I've studied (from afar) all the mounts available on the market for large DSLRs. I quickly came to the conclusion that nothing out there would meet my own particular requirements, that most of them have glaring engineering "issues" that would prevent them from ever excelling at their role, and that (as with multirotors in general) many people skip right by the real issues and focus on the wrong aspects.
(Please excuse me if I'm sounding a bit short tonight - somebody upset me greatly and I'm still simmering
)
The gimbal has two principle roles: to isolate the camera from vibration and to keep it pointed in the right direction as the airframe rotates about its axes. Unless and until it succeeds on the first task, its performance in the second is largely academic. The engineering profession has been dealing with vibration since forever and the techniques for controlling it are widely understood. Suffice to say that none of them involve pool noodles
. The vibrations that need to be kept at bay involve a range frequencies between "low" and "medium". This is highly significant, since its extremely difficult to identify/apply a single solution that would cover that particular range effectively. So
at least two "stages" are required. Straight away that's difficult with the currently available units since they are designed as "universal" products, i.e. there is little or no accounting for what frame they're being mounted to or the amplitude and frequency range they need to cope with. To be truly effective, the gimbal must be designed as one with the airframe.
Ask an engineer to calculate a solution for any vibration problem and their
first question will be "what sort of weights are we talking about here"? Mass is extremely important - a solution that works for a 1kg payload may be completely useless for a 2kg payload. In fact, it's
much more difficult to isolate a light object than it is a heavy one.
So, assuming we've managed to block
any vibration from reaching the camera, its sole remaining task is to keep the camera "still" while the object it's attached to moves. The mechanisms employed in all the commercial units I've seen are, by definition, incapable of succeeding 100% in that task. Hobby servos have "backlash" (a dead-zone where the rotated mass is "uncontrolled"). This backlash means that every time the servo changes direction (which is obviously happening constantly) there'll be a jolt through the drive system which will reach the camera. The servo is (usually) driving a gear system or a belt drive, either of which also have backlash in them, so the problem is multiplied. Then there's the accuracy issue: most servos are very low-resolution devices (for this task)
and they can have poor "repeatability" (i.e. ask them to move and then go back to
exactly the same place and they can't).
Speed is obviously vital: speed of reaction (latency) as well as speed of rotation. If the servo doesn't react quickly enough to a position command hten it will
never be "up to date" with the stream of commands coming to it 50 times a second. Once it is moving, if it can't get to the commanded position fast enough then once again it will always be behind the game. The heavier your camera is (plus the weight of the rotated parts of the gimbal) the more power will be required to get it moving (against inertia) and then accelerate it (and stop it).
So how come the "manufacturers" don't fit the best, most powerful, fastest and most accurate servos available? In fact, I've not seen
one even disclosing/discussing the servos they provide, or offering any meaningful advice if they don't supply them. All the shiny carbon fibre or laser-cut logos are just bling if there's nothing to match it "under the hood".
I mentioned the command rate (50Hz) above - the rate at which the controller issues new positional commands to the servos. Most digital servos will accept a PWM command rate of at least 250Hz, and should be happy being driven at 333Hz. Doing so will make a
big difference, but I don't see it being discussed. Voltage, too - makes a huge difference - your servos should be supplied from an efficient switch mode power supply (at or slightly above their rated voltage), otherwise their performance (and your results) will decline as the battery source drains through the flight. Lots of power is needed, too - a pair of powerful servos can easily pull 8A peak between them. If the supply is not up to the job (like if the servos are powered only from the controller) the voltage will dip - which may account for many unexplained FC "freak-outs".
Coming back, briefly, to the frame vs. gimbal issue and the
de facto practise of hanging the gimbal under the centre of the frame: that's
not a good place to put it. In fact, it's the worst place, by far. Aerodynamic drag, displacement of the vertical CG, increased moments of inertia, mechanical stresses, crash vulnerability, redundant supporting framework, compromise to the roll axis mechanism, etc. - there's a long list of reasons.
I don't mean to be discouraging here. I guess its a question of managing ones expectations. In spite of some brutal price tags, there is no acknowledged "reference" gimbal, a standard to which all others are compared. I find that significant - it confirms an immature market were neither suppliers nor buyers are yet confident enough to pronounce a clear leader. In my opinion, no clear leader
will emerge as long as we're locked on to the current pattern.
In terms of what to buy now, the "usual suspects" are known to this group and there's plenty of discussion about how to make them actually deliver what they promise. But I'd tend to agree with Ross - build your own, at least for now. That's what I'm doing - the new CNC router arrives in 3-4 weeks
