I have since built a new, far better, motion control head using an internal belt mechanism, pictured above. Read about it here. This post is left here for those interested in a more simple arrangement.
In May I came across the OpenMoCo project which is a community that is producing an open-source photographic motion control system, concentrating mainly on the software side of the problem. For a while I have been trying out timelapse photography and the idea of creating paning, tilting and sliding shots has been interesting but the equipment to do this can cost more than £1000. Being an engineering student and having access to some machining tools I thought it would be fairly easy to come up with and manufacture a system for myself. Initially I had a look around the OpenMoCo forums and found rigs other people had built, bearing in mind I had only a few days to build the system I decided to base my design heavily off of this pan-tilt head one user built. I had a very limited time and budget so couldn’t risk going with a totally novel design; I knew this had been tested and would work and I also had access to a laser cutting machine so it made sense to heavily take advantage of this prototype.
I quickly modelled a first draft in a 3D cad program, luckily most of the suppliers who I purchased the components from offered cad models so this allowed me to get on with completing the design before I had even received most of the parts. The cad software made it easy to quickly change the design and I had the whole thing finished in two days. The final design comprised of 10 flat acrylic plates which were laser cut using s 50W CNC laser, courtesy of the University of Bath. Four of the plates needed a small amount of additional machining to add threaded holes and to ensure that the ends were properly perpendicular, which is important since the laser cutter leaves a slightly tapered cut.
A render of the cad model for one of the earlier designs is shown above, subsequently the profile of the base sections was changed to create a fully enclosed base. The manufacture was fairly straight forward and was nearly all finished in a few hours, other than the electronics, which is still ongoing. I still plan to add a slider to the system and have the an extra motor and driver board ready for when I can implement this. For the slider I am planning on using either an igus slider or percission rail and DIY carriage. We shall see.
The head is driven by two stepper motors through a worm gear which reduces the motor speed by a ratio of 30:1, this allows 6000 steps per full rotation of the head, and since the motor drivers I use break each step down into 8 microsteps the rig can have a spatial resolution of 48000 points per rotation. The whole system is controlled using an arduino UNO, an electronic prototyping platform which allows a cheap and easy way to program your own microprocessor. Initially I was planning on using the OpenMoCo timelapse engine to control the head but after some time a friend and I decided to come up with something a bit different that would allow manual control as well as programmable time-lapse moves which could continue even without a computer attached to the arduino. The firmware is still a work in progress but in the meantime I will be using OpenMoco.
A picture of the final system is shown on the left, I will update soon with some test footage once I finish the electronics and software side.
I will post more information once I have all the software sorted out, at the moment the electronics have just been (semi)finished and I am doing a couple of tests, getting the time-lapse function working with some quick and dirty code but will update when I have something better and more stable. Here is a picture of the messy electronics, the opto-isolators are doing nothing at the moment, they are for later. Enclosure coming soon…
I also hope to use the head for giga-pan photography as well, this is when you take a large grid of pictures at different angles and stitch them together automatically into one much larger image. Also I might try some experimentation with real time macro videography. Any ideas?
Edit: Added first video tests below, just got to iron out the sticking problem at the beginning of the move ramping and things should be working out well.