On my 3rd head. One crack was user error, the other 2 just appeared from nowhere. I have a stainless laser cut ring and some knurled screws on it now. Those lift and turn things are gash.
Leaking cells were the main cause of loom rot. I had a 100% failure rate of AP14s. Not had a single problem since switching to the 16s (or the N@90 ones).
The AP units really really need another round of development to improve the engineering on them. TBH if I didn't have to have one to teach on it then I'd probably be diving something else. Things that need work:
Cracking heads.
Lift and turn things (try moving them with cold wet hands).
Stainless screws mashed into the threads next to the mouthpiece. If one of my apprentices came up with that they would get a telling off.
That case. It's buoyant, too big, and cracks in normal use. Just pay whatshisface for the GBox design and use that instead. Or make the can structural.
The harness. Too many clips and dangly bits and all the D rings are in the wrong place (and are not adjustable).
Wing weight pockets. In the wrong place, make the wing deform when you put weight in them.
Back mounted counterlungs. Get trapped between the case and your back so when you inflate the wing on the surface, you can't breathe.
BMC dump. Just extend it with a clip to the wing D ring as standard. The human body doesn't have enough elbows to reach it.
Lungs you can remove easily for cleaning.
The "not quite international" AP QD fittings. Everyone else in the world uses a set standard. Why is the AP one just everso slightly different?
Manifold bar. When you are machining it, please put a bigger flat on it then you can use any hose not a special one.
Batteries that you can replace without stripping the whole unit down.
Engineer out the spacer ring.
Engineer out the external scrubber dip tube.
Engineer out the injection ports on the counterlung and put them onto the T pieces instead.
User configurable startup menu.
A timer that is actually useful (only counts up when diving so you can measure the scrubber life properly). Having it count up when it's on is useless.
Change the crap handset strap for 2 bits of bungee.
Change the internal cell wiring so that you get sacrificial loom parts that are user changeable and aren't a birds nest.
Solid state O2 measurement.
Change the cylinder valve slugs to brass instead of plastic that go stiff with use.
Sell Sofnolime in small cans where you can get 2 fills out of one can, not 1.9 fills.
A head up display that works like a NERD.
Reposition the cells in the head so face of cell 3 doesn't get wet.