It is absurd to assume that because the 600 and 650 are similar that it did not take massive R&D $$$ and effort to produce the 650. Or that even though the 580 and 600 have the same drivers that the same doesn't hold true for the 600. I work in R&D at a tech company and can tell you first hand the work that goes into prototyping and pre-production, but I'll try to apply my experience in terms of upgrading a headphone model line.
Engineering and design is not an off-hand endeavor. You meet and discuss and argue for a good amount of time before you even settle on parameters for an even incremental improvement. What do you want to achieve? What is the end result? Can we improve on the existing design and if so how? Or do we need to do a blank-sheet design? The difference between an HD580 and an HD 600 might seem minimal to you, but it most certainly is not. The most obscure change in the shape of the enclosure will surely have drastic consequences to the sound of the item. You don't just change the shape for aesthetics. It serves a purpose. And rapid prototyping is does not work in this instance. It only works for shape, form, and aerodynamic work. Doesn't work for materials testing and so forth. So each prototype is hand built. Changing to carbon fiber? Hypothesis: Carbon fiber will reduce resonance. How do you prove this? Well, you experiment with density, shape, composition, etc, until you come up with what you think is the solution. But it might not be. So you hand-make another prototyple part. On and on and on and on. And then on top of everything else you have to make the end product pleasing to the eye of the consumer (subjectively or course, even Apple products have their aesthetic critics). So you have to accomodate industrial design engineering as well as material, electrical, and audio engineering. Next you have to take mass production into account. So you have to adjust your design further to accomodate mass production while producing the required end result. And we haven't even produce tooling for the plant yet. Do we have space for it or do we need to acquire physical plant or sign a co-pro deal with another company? Then deal with teething problems, QA issues, marketing, sales, so on and so on...
Still with me? And this is something as inoccuous as a pair of dynamic headphones. Try putting 300 million transistors on a piece of silicon the size of your pinky fingernail...