Building a Miniature Gyroscopic Stabilizer for Video Camera?

try this

http://www.google.co.uk/search?q=gyro+stabalised+binoculars&sourceid=ie7&rls=com.microsoft:en-US&ie=utf8&oe=utf8&rlz=1I7GPCK_enGB411&redir_esc=&ei=y3YYTeqDBYPOhAfF-rG3Dg

These are really interesting, though maybe a bit pricey. Of course the R&D is done and that's worth a lot. I'd hate to take a new pair of these apart for the "fun" of doing it, but maybe I can find a salvage unit to explore. Thanks!

The 2 easiest ways I would recommend is attach a small pole to the bottom, and add a small mass on the end. If the rc aircraft can support the extra load, it should provide some stabilization, which you could easily alter.

Another way would be to get (or make) a gyro, just a heavy spinning wheel with an axle which is then connected to the motor. Problem with this is that it uses up a lot of space, and could also be quite weighty.

This is a gyro for a helicopter

Or this

Can you tell us more precisely what you aim to do with this? Stabilising shake on a hand-held camcorder or still camera has largely been solved by the manufacturers anyway. If your intention is to keep pointing in the same direction while the holder goes every which way, then you will need a gyroscope arrangement. Mechanical gyroscopes are heavy, and would be completely impossible in a model aircraft. I suggest you investigate the automatic pilots which are available for the inexperienced RC modeller, such as this one from Futaba http://www.futaba-rc.com/radioaccys/futm0999.html which works by sensing the difference between ground and sky. If you want to work in the dark, then you will need the direction-stabilising capability of a helicopter gyro (which is actually a rate sensor).

For RC planes, you could always get an expensive radio and system, and put a few gyros on the plane, I presume like the auto pilots, but you fly the plane whilst the gyros keep it all steady.

Many thanks for this and the many other emerging replies to my post! These are very helpful and point me in multiple, useful directions.

To be more specific about the stabilization concept: The goal is to stabilize a lightweight (approx. palm-sized) camera w/optical zoom, given compound/multi-directional vibration and jostling. (Imagine hand-holding or rigid-mounting/gaffer taping the camera while/for riding in a car, on a train, in crowd, while running, &c.; occasional mono-pod use is a possibility.) A small, unobtrusive form factor is important, so consequently the typical mini-steadicam solution becomes too awkward, also reducing options for low angle shots (too little ground clearance and just too big and "obvious" for tight situations). Low cost is also a goal.

At this point, my primary questions stem from ignorance of how the mini helicopter gyros are actually designed. Are they essentially 1- or 2-axis? If I should want 3-axes (XYZ space), do you think one or more units could be ganged together using a unifying frame for everything? I have no idea what range of vibration frequencies would be damped, but I think empirical test would be the easiest way to answer this one.

You do not need x,y,z stabilisation but the related angular movements.

In any usual moving or vibrating situation the angular movements result in much bigger image movements than the linear movements.

First question to be solved: which bandwidth?

Any stabilisation system has a limited bandwidth: from milliseconds to months,

how long shall the image be stable to within which angular limits?

If you can live with seconds there may be a possibility of passive stabilisation by flexible mounts.

If for longer periods you will need a gyro stabilisation.

Gyro: depending on type 2 (2axis each) or 3 single axis needed. Military systems are near or above 100K$ and heavy. So no chance.

Miniature gyroscopes are often Coriolis devices: no true rotation.

BOSCH has some for stabilising the steering of cars (after turning upside down the newest Mercedes this became a self-selling item).

You will need an angular decoupling of your camera from your support system (car, hands ...) Torque motors in good ball-bearings are clumsy and costly.

A flexure mount as the below 2-axis flexure hinge can be simplified to "do it yourself"-construction that is realised by brazing. (Welding, gluing and clamping is possible but will not achieve good quality). Welding is distorting, gluing and clamping is severely limiting the load capacity of the flexures. Design shall be done so that equal load condition and equal stress condition be met under any of existing load directions!

Have success - this design usually is a task for an engineering team of 7 qualified MEs and EEs and ... for 1 to more years!

RHABE

Dear Rhabe -

Well, I am the first to admit my lack of qualifications for this discussion. But here are three questions (and - sorry - subsidiary questions) that come immediately to mind, after reading your comments and looking at the image of the 2-axis flexure hinge you attached.

1) Can I assume that the second axis is treated in roughly the same manner as that shown in the quadrant displayed? (i.e. -- If this quad is at 6:00, can I assuming it is structurally "mirrored" at 12:00? ...and that aside from the different mapping of what I'll call the isolation slots, the that 3:00 and 9:00 quads are mirrored as well, and similar?)

2) Could a related (and useful) hinge structure be prototyped using a 3-D printer and some particular type of thermosetting plastic? (You mentioned brazing, but this assume various alloys and temper.. yes? ...and therefore a piece that was fabricated from components that are both mechanically & chemically machined? )

3) Are there CAD drawings of such "hinges" available for download? (.DXF or .3DM?)

I want to note that I have no commercial intent in mind. I'm working on this project with my grandson, who is 12 and whose imagination is way ahead of mine... )

Having just looked at some of the flexure hinge "math" in the literature, your caveat is surely validated!

Thanks for your thoughtful and engaging post.

1) Mirrored: ok, yes, but there are different mirror planes to create a true 2 axis hinge.

X- and Y- hinge rotationally displaced 90 degrees (usually).

If you turn the imaged side of the hinge by 90 degrees you would see the same structure of the hinge but the upper slot will then be a lower slot.

2) 3Dprinting - I never tried, maybe working. I would try sawing and filing in steel or aluminum and used saw-blades as flexures screwed at the ends. Brazing does not need a temper if you are content with medium load capacity at given deflection stiffness. Ball bearing steel would be best but any other with a good strength would be good. If you use an air-hardening steel you will get considerable good quality after brazing.

3) CAD drawings - no.

Math is not really complicated if you avoid buckling loads/designs.

Just calculate for bending stress and shear stress and estimate buckling. Estimate an allowable stress level, make equal the load in x and y (the axis directions) and make equal the stress and the von Mises stress from shear. (Tension is not accounted for as tension is not at stress limit). Put these equations together and you will get as input the material data, the load is determined by your requirements. Output: geometry can be calculated from these and deflection stiffness and any of the linear stiffnesses subsequently. Math is not complicated if you stay with constant cross section or as in the photo with end-thickness twice the mid-thickness.

With brazing it is necessary to realise the Bendix-type cross-flexure hinge. With edm or 3D-printing you can stay with the shown type of flexures linked to each other in the center.

Have fun and a Happy New Year

RHABE

OK! This is helpful, too! More questions:

1) Are the "arms" of the "X" also attached/fused at the crossing point, or do those diagonals simply by-pass each other?

2) Is the cross-section of each "hinge arm" twice the thickness of the center point in both X and Y dimensions? (i.e. A 4-sided pyramidal form having a slightly curved "thorn-like" attenuation as it approaches the center)

3) "Saw blade" implies a source of tempered steel. Are you also suggesting that the tapered contours be ground from blade stock, as opposed to assuming a uniform -- albeit very thin) gauge is OK?

Thanks, yet again.

Replies:

1) Fused at crossing point: in the photo-example yes but not necessary, if not then more space is needed. "Flexures that pass by" is known as Bendix-cross-flexure hinge although much older than the "invention" by Bendix.

2)The hinge arms = flexures have constant width (nearly equal to length) and nonconstant thickness if optimised. Optimising against bending loads (in direction of hinge axis) results in end-thickness equals twice the mid-thickness. 4-sided pyramidal form would be good too but requires much more effort to realise.

Nonconstant width and constant thickness would be a good combination too - I never tried. End-width would be optimum near 1.4 times mid-width.

3) I would start with constant thickness and constant width. If volume (space) is not an issue then I would switch to contouring the width but stay with constant thickness - much easier than the other way. If large angular deflections are wanted then constant thickness may be better. See below example for a 1axis-hinge. 4mm from axis to end, 40µm thickness, 2 hinges in series (not a too good idea but sometimes a necessity. Photo is from first example so some thickness nonuniformities.

RHABE)

Dear Rhabe -

Thanks so much! Very helpful. Of course one answer always leads to more questions, if you can stand another few...! I'm sure it's obvious, but I am wondering what part of this device gets "anchored" and how. I other words, where is the vibration source, as opposed to the damped "load"? ...and on which axes?

TIA, Gipfel

Here is a better photo where you can see the two hinges in series connection and the 2 holes where to screw to baseplate and rotatable part. The lower "triangle" has no separating slot so connecting the front hinge to the rear hinge. Two of these in some axial distance needed for 1 axis to prevent forbidden torques. Only forces (x,y,z) and torque around axis are allowed.

RHABE

Thanks so much! I have a much better sense of this device/system now.

Happy New Year!

Thanks too, for this info. I'll check out the Futaba options! In the back of my mind is the weight of a WWII torpedo gyro I once had a chance to play with... That form factor is out of range!

Why don't you develop just an electronic correction of the image so it would be stable under any 3D movement? It is mostly used in professional video cameras and easy to develop.

This is an engaging idea, but I suspect it's well-beyond my tech abilities! If I understand your thought - you are suggesting stabilization in "post" -- i.e. - after shooting and before editing. Or are you referring to another form of EIS at the time of data acquisition? I'm open to any and all approaches! Thanks!

the granbdson of the inventor of the autogyro (Like helicopter) originally from Spain -family moved to Cuba then to Miami has several patents on gyroscopic camera stabilizers. His company is called or was called Scitech-may still be in existence. His name as his grandfathers is Juan De La Cierva- may be able to look him up or look up his camera stabilizing patents on Google Patents. I believe thatyou are allowed to do a copy of patent for personal use- need to check with a lawyer. If not get a Hold of Juan very nice affable bright engineer-he also invented a VCR editing machine and guidance system as well as many helicopter stabilizing devices world class expert in gyroscopes

Thank, you! Will look into this...

Hello Gip,

Go visit www.rcgroups.com and seek out the aerial photography forums.Lots has already been accomplished in this area.

Skies.

Jay

Thanks for this lead, too! Have started exploring. Certainly found some impressive in-air stills!

Would the Steadicam approach be useful? These just use counterbalance weights to stabilise the camera but I've seen film of cameramen running with the units & obtaining steady video results.

Thanks for this thought! The steadicam concept circumvents many nettlesome constraints and solves many of the problems at hand. There are however, some nagging issues such as 1) Limited ground clearance and 2) The typical size/scale of the steadicam geometry is workable with free space, but tough in an unfriendly crowd!

Best regards!

I'd research MEMS chip accelerometers like the adxl312

http://www.analog.com/en/mems/low-g-accelerometers/adxl312/products/product.html

and build from there. These are tiny sensors which might fit your application. If the motion displacement is relatively small, voice-coil motors (CD laser focus) should be the smallest and lightest actuators that could provide stabilization. Good luck.

You will need 6 good accelerometers and 3 pretty good integrators to get rotation rate around x,y,z.

Then amplify and feed back to torque motors.

Gyros are much better as a gyro is directly measuring rotation rate.

Inertial navigation systems use gyros to measure (incremental) rotation rates frame (aircraft or else) to Earth. With these update the coordinates frame (stored in some computer). Then transform the data from the 3 accelerometers into the Earth (or stellar) coordinate system. Then integrate the accelerations. (This is allowed only in an "inertial" coordinate system). First integration yields velocity (north, east, up). Second integration yields distance from start point. Satisfactorily accuracy is 1 nautical mile per hour of flight. Then the pilot will find the airport without much further support.

RHABE

OP seemed to only need image stabilization and not an accurate long range inertial navigation system. MEMS rate gyros/accelerometers and low cost PIC microprocessors can satisfy some of these less critical applications.

One example is proposed here www.waset.org/journals/waset/v42/v42-109.pdf. Similar configurations are used to measure flight data and trigger apogee parachute recovery systems on High Power model rockets. Various hobby assemblies, like a 2-axis linear accelerometer plus 3rd-axis rotational rate gyro MEMS chips on one board, can be found with the right web search criteria.

Thanks, but a=dv/dt , v=dx/dt, and the reverse integrals are already old friends. ;-)

Note: Links are provided for informational purposes only. No connection to or endorsement of any products.

Hello, mjb1962853!

As I mentioned in my reply to RHABE, this dialog is catalyzing lots of materials for me to think through and to synthesize. I'll pursue the links you've suggested!

Many thanks to you and all those who have posted.

Happy New Year!

http://www.waset.org/journals/waset/v42/v42-109.pdf

You are right, image stabilisation would suffice.

But be cautious, depending on type of movement this may actually require angular movements to be stabilised. I agree that full inertial navigation would not be suitable but depending on accuracy and bandwidth a simplified system (cheap components but the same system architecture as in grown-up-systems) may be a good idea.

In the article of above link there is an important flaw in chapter E: calculating tilt angles with respect to the horizontal plane (helicopter-referenced) without linear acceleration correction and without correction for x and y-tilt. This will work only in an helicopter that flies perfectly horizontal at any time and flies without any horizontal acceleration. So this article is a fine paperwork about principles of sensor use. Maybe the authors are not allowed to tell more about a real system, maybe there is no real system.

Have a happy New Year 2011

RHABE

Dear RHABE!

Thanks too for this extension of the Gyro-Stabil thread. Although a bit too nuanced for me, combined with the reply this post provoked, an increasingly viable synthesis appears to be emerging! I'm learning a lot...

Regards & Happy New Year!

I just wanted to find out if you have ever used a Kenyon gyro?

This is where you can find out more --- www.ken-lab.com

They are the only true Mechanical gyros made at the moment for the use you suggest.

Also something to check out is the steadicam tango --- http://www.tiffen.com/steadicam_tango.html

This can provide better options for perspectives.