes, try water cooling out and come back with the results please.
Why not put an XM-L2 onto an already existing water cooling setup for a CPU and see what it does? Anyone done that yet? At least it would tell you if there’s any huge benefit. I am kind of doubting it will be enough of a benefit to be worth it. There’s always an upper limit, no matter what, but finding it can be something very individual and necessary. Sometimes one has to just try it all, to realize what the limits are and that’s all part of modding.
The difference between modding and real life, is 6 amps.
I like the questions you're asking NightSpy. I probably am missing some critical info because I don't think we know the limits for sure yet. The fantastic work performed by djozz and Match is much better than anything I could do, but the testing I have seen has been with very small samples and has sometimes been limited by available resources and imagination of the possibilities of the time. For instance. Match tested directly on copper like you propose in the OP, but he only tested up to 6+ amps. There are folks now pushing xml2 on copper emitter bases to around 9 amps now.
I think it is good to have people with different perspectives and knowledge bases to ask these questions and possibly design new tests or approaches. I hope djozz drops by to share his thoughts. He is by far the most qualified person I know of in this area.