Ive never understood the fascination with con based casting in d&d.
It's not just D&D. "Blood magic" and other such themes are popular with a lot of people, because it feels thematically strong. But, well...
My group tried some homebrew HP casters back in 3e/3.5 and it never ended well. Either too much of a death spiral, or the cost was so minimal it was incredibly overpowered.
This. As with most design conundrums, the problem is that this creates an extremely narrow balance point which may not be replicable across different situations. That is, the design space is dynamically unstable. Costs too high, characters just die trying to be normally effective. Costs too low, there's no actual risk, and so you've just been great power for no
actual risk.
4e did (sort of) solve this problem, with its Vampire class, but I find that for at least a portion of the fans, the fact that it IS a problem is what makes them interested. As soon as you solve it, you take away the "ooh I'm being baaad and breaking the game!" feel, which means even though you're casting from HP, you're not
actually any stronger than someone else with similar optimization skill who doesn't do that. I knew a couple friends back in WoW who felt that way about the Warlock, it was specifically the
power of being able to turn damage into healing and health into damage that drew them in, if it had been properly balanced compared to the far more fragile and risky Mage class, they would have lost most of their interest.
Ironically, 4e also had an
actual Con caster, the Elementalist Sorcerer subclass. And it was quite good! Actual simplicity in a caster class. Pick your element. You have now made one of the like three actually relevant choices (other than feats) you will make for your character's entire career, congratulations. It was as simple as the Champion is simple, and yet better-designed because it didn't fall behind other Sorcerer subclasses.