To do this, I started with plain, flat curtains on a standard static bar. I sewed rings into the back of the curtain in line from the inner lower corners to outer upper corners, then ran a control line through them (dead-ending on the inner lower corners), through blocks on the curtain rod supports, then down the side. To make it all controllable from one side, I have the right side running back across the top, then joining to the control line on the left side, resulting in a single point of control for both sides. Since I'm using 2mm control lines, I needed pretty small blocks with tight clearances, so I ended up using ones from the Harken 16mm line - utter overkill, but they get the job done. To make sure that the curtains closed tightly when closed (since there's not much overlap), I sewed magnets into both sides, so that they 'zipper' it shut when the curtains are down, and can easily pop apart when opened.
That's the state that the curtains have been in for about a year and a half. My next step was to automate them, but I ran into a snag. If you look back at the Card Swipe Unlocker project I did 7 years ago, I used a low-RPM, high-torque motor - that motor had been missing, blocking my next step. I looked into other linear actuators and curtain automation systems, but none of them were suitable for my purpose. When going through a random drawer a couple weeks ago, I happened to re-discover the missing motor - joy! Once I had that element, I went about designing the circuit to control the raising and lowering (I also recycled the relays and plastic box from that project). I wanted to use the same control system as my lights so that it could be automated using the same application (which includes scheduled events - ie open them in the morning), which meant that it needed to work via a DMX-controlled relay. In addition, I decided that including a dedicated DMX control channel for direction (ie one channel for motor on/off, one for direction up/down) was excessive. After much pondering, I came up with this circuit that does the job beautifully.
There are 2 switches that are triggered by knots in the control line - one indicating that the curtains are all the way open, and one that they are all the way closed. The switches have to be placed facing in opposite directions in order to be triggered correctly by the line, and become circuit interruptions for the motor power when triggered (so normally forming a closed circuit when at rest). To make the reed switches activate by the knots in the control line, I added a housing made from InstaMorph (mouldable plastic) around the switches, with a channel for the control line, and a 'hood' to ensure that the knot would hit the switch.
The power supply coming into the circuit is always active. When the DMX-controlled relay is flipped, it triggers 2 DPDT relays. One of the relays determines the polarity of the motor circuit's power supply (so changes the direction), and the other relay determines which of the switches affects if the motor circuit is open or closed (we want the 'passive' switch to be removed from the motor relay circuit when switching direction, otherwise the motor would never start, as it is activated when the change happens). Ergo, when you change the state of the DMX controled relay, it activates the motor in the opposite direction, and changes the switch determining when the motor should cut off. Add in a DMX-controlled relay (http://www.amazon.com/dp/B00CEOLYZ0), and you've got yourself a DMX-activated curtain control box.
One issue that I ran into with this is, when the curtains are approaching full closure, there's very little (or no) upward pressure on the control line. Unfortunately, this means that the switch indicating that the curtains are fully closed is never triggered, and they start winding back open the other direction. To get around that, I added another control line spanning from the junction point of the two primary lines on the left side across to the rod support on the right side. I then added a small weight on an additional block to that 'tension' line with a small weight on it, and it worked perfectly - that tension line gives just enough pressure as the curtains are closing activate the switch, and pulls up and out of the way when the curtains are opened. The length of the tension line is such that it is taught along the top when the curtains are fully opened, and drops down when they are closed (the image has it taught across the top when they are closed, which is not entirely accurate, but you get the picture).
And there you have it! Curtains that open like those on a stage, and are automated using a single DMX channel and one power supply. I plan to swap out for a stronger motor to make the opening process faster, but this fully proves out the automation process.