In my opinion, the sweet spot tends to be a bit wider than 5-10, and possibly as wide as 3-12.
All editions except 4th tend to have the problem of low hit points make PC death too much a matter of random bad luck.
All editions before 4th tend to have the problem that at high levels, increasing prevalence of save or die tends to bring the random bad luck back.
Additionally, all editions before 3rd tended to have the problem that low level wizards weren't very fun to play because they had too limited of spellcasting options.
Additionally, all editions before 4e tended to have the problem that 5th level spells were such powerful game changers, that the party that lacked access to them was playing a different game than the one that did.
A related issue to this is that all editions before 4e tended to have the problem of non-caster classes with linearly increasing power and, and spellcasters with exponentially increasing power. This was most pronounced in 3e, where many of the historical restrictions on casters got tossed and high levels weren't play tested well, dramatically reducing the sweet spot in this edition.
Additionally, all editions tended to have the problem that combat at high levels became fiddly for one reason or the other. Either the problem was the players extended strategies from low levels to gain increasing numbers of henchmen and retainers, resulting in very large combats. Or players hit on the strategy of summoning large numbers of allies. Or spell-casters realized that they could very powerfully influence combat by using buffs and debuffs. Or all of this at once, which could just slow resolution to a crawl.
There are also a number of edition specific problems.
In 1e-2e, one problem was a lack of published content suitable for characters over 10th level. Particularly post UA, in 1e a 15th level party with good equipment could reasonably take on just about any published unique foe in a stand up fight. The options for challenging parties just diminished, and the few published examples usually relied on breaking the rules in some fashion to nerf high level PCs - making spells not work or be impossible to recover, making magic items cease to function, and so forth. That successfully created challenged, but only in a method that seems like 'cheating' and is frustrating for players.
Third edition spellcasting simply breaks the game. Period. Too many restrictions were removed. Too many options for avoiding problems were provided at too low of a cost. And too many spells are simply unbalanced. Depending on how quickly the player gets system mastery, this can just break the game even by a fairly low level. But even players without a lot of system mastery can break the game by 9th level. 3.5 edition, despite a few obvious needed nerfs, actually ended up just making the problem much much worse, by breaking numerous formerly reasonable spells for no obvious reason and clearly without there being much in the way of playtesting to prove the concepts.
4e has the problem that despite attempting to make 30th level play exactly the same as 1st level, the math at high levels wasn't well tested resulting in long, tedious combats, where players were often running out of encounter powers and reduced to 1e style slap fests (but without 1e style speed and short combats). Also the end results were often quite obvious long before the combat was over, resulting in pointless grinding.
4e and 5e in my opinion both suffer from too much number inflation. There has been a consistent trend from 1e to the present that each edition increases the hit point pools and damage per round, particularly at high levels. All this number inflation results in increasing math tedium and die rolling.
All editions run into problems of the scope and scale of the adventure becoming too large to be easily relatable or too big to easily cover. For example, as players become true powers in the world, the scope of their influence and possession can extend out to whole nations. It becomes very hard to then easily track what they 'own' and what resources they can draw on, and if the PC's are allowed to draw on them, you can run into situations where mass combat rules (or taxation and economic investment rules!) are the most important part(s) of the campaign. The game can quickly become Accountants & Spreadsheets. Plus, the foes of the players more and more become superhuman beings with superhuman intelligence, senses, and knowledge - and this can be a challenge for DMs to properly run, as by definition such creatures are beyond human understanding.
