It's time for a triple popcorn Fluxer update - an update so wordy I needed to split it into two parts.
In the next two posts, the Flux Deluxe's MOSFET saga takes a few extra twists and turns - and my write up is extra wordy, too - but I finally (!) determine a true root cause of my MOSFET issues and reach some conclusions about the path forward. It was a full weekend, so pack yourself a fresh bowl and settle in for a long-ish update.
Part I: New PCBs vs. Old Issues
As I mentioned last week, I redesigned the FD's PCB to add some copper heat sinks near the MOSFETs, on both sides of the PCB. I wanted to evaluate a few potential changes, so there are several different versions of these new PCBs coming. The differences between them are very minor, however, and their heating circuits are identical, so any of them should be capable heaters, and all include the latest fixes, etc.
The first two versions of the new PCBs came on Friday, and that evening I built up a few of the new boards for evaluation. The results were very revealing, and forced me to reconsider some previous decisions.
Or as I put it at the time:
Fuck! 
Which is par for the course around here...
Surprise PCB revelation #1: The new thermal dissipation/heat sink areas I added to the circuit boards work very well. They transfer heat even better than I anticipated they would.
Surprise PCB revelation #2: They did NOT, however, solve the issue of the MOSFETs killing themselves. If anything, the new heat sinks made things even worse! The new PCBs were
extremely efficient MOSFET killing machines. Which, y'know, was not part of the game plan. What fresh hell was this?
Not wanting to accept these results at first - and being a stubborn person - I spent the next several, very frustrating hours confirming all of the stuff I just mentioned: the new PCBs
were performing exactly as I expected them to as far as their heat transfer capabilities - they transferred
a lot of heat to the new copper surfaces I added. But that add'l heat transfer didn't mitigate MOSFET failures, and if anything it exacerbated them! WTF was going on?! The observed voltages remained good throughout these failures, BTW. So if it wasn't voltage and it wasn't heat, then what element was I actually fighting here? What was my true enemy? It had to be something pretty obvious, as the new PCB was killing the NVD5890L MOSFETS at an alarming rate.
Thinking hat time.
I looked at my clues: The new heat transfer areas of the PCB were getting hot to the touch - which I took to mean they were working as I expected them to work - but the MOSFETs and the areas around them became
extremely hot, very rapidly. And the MOSFETs were dying, too, also very rapidly. Each test was taking 10+ minutes to setup and less than 3 seconds to execute, which was also getting old. I needed to figure out what was going on ASAP, before I ran out of MOSFETs and/or patience.
When I considered all of the above evidence, it looked like the MOSFETs
themselves couldn't dissipate power quickly enough. The new PCBs allowed the MOSFETs to offload a considerable amount of heat (meaning they could run harder, as they has less thermal resistance), and now they were now running so efficiently they practically sizzled - and then died. Hmmm - this wasn't a problem of shedding some "extra" heat. This was looking like a straight-up miscalculation. Just how much power was being pulled through these MOSFETs by the 14mm coil?
I danced all around this question while I was evaluating the voltage spike situation, but for some reason I didn't ask this
specific question, and I should have. In retrospect it was an obvious blind spot.
I took some fresh measurements and then re-calculated the heater's
actual power output with the 14mm coil. I then compared those measurements to the datasheet (specs) for the MOSFETs I have been using.
I observed that the 14mm coil is capable of pulling quite a bit of power (i.e., current) through this circuit: ~9.6A, which is consistent with its performance.
The spec for power dissipation is given in Watts, and Wattage is calculated by multiplying Current X Voltage.
The heater's voltage is nominally 12VDC, but a fresh set of batteries is actually a few mV higher than that. 12.6V is probably the highest state of charge these batteries are likely to exhibit, so I used that value for these calculations. It's the power level one might encounter with a fresh set of Sony VTC5-D (high current discharge) batteries.
Using that figure, the math says that a 14mm Flux Deluxe is capable of pulling about 121W of power through the heater MOSFETs
in about an instant. Which is a lot.
NB: This is a reminder that 18650 batteries pack a lot of amps, so treat your batteries with respect, and inspect your heaters regularly for potential issues! Silicone wires nick easily. If you notice anything that needs attention, please drop me a line. I am happy to work w/you to fix potential safety issues.
To this point, I have been using the NVD5890N as my heater MOSFET. This is the spec sheet for the NVD5890N, and I've circled its Power Dissipation limit in red:
The NVD5890N MOSFET is rated to dissipate a maximum of 107 Watts.
As I just showed, a Flux Deluxe w/a 14mm coil is capable of pulling about 121 Watts.
![:\](/i/s/hmm.gif "Hmm :\")
I am over the power dissipation limit of these MOSFETs by >10%. That's obviously not a good place to be,
but nonetheless, I am grateful to discover this, as it's an answer that explains just about everything! No device performs well at its limits, and when you exceed a device's limits by >10% - and
especially when that limit is something critical
like the fucking power dissipation limit of the MOSFETs!!! - you are going to to see some bad things occur.
This is
exactly the sort of smoking gun I was looking for and hoping I would eventually find. I am thrilled I finally found it! Hindsight is 20/20, and this sure seems obvious now, but I somehow kept missing it until I had cleared away all of these other issues that I mistook it for.
So this is it, folks. This is the root cause of the Flux Deluxe's MOSFET issues: I am drawing more power through my MOSFETs than they can dissipate. THAT's what's been killing them. 
I am not proud of this oversight, but I am SO happy that I finally (!!!!!),
positively identified it. This is the answer I was looking for, and it was right in front of me the whole time. D'Oh!
Thankfully, this is an easy problem to address: there are
many other MOSFETs out there in the world, and some of those have power dissipation capacities that can easily handle the power demands of a 120W device. And you can even search for them by their desired specifications.
I have been using the NVD5890N MOSFET since
@stardustsailor recommended it back in October, and it is a
great chip for an IH; it's been one of the secrets of the Flux Deluxe's quick performance. I was so impressed with the NVD5890N's performance that I have been taking it for granted that this chip could handle whatever I threw at it,
as it was a beast! 
Unfortunately, even in Beast Mode, it tops out at a max of 107W, and I am regularly exceeding that. No wonder it's dying! The wonder is that it is working as well as it is for so many of you!
Continued in Part 2: A New Chip in the House
It took a whip and a chair (and the proper TVS diodes) to tame those chips.
