I was working on my magic item chapter and, as expected, pretty much copied over the traditional charges structure from ye olde D&D. A lot of items have charges, and you spend one or sometimes two to make them work. But after a session or two where my players received their first items with charges I remembered that those rules annoy me. I find two issues with them:
1) They introduce per-item tracking. Each item that has charges has to be tracked.
2) I like the general idea of magic being mysterious and unreliable. However, a fixed number of charges works against this. You can partially mitigate this as the GM by taking over all charge tracking yourself, which leaves the players never sure as to how many are left but still requires that they be tracked, with the burden shifted to the GM. It also only adds an element of mystery for the players: the items themselves are still operating on fixed charge amounts.
So what I've done is gone with a roll-based system. Whenever a charge is fired off from an item, the user rolls two dice. On double 1s, the item fails to function because it's out of charges.
Which dice you use depends on your inclinations:
1) If you want something that varies in a way approaching canon charge amounts (per the DMG, staves have 20-25 charges, rods 41-50 charges, and wands 81-100 charges), roll 2D4 for staves (1 in 16 chance of failure), 2D6 for rods (1 in 36 chance), and 2D8 (1 in 64 chance) or 2D10 (1 in 100 chance) for wands (whichever you prefer). For staves, this is a little less than canon, but not by a whole lot.
2) If you want something fast and simple, just use 2D6 for everything. This boost staves a touch and shorts wands in particular, but I always felt that wands had too many charges. I favour this approach.
For when an item requires the expenditure of two charges, you can just roll each time for each charge expended. However, I like to treat this as more demanding on the item. As such, for the use of two charges, I have the user make a single roll, but replace the
two dice with one die that's a step higher. So if you're using method
1, a staff rolls 1D6, a rod rolls 1D8, and a wand rolls 1D10 (for method two, everything just rolls 1D8). On a 1, the item fails as normal. If using three or more charges, re-roll the die once for each charge past the second (or add another die per charge used—same thing); any result of 1 exhausts the item as normal.
If you allow items to be recharged in your campaign, then it becomes a binary process: items either work or they don't, and instead of adding individual charges, recharging reactivates an exhausted item's ability.
I like these charge replacement rules because they remove bookkeeping, which is nice, but they also introduce a point of drama: everyone watches when a charge is fired off to see if that disastrous roll comes up. It's also readily configurable to your own campaign, as you can see: just alter the dice rolled to get the particular odds you want, if the odds given above don't quite suit you. And it adds a bit of that fickle, capricious status to magic items that I enjoy.