Can you solve this puzzle?

A simple Meredith, no more, no less. Hope you like it

@Arisktotle's version

It's surprsing that white can still get a draw if black gives up its queen for the bishop. Those are the details that make a composition attractive!

A second version with more complex and thematic play. Note a stalemate occurs from both sides!

That's excellent with many pluses compared to the first version! I think a bit more can be done with the initial positions of wB, wR and bQ, e.g. : FEN 4n2r/2p3kp/2P1P1p1/2PK2PP/8/5R2/B6q/8 w - - 0 1  Perhaps it's even possible to get bQ from promotion!

Here is a version with promotion, although I think your version is better: it has more complex opening play, and I would have never expected 1. Bc4 Qd2!?+ to be a draw.

That's the advantage of engines. We humans still need to come up with the good ideas but they help us get rid of bugging variations. [edited here]

The approach in your new diagram is one of the options. The queen comes natural and the explosive starting moves are replaced with a quiet bishop manoeuvre. This is a matter of taste. I personally like the rook sacrifice but that is much harder to combine with a queen promotion. What I don't like about your setup is the e2/e3 pawn pair. It is non-functional in the remaining solution part and sort of "clutters the space". Ideally the extra material in the diagram should be removed before the final dances start.

I agree!

The version below has a promoted bQ against wR and wB and should have the best economy.

That's a good setup in the category "queen promotions"! I like that you kept .Bc4 and .e7 in the solution. You can also keep .h5-h6+ Kg8 as starting moves. There is no minus in starting with a check move in endgame studies (and the engine refutes ...Kf8)!

True, although it (like most things) are a matter of taste. I personally am indifferent.

Pawn c5 to c3 might be an interesting change.

That appears to be a nice extra! And the diagram looks better as well.

There is a good reason to add the intro move h5-h6+ to the solution if you can. Not just here but in any similar situation. The core theme of the endgame is to dominate the superior black forces with minimal white material. This can only be achieved by locking some material behind bars as in the upper right section of the board. Prisons such as that one always look a bit artificial and contrived to chessplayers since everyone knows that pieces need space. As a study composer you therefore want to "explain" how the prison was created from a more natural position with "breathing space". The move h5-h6+ is perfect for that. Even better when the white move before that could have been e.g. g4-g5+. This is an example from a  list of items you prefer not to see in a diagram but love to see during a solution like a quadruple pawn, underpromoted units, multiple pieces under attack, weird contorted positions and as mentioned the prison. So everything that aids setting up a prison during the solution is a good thing!

Btw, about 30% of all endgame studies start with a checking move. That's definitely not an issue as it is in direct mates!

That's a good point! 

I experimented some more with the position at hand, but so far I can't get g4-g5 check in. Well, that's the life of a composer, I guess

That was sort of a theoretical suggestion. Just to indicate that you can try to undo a prison to a point where nobody sees it coming! Retracting the h-pawn already does a great job in making the position look more natural.

Recently I made an endgame study where the black king locked itself in a corner prison as the only option to fight for a win - and then white tooks advantage of the prison! After one of the studies in the great series lodrac91 is posting in the Endgame Study forum.

The version with c3 is much harder to create. For example, in your version, Qd6 wins while in mine, Nd6 gives black a huge advantage.

[Edited here]

You have done an excellent job in persuading me to add h5-h6 [edited]

Whoops! Looks like I'm mistaken. Its h6 before Bc4, not after

This works

(a) I love the version with c3! Did anything make you think otherwise?

(b) IMO h5-h6+ can be added to the Q-promotion version. 1.Bc4? gxh5! (or even Rf8) kills white's ambitions.

In general I think there are probably a number of good candidates to become "the one". I sometimes gather "good versions" and sit on them for 10 years before I choose

Sorry for those misleading posts.

1. I do like c3, but It is much harder to create good opening moves for it. For example, in the top version, I had to cut out 1. Bc4 g2, which I really liked.

2. My bad. 

3. Good point!