Well, the head has a fly height of less than half a microinch (a human hair has a thickness of over 2000 microinches). I wouldn't say that "Displacement of the disc axis would require that the bearings have a significant amount of play."
Drives can take "shocks in the tens of g's" because the disc surface has a protective layer in case the head touches it. Actually the head is protected by the slider which keeps it airborne.
If you mean it would actually be able to target data as usual "during these same shocks in the tens of g's" then you're dreaming my friend
I'm thinking that if the head operates continously under strange conditions the slider might now and then get discreetly grinded by the disc over a long period of time. One day the slider has been grinded away enough for the head to be exposed and when that happens the head will be destroyed on impact, so much for happy data retrieval days
Please take another look at my point also:
Horizontal = ONLY arm y-axis gravity
Vertical = ONLY disc gyroscopic precession
Angle = BOTH arm y-axis gravity AND disc gyroscopic precession
An angle mount will catch both forces simultaneously which might create unstable operating effects as described.
I am aware of the fly height of the head. This does not change depending on the orientation of the drive. It is determined by forces larger than that exerted by gravity, otherwise merely turning the drive upside-down would render it unoperational.
I stated that "Displacement of the disc axis would require that the bearings have a significant amount of play." I did not state that the axis does not move. I believe everything I posted was merely an easily verifiable fact. I did not write with any particular conclusions in mind, so as to be objective. And by significant I do not mean large, or even perceptible on a human scale, I mean large enough to be a factor. For example, there is a myth that the coriolis effect determines the rotation of flow in a drain, or toilet, or what have you. In reality, this effect is so tiny that the rotation is much more sensitive to other factors.
I assume the "protective layer" does not lose its protective properties when the drive is mounted at an angle, and that the slider's durability remains the same regardless of drive orientation.
Let's say the head and arm assembly has a mass of 10 grams and the arm length is 10cm. The torque produced by gravity has an upper bound of 0.0098Nm if we also assume all the mass is concentrated at the very end of the arm. Let's say the axle length is 2cm, and the weight of the rotating assembly is 1kg, concentrated at the center of the axle (a reasonable estimate for center of mass). Then torque on the axle is 0.098Nm at most. The arm torque is inversely proportional to the drive's angle from horizontal, the axle torque (which would be the source of any gyroscopic precession) is directly proportional. Thus, as one rises, the other falls. Note, however, that the axle torque is an order of magnitude larger than the arm torque, so the sum is dominated by the axle torque over the majority of the range of possible mounting angles. Thus any combination is no more significant than the maximum torque on the axle alone.
Now let's say the rotating assembly does indeed exhibit gyroscopic precession. The frequency of precession for a 7200rpm drive will be about 1 full rotation in 5 seconds. This is based on my grossly exaggerated torque figure, the real frequency is likely to be less than half that. I would show you the math but it's long and I hope you don't think I have any reason to lie to you
(though if you really insist I can post it). Let's say the radius of precession is 1mm (meaning, the bearing allows the end of the axle to be 1mm off axis). Pick a spot at the very edge of the platter. Deflection at this spot is around 1cm (peak to peak), once every 5 seconds. Already, I can see that 1cm is a ridiculous overestimate, but let's go on. Pick a spot on the very edge of the platter. Average magnitude of its y-velocity is then 0.002m/s. Its average y-velocity is zero, obviously, otherwise you will not find your drive where you last left it. The position of this point over time can be described sinusoidally, so with our average velocity figures, we can roughly estimate the maximum velocity achieved and we know the period over which this velocity increase occurs. Let's overestimate again and say y-velocity goes from zero to 0.01m/s in 1.25s (which is 1/4 of our period of rotation, the domain over which the wave goes from zero to peak). This is 0.008m/s^2 of acceleration. The head already counteracts acceleration of 9.8m/s^2, and can deal with 9.8m/s^2 in the opposite direction as well. This is a few orders of magnitude larger.
Even in this horrendously overestimated worst case, the changes brought about by the combination of gravitational and gyroscopic forces do not appear to be significant. If you disagree, please show me where I have made a mistake or invalid assumption. I'm sure there are bigger geeks out there than me (I hope...) or at least people much, much better with math and physics.
My hat goes off to you, sir.
Likewise, thank you