After more than 3 years of planning, thinking, reading and experimenting stopmo, i decided it is time to start the big project. I have an idea of a short video and in the time passed i refined it untill now when i know exactly what i want to show, how i want it to be filmed, what equipment i need,etc.
Part of the project, there is this dollie/crane assembly that is required to move the camera around the stage smoothly and swiftly. The particularity of the crane is that it is built with readily available materials from hardware store and...parts that are 3D printed. From my calculations, the crane will be able to support a camera up to 5 Kg along with the required automation like follow-focus.
For the music video i am planning to film, the camera jib and several other MOCO systems will be controlled by a computer. The same computer will also control DF remotely. The reason for this setup is because i need to plan the shots and see them in real time and make the required changes to camera path and props before the actual shot is made since there is a lot of automated motion on the set.
It looks really good, clearly the product of much thought (mine has been the product of much experimenting, then a bit of thought!)
The axes that seem to be most problematical are crane and rotation. It looks like you are using a belt for the crane. Does this operate without significant backlash? I found a belt and geared motor gave me unacceptable backlash, so have changed it for a vertically-mounted ballscrew, as in the pic. This gives zero backlash.
As for rotation, I can't see what you are using to power this. I am in the process of changing my whole arrangement so it becomes a 450mm diameter lazy susan type bearing (routed into birch ply and using 14mm bearing balls). In one of your pics, it looks like you have opted for something similar. Around the perimeter of the rotating plate I will fix some flexible plastic gear rack, and have a stepper mesh with this. The whole arrangement should be solid and stable on my wide track, rather like yours except made from an alloy ladder!
I would be very interested to hear some of the details of your rig.
Yes i agree, the most problematic axes are the crane and the rotation. Initially i wanted to go with a setup similar as you did in the picture but later on, calculating (BPE) the forces involved, i opted for the lazy susan approach. Initially i wanted to machine the circular bearing on the big circumference of the bottom pan disk but i had only one piece of aluminum that fitted the bill and i could not find the second. You can buy from e-bay or AliBaba(China) larger thrust bearings that could help but at the diameters required to counteract the focres involved, they are very expensive.
All rotational axes will be direct drive. If you use HTD belts of adequate size, there is no backlash. Most of the backlash you experience is because of the tolerances in the geared drives because you are using generic drives. Industry manufactures backlash-free geared drives but the cost is prohibitive for this kind of setups. Look for example at Kuka demonstrations on Youtube. You will find out that they use less gears and more belts inside their high speed precision robots. I am not using geared drives for rotation and crane. I am only using them for camera yaw, pitch and horizontal positioning, since the arms are very short and backlash is right now very small (i might ditch them if in the trials i will find that i cannot compensate the backlash).
The machine is a prototype since i am expecting to make some changes especially on the camera end. I am using High Torque nema23 motors with small pulleys(15-25) teeth. Right now, the crane has a single motor fitted but if i have a problem with that, crane not holding position, there is provision for a second one on the other side. All the belts i use have steel core wires so there is no stretching backlash. I will use micro-stepping but i am still in tinkering phase with the software.
The Dollie/Crane was built with flexibility in mind and usage in crammed and small places. First of all i wanted to be able to use materials readily available at local hardware store, without too much fuss. All the rails are U shaped sections of 20x20x20mm. However many of the parts are either 3d Printed or cut on CNC so the equipment required to pull this off is not trivial at all. Back to the flexibility, the idea was to use inexpensive or less expensive materials to manufacture a semi-pro device that can be used and abused and expanded if required.
The wheels are built exactly like inline skates but with side mounts instead of more common top mount. The dollie base is wider than the track width and by having both i can by simply reversing the wheel assemblies, transform everything from a floor dollie into a suspended one. This is beneficial if your studio floor space is crammed or you have little space.
Another different thing is the moving crane arm. In most of the setups, the crane arm has a fixed length and you can manually adjust the length of it. In my setup , the arm will be also controlled by a motor and the length will be varied accordingly. The biggest reason for that is again small studio. if you have a stage that occupies most of the room and you need to do a full stage traveling move, with a jib of 2 meters you need on each side of the stage the extra 2 meters of track. But by having a dynamically adjustable jib arm, you need only 1 extra meter of space available at the back of the crane to accommodate for the arm retraction. The rest of the setup is common to other installations in respect to camera movement, attachment and accesories.
The crane will be controlled by some custom designed software and not by Dragon Frame. The main reason is because the crane will be synchronized with other props and DF has limitations in that sector. If you remember the destruction machine in Boxtrolls, that was controlled by a custom MoCo rig. What i am trying to do is a similar rig but for another purpose. So i am starting with the most important piece of equipment which is the dollie. If this is a success, then i will proceed to build the rest.
Right now, what i am planning to do is finish the build with the current setup i have in mind and film some 30 seconds of intro or title or something, to get a feeling of what is wrong and flimsy and does not work and correct them. And with all of those corrected i can start the big shooting.
Another idea i have is to share the build files, STLs, and CNC for anyone to use and built their own rig if they want to or if they cannot, maybe... commission one from me :)
I will update this as i go but if you have more questions do not hesitate to ask.
Thanks for the detailed reply. You are right that most of the backlash is down to the gearboxes on the steppers. I went for these initially because I read a thread where someone had used them. So now swapping out with high torque ungeared motors. I reckon Nema 17s are adequate for the pan and tilt functions, and keep the weight down.
Really like your extending arm idea. It certainly will make it more versatile in a confined space, but probably quite a construction challenge. My arm is also quite long, so I have designed the head section to be able to detach and operate separately, on a smaller mount (as yet unbuilt).
With the lazy susan, I assume the black blocks with bearings are your 3d printed pieces. How many bearings are you using round the circumference? My plan is to have a groove in the ply for the balls, as I can machine wood but only do basic metalworking. Accurate routing ought to prevent it from slipping off-axis, but I will also have a shaft to prevent any movement and against the risk of the whole unit tipping.
Great that you will be updating. Information is quite hard to come by, and it's very useful to be able to compare experiences. Might have saved myself some money on those geared steppers!
Please see attached files for details on how the bearing blocks are constructed and how they run on the disk. I use 4 bearing blocks, two of them are positioned at 180 degrees and the other 2 at 45 degrees to one of the blocks. If you notice, each block has 2 bearings. One is used to support the disk and the other one is used to prevent the disk from slipping. In this way, the whole set of bearing blocks form a rolling/centering assembly with virtually no play at all. The center bearing is used only to prevent the entire thing from falling and keeping the assembly tight. One of the blocks has slits instead of just holes so you can move the block and tighten or loosen the play in the assembly as you wish. One of my main problems with the disk is that it is still made of wood. Even if it is MDF. As you see, i also used a groove cut in order to keep the bearings somehow aligned. But that groove will have a different purpose. I plan to cut an aluminum disk , same size as the groove and glue it in with some epoxy. In this way i will harden even further the support surface and prevent the grooves that might appear if the weight in the jib is too high and ball-bearings will sink into the wood.
The bearing blocks are only some of the 3d printed parts. The biggest parts i'm proud of are the wheels and the wheel assembly. They are 3d Printed too. A single wheel is composed of 3 main parts: A core, a running cylinder and a rubber layer. The core and running cylinder are printed first. The running cylinder is then machined on a mini lathe and then put together with the core in a molding jig. Then the rubber is poured in, et voila, a full nice smooth wheel. The rubber helps in reducing the vibration caused by small particles that can end on the rails and gives more flexibility to the otherwise harder, run surface. At the same time helps greatly when the dollie is used without rails, on hard surfaces.
Hi Julian. Very neat bearing blocks, a good solution. I did notice you were using MDF. Not a fan of it, as it absorbs water and sags quite readily. Probably ok in the doubled thickness you are using. One way to stiffen it up a bit would be to coat it with some thin epoxy, which would soak in and make it more impervious. This is what I shall be doing with my ply dolly base. I'm just contemplating now whether to abandon the big ball race and go for something like your arrangement.
I have just spent an hour or so in the workshop swapping over motors to see if I can sort out the backlash issues. My geared tilt motor miraculously seems to have zero backlash once it is powered up. But the pan motor has a significant slop. All gears are tightly meshed, so it's not that. I swapped it out for an ungeared motor, and sadly this is not up to the loads. I think my solution will be to apply a bit of tension in one direction with some bungee - crude, but it works if I don't try and do a whole revolution!
Your description of making the wheels sounds awesome. I have used scooter wheels with the rubber machined off, so just the bare metal rims. The problem I found when I was using skateboard wheels was that they developed a flat area when the dolly was left in one position for a length of time. This translated into a noticeable 'clunk' and a jolt to the camera. I might have opted for nylon wheels if I had found them on eBay before the scooter wheels.
Thank you for the advice. I am planning to coat all MDF parts in 3 -4 layers of marine paint so it won't sag too much. If ,after the title test, i find out that MDF is unsuitable, i will redo the entire thing in marine plywood which is much stronger and more resistant to sagging.
From my experience, the geared motors are used to further reduce the rotational speed of the motor and gain more torque. But at the same time, the assembly the motor moves must be very well balanced so the motor puts minimal effort into moving that part. Also , in DF code, i do not see any backlash compensation so that might also add to the issues you are experiencing. What you can try to do is, instead of using gears, use timing belts. There are some designs on the web that use that kind of setup but you need to find belts that will support the assembly.
Coating the MDF in anything impervious is a good plan. It should be fine. If you do change to plywood, I would recommend birch ply as it is very consistent. Marine ply can be good, but there is also some rubbishy stuff around, and the lamination are thicker.
Thanks for your thoughts on timing belts. I opted for gears for the head, and can't quite see how belts would help to eliminate the backlash, which is in the motor gear train. And I worry about the sideways loads imposed on the motor bearings when the belt is tensioner, unless one goes for a supported axle to which the motor is attached. But this all adds to the bulk.
Completely agree about balancing. I have used gym weights, and was surprised by how many I needed in the end. This all adds to the loads on the crane and rotation bearings, and makes for lots of stiction. Nothing's easy!
Sorry for not posting more often. I managed to finish the mechanism for keeping the front arm (the camera arm) horizontal. I favored this mechanism instead of the common control bar because the arm will have 2 configurations. One common where it will be fixed on the crane, where adjustments can be made only before a new scene is shot, and one mobile where the length of the arm on the camera side can be varied with a motor inside the lift disk. The second option will require more mechanics to be developed but that is reserved for phase 2.
Also in one of the photos you can see the pitch and yaw assemblies along with the 1:5 stepper motors mounted. The assembly i think is quite heavy and most probably will be reworked from thinner aluminum profile which will allow also for another roll axis to be installed. Again that is in phase 2.
Firmware goes slow because i could not find many ready made libraries for acceleration control and motion control for my MCUs so i was forced to write them myself. Right now the speed tables work, i can accelerate and decelerate a motor quite smooth. I need to implement a lot of functions but the most complex will be synchronization.
It looks really good, and very solid. I can see now how you can extend and retract the arm, which will be a very useful axis as you said for confined areas. Very different from my more conventional jib, but I can see yours working well. Many ways to achieve the same end!
I am envious of your 3d printer, as it is ideal for all those joints. One suggestion for reducing weight without changing all the components is to cut large diameter holes. I did it with some of my aluminium parts, but it is a lot of work for small gain.
finally the first video taken from the rig itself. No motors were used because even if they are already mounted, i am still working on the firmware. I couldn't resist to try and get a video to see how it behaves. There is still some wobbling because the track is not fixed yet and the table on which the dollie runs is not straight.
There is still a a lot of work to be done but i think i am on the right track.
Nice! Now it gets exciting!!