Neills DIY Gimbal Design

BIGTACO

Member
So I decided to design and build my own two axis gimbal. This way I can tailor it to exactly what I want, save some money, and have some fun doing it. I start getting anxious if I haven't built something for a while so here we go again! Anyways, I took some of my design cues from the ever popular and proven "Askman F550 Gimbal". My gimbal uses the same reduction ratios driving the rotational axes. The structure, geometry, materials, and tolerances of my design will be different. I plan on making this a very rigid design with zero axial play and very little radial play along rotating axes at all bearing assemblies so camera movements are smooth as possible. The structure will be made of 2mm thick carbon fiber plate separated by aluminum standoffs that are tapped on either end. Exact same stuff used on standard RC heli frames. I have completely balanced all 6 motors/props on my HEX so I have no need for vibration dampening of any kind. Because of this I will not be adding in any grommets or vibration reduction bits. I will be using landing gear from quadframe.com http://quadframe.com/collections/frontpage/products/200mm-clearance-landing-legs and integrating this into my design. I have just completed the 3D CAD solid model of my gimbal. If anyone wants to view the "live" solid model you can go to the following path

https://dl.dropbox.com/u/91287103/NEILLS DIY GIMBAL.EASM

You can then save the file to your computer. This file is an edrawings file. You can then use a free version of edrawings to view the solid model, rotate, zoom in, zoom out, etc etc...
The free version of edrawings viewer can be downloaded and installed from: http://www.edrawingsviewer.com/ed/download.htm

Maybe my design will give you ideas for your own gimbal! I am in the process or ordering all my building materials and electronics right now and will begin building soon. I will post up my progress here, as well as setup and flight testing. The purpose of my gimbal will be for aerial video work for our small videography business "Gearheadflicks.com".

Cheers! Neill




(EDIT 8-03-12): My design has evolved quite a bit from my original REV when I posted this up. So the above description and model isn't completely accurate. Later on in the post you can find out what the gimbal has evolved into.
 

Attachments

  • NEILLS DIY GIMBAL.pdf
    774.4 KB · Views: 866
Last edited by a moderator:



jes1111

Active Member
Hi Neill -

zero axial or radial "play" at any of the moving elements

There's nothing showing on the eDrawing about how you intend to accomplish this. Can you give details?

Jeremy
 

jes1111

Active Member
Some other thoughts/observations:

- why not extend the bearing block so it meets the opposite plate? This would improve the rigidity and save a couple of alu spacers. The roll bearing block looks wider than it needs to be.
- I presume you're going to fillet the sharp corners?
- Rather than those long alu spacers between the landing gear plates, why not use truss-cut CF with "on-edge" screws & nuts? Would probably be lighter and stronger.
- I'd also consider some CF pieces to brace (at an angle) between the top/bottom plates of the U-piece. Your biggest enemy will be torsional forces trying to twist the arms - the alu spacers won't significantly resist that.
- I've spotted the little bearing shims on the pitch axis now - that method of pre-loading the bearing relies on the nut tension to apply the pre-load: tricky to get (and keep) right, particularly since your camera tray is also relying on the tightness of that nut to retain its position.
- obviously the component spacing on the roll axis is shown correctly on this model - same issue applies, though, re: the bearing preload. I'd consider wave springs for both axes (for applying the preload).
- using threaded bolts as the axles is difficult, since the contact between the shoulder (unthreaded) portion and the bearing inner bore is not likely to be a good, toleranced fit.
- I'm not sure that two 0.21" CF tubes will be rigid enough fr the camera tray, although I don't think you specified the target camera weight.
- my general experience has been that 2mm CF sheet is still "a bit flexible" - whereas 2.5mm is stiff as hell. Remember that rigidity is cubed for each doubling of the thickness, so the gain is rapid as you go thicker.
- IMO you'll still want some vibration damping in there somewhere :)

Pre-loading the bearings is a great idea, of course, but it will only reduce/remove axial play, not radial play. The radial play will still be dependent on the bearing tolerances (and a few other factors). You're likely still to suffer from backlash, since there will be a handful of it in the servo gearbox and the belt drive itself has backlash. Setting the tension appropriately will help minimise it but it's still likely to be the limiting factor in the overall results.
 
Last edited by a moderator:

BIGTACO

Member
jes1111 many thanks for the review of my design and suggestions. I like it! I will answer all your questions one by one. You have made some good points. This design, as with any design can always use improvement.



(- why not extend the bearing block so it meets the opposite plate? This would improve the rigidity and save a couple of alu spacers. The roll bearing block looks wider than it needs to be.) The reason for some of the decisions I made is I am on a very tight budget due to a new baby on the way! For this reason I am using as many parts and raw materials I already have in an effort to save money. The aluminum standoffs and nylon blocks I have already. The aluminum standoffs are a bit thicker than the blocks. But, you are right about the rigidity so what I will do is just shim up the blocks as needed @ the 4 fastener holes. I have plenty of shims of varying sizes from other projects I can use. I will run screws all the way through clamping the assembly together with nuts on the other side. Good suggestion. I am glad you pointed this out because it has been on my mind too.



(- I presume you're going to fillet the sharp corners?) Oh yes.



(- Rather than those long alu spacers between the landing gear plates, why not use truss-cut CF with "on-edge" screws & nuts? Would probably be lighter and stronger.)
The landing gear I am going to buyout from Quadframe.com. This isn't my design. I am only integrating it into my design. Hopefully it will be rigid enough. But if not, updating and modification to the landing gear structure will be required.



(- I'd also consider some CF pieces to brace (at an angle) between the top/bottom plates of the U-piece. Your biggest enemy will be torsional forces trying to twist the arms - the alu spacers won't significantly resist that.)
This is a very good point. What I will do is put her together and test the rigidity. If its not sufficient with regard to torsion I can always add in bracing. I purchased extra carbon fiber so I will be able to swing that if need be. I can see how torsion may be an issue.





(- I've spotted the little bearing shims on the pitch axis now - that method of pre-loading the bearing relies on the nut tension to apply the pre-load: tricky to get (and keep) right, particularly since your camera tray is also relying on the tightness of that nut to retain its position.) I plan on using "lock nuts" to apply preload to the bearings. In my experience lock nuts allow you to make small adjustments to preload and they do well to keep their orientation on the screw where you need it for low load / zero RPM applications like this. I won't need much preload on these bearings but I agree a precise amount will be required for smooth movement W/O endplay. Its a fine like between removing all the endplay while rotating w/o binding. I can see where those wave washers would be beneficial to allow you to make more precise adjustments to preload though. I will see how it goes without them and consider them if I have issues.



(- using threaded bolts as the axles is difficult, since the contact between the shoulder (unthreaded) portion and the bearing inner bore is not likely to be a good, toleranced fit.) I agree. I have some reservations about this method as well. Ideally I would go with a precision shaft turned to a tighter tolerance and then run a "tight" slip fit or possibly even a very light press fit onto the bearings. This may be what I end up doing if the screw idea doesn't pan out. But, again these are in stock already in my tiny garage workshop... Also let me explain that my intro paragraph above is a bit over the top with regard to axial and radial play. What I should have said is I plan to remove all axial play and leave very little radial play. These ball bearings will have some radial play and hopefully the dim between screw OD and bearing ID isn't too large. For bearings ideally I would go with a tapered roller bearing on one end and radial ball bearing on the other to reduce radial play.



(- I'm not sure that two 0.21" CF tubes will be rigid enough fr the camera tray, although I don't think you specified the target camera weight.) I think your right. I will scrap this and use carbon fiber plate for the tray. If need be I can truss it but I don't think I will need that. My camera will be a Sony NEX-5N. So its very light.



(- my general experience has been that 2mm CF sheet is still "a bit flexible" - whereas 2.5mm is stiff as hell. Remember that rigidity is cubed for each doubling of the thickness, so the gain is rapid as you go thicker.) I already ordered my 2mm thick sheet! Hopefully it will do the trick. I took a look at the Trex 600 heli frames before I ordered. This is a large bird. They use 1.6 mm ~ 1/16" thick carbon fiber plates. I believe 2mm thick will be more than enough rigidity given the size of the gimbal and the weight of my camera. 2.5mm and up to 3mm is used on 1/8 scale onroad and offroad RC car chassis. The reason is these assemblies see HUGE loads. My assembly will see much much less.



(- IMO you'll still want some vibration damping in there somewhere :) I have zero vibrations dampning on my hard mount I fabricated and I get zero jelly in the footage. But, I spent a loooong time statically balancing the props. Then dynamically balancing the props on the motors. The motors are avroto's and the bells are damn near perfect right out the box! My theory is get to the root cause of the vibration and reduce it to a level where it doesn't cause issue with the footage instead of masking it with vibration dampners. That being said, if this were a gimbal I was designing for the masses I would for sure put vibrations dampning in because others aren't as anal about balancing as I am .

jes1111 again your suggestions are appreciated and will help me make a better design. Thanks again.


Neill
 
Last edited by a moderator:

BIGTACO

Member
jes1111 I went ahead and applied a few design changes based on your suggestions. Check out the new model when you get a sec.

Also in regards to the topic of axial play between the OD of the "bearing assembly screws" and the ID of the bearing bores...I took a caliper to the 5/16"-18 screws I will be using and they came in @ .310" on thread major diameter and .309" OD on the shoulder. My bearings (which are still in the mail) should be right at .312" ID. So we are looking at .003" air between them. So not as bad as I thought! But, even if it was worse I dont think it matters too much for this bearing assembly application. The reason is this rotating assembly never makes a full rotation. Also, the "pivoting assembly" has a very low mass. So inertia is very low.

A super tight fit or shrink fit would only be required for a rotating assembly where we have RPM's. Once I apply preload to the bearings there will be enough friction between the faces of the screw head, washers, shims, nut, etc that they will not move radially, with respect to each other, through the entire range of motion. Why? Because mass is low and we aren't rotating at speed.

Now if we took this same assembly with the same preload on the bearings and rotated it at say 500 RPM, then we may have problems.

View attachment 5180
 

Attachments

  • section.jpg
    section.jpg
    103.8 KB · Views: 477
Last edited by a moderator:

jes1111

Active Member
Yep - sounds like a good plan - although it wouldn't hurt to use some (appropriate) Loctite to hold the bearing inner race on the "shaft".

Keep us up to date :)
 

BIGTACO

Member
I made some progress this morning on the tilt assembly. Cut the carbon plates on my band saw (pain in the _ss). Used my drill press to drill all holes for the standoffs with both plates stacked on top of each other and then clamped down. Also cut and drilled the nylon block that will connect the tilt assembly to the roll assembly. This block will have a 5/16"-18 screw running through it, through the drive pulley, and then into the bearings of the roll assembly. Also running through this nylon block will be two 8-32 screws that will be tapped into the drive pulley.

I still need to fabricate the two nylon bearing blocks which will be fastended down to either end of the tilt assembly. I am very happy with it so far. Its very rigid and light. Some pics are below of my progress. Sorry for the low resolution smart phone pics. My real camera is out on a job.

tilt1.jpg


tilt2.jpg
 
Last edited by a moderator:

BIGTACO

Member
Made some more progress. Got the (what I call) roll assembly mostly constructed. Plus I finish and install both bearing blocks into the tilt assembly.

DSC00362.jpg

DSC00361.jpg

DSC00363.jpg

DSC00364.jpg

DSC00365.jpg
 


BIGTACO

Member
More progress on the gimbal...

I am getting close to completing the camera tray. Went with a tubular design similar in concept to Cinestar tray. Also, over the past few days I have been brainstorming some pretty big design changes with regard to external potentiometers. I have decided to use geared 3 turn pots on both axes to increase servo resolution. It was tricky getting the gearing right with regard to correct ratio(s) while staying in the confines of my existing structure. Looks good on CAD. Will be a challenge to make it a reality. I'll keep posting progress.

Sleep, eat, build!


1-20.jpg

2-21.jpg
 


BIGTACO

Member
Looking good :)

Thanks Jes. FYI I decided to change my rotating axes to 5/16" OD precision shafts and matching collars instead of the fasteners and lock nuts. Turns out you were right. The fit between the bearings and screws isn't tight enough and causing problems. New parts in route from Mcmaster Carr. Thank god for Mcmaster Carr. Best online shop on planet earth IMO.
 

jes1111

Active Member
Turns out you were right.
Must be a "law of averages" thing ;)

Looking at that last pic, you've got no stand-off between top/bottom plates on the right arm, whereas there is one on the left arm. Is something else going to fit in there?
 

BIGTACO

Member
Must be a "law of averages" thing ;)

Looking at that last pic, you've got no stand-off between top/bottom plates on the right arm, whereas there is one on the left arm. Is something else going to fit in there?

Well there actually is a standoff on each side but the left side is just up closer to the bearing block. Had to make room for a Savox servo to be in that location. The servo will be a slight press fit between the plates.
 
Last edited by a moderator:

BIGTACO

Member
Progress update:

I ended up having to redesign my tilt assembly bearing blocks some and change material. Originally I chose a nylon grade that has too much elasticity. This caused some compression on the bearings when the whole assembly is clamped together. Also, the material wasn't ideal for keeping my axes in one location under load from the pulley/belt. And last, the faces of the nylon bearing blocks I cut were the same face the bearing bores were machined. Thic caused non-parallelism between either bearing, compounding my bearing problems. b To resolve all this I updated the bearing blocks to annodized 6061 aluminum instead, which I should have done in the first place. As you can see in the photos below I also changed by rotating assemblies from screws and nuts over to precision steel driveshafts and collars instead. After assembling the rotating bits I checked for slop and cannot find any. Atleast none that I can decern with my hand. The bearings are a tight slip fit onto the OD of the shafts. And the bearings are also a tight slip fit into the bores on the bearing blocks. The bearings themselves have a high ABEC rating so they run smooth and W/O much "slop". I ended up going with GT2 tooth form belts and pulleys to reduce backlash as much as possible.

DSC02602.jpg


DSC02603.jpg


DSC02592.jpg


DSC02595.jpg


DSC02600.jpg


In the face of either collar I am going to drill and tap two 6-32 or 4-40 holes (haven't decided yet) to fasten the tilt tray carbon plates to.

DSC02601.jpg



DSC02595.jpg


DSC02594.jpg


DSC02593.jpg
 
Last edited by a moderator:

BIGTACO

Member
Made a good bit of progress this weekend.

-Got the tilt tray mostly completed and installed onto the gimbal. I still need to cut the slot in the camera mount plate for the camera mount screw.
-Mounted the gimbal to the landing gear.
-Mounted the entire landing gear/gimbal assembly to the hex frame. I decided to go ahead and put some vibration isolation into the design at this stage. I used some rubber grommets to separate the landing gear/gimbal assembly from the hex frame as you can see in the pics. This should provide enough dampning since I am well balanced on my props/motors. If not, I plan to use 4 of these instead of the grommets I have http://photoshipone.com/shop/index.php?main_page=product_info&cPath=4_36&products_id=212

Things left to do:
-Cut down the extra lengths of shaft I have sticking out the ends of my pulleys. I had left the shafts longer than needed as a "just in case" measure.
-Fabricate battery mount tray.
-Locate servos in their spots. Drill holes for servo mounts and fasten down servos.
-Install pulley shafts to servo output shafts. Then install the pulleys and belts.
-Once proper servo location/belt tension is determined install potentiometer gearing on either axis. Then locate both pots and fabricate mounts for them. (I will be using 3 turn pots for both axes. Pot rotation will come by way of gearing running off servo input shaft. Gear ratios are determined by desired travel per axis.)
-Modify servo wiring to accept the pots.
-Select and purchase standalone gimbal controller
-Testing testing testing!

5-10.jpg

1-21.jpg

3-16.jpg

4-14.jpg

DSC02618.jpg

DSC02608.jpg

DSC02615.jpg

DSC02614.jpg

DSC02616.jpg
 
Last edited by a moderator:

Tomstoy2

Member
Really nice, BT! Curious, what do you think your all up cost will be on this project? Excluding stabilizer.
 

BIGTACO

Member
Thanks for the kind words. In regards to cost somewhere around 400 - 450 bucks (servos & potentiometers included but gimbal controller excluded). There has been waste due to mistream design revisions. As with any initial design nothing turns out 100% like you plan. If I knew what I know now when I first started I could have shaved off 100 bucks or so.

Really nice, BT! Curious, what do you think your all up cost will be on this project? Excluding stabilizer.
 
Last edited by a moderator:

Top