I haven’t tried this yet, but I think you can build a reasonable monochrome raster laser display out of commonly-available parts for a few dollars.
12kHz is a horizontal scan on a 320×200 60Hz-refresh display; 28.8kHz is a horizontal scan on a 640×480 60Hz-refresh display. The newly-fashionable 16:9 aspect ratio gives you more screen space than 4:3 for the same number of horizontal scans; 1024×600 is the same number of pixels as 900×680, but at a 12% lower horizontal scanning frequency. You could carry this much further, e.g. with a 8:3 or 16:3 aspect ratio. The cones of cheap dynamic earphones can move hundreds of microns at, I hope, tens of kilohertz. A piece of aluminized mylar parallel to the plane of the cone, rigged up so that one end is pushed and pulled by the earphone cone, while the other end remains fixed, could transduce that translational motion into an an angular displacement. (It needs to remain taut, which can be achieved by a little spring tension.) Scanning a 37cm horizontal from 1m distance requires 2 atan (37cm/1m/2) = 0.37 radians of angular deflection of a laser beam, which means 0.18 radians of angular deflection of a reflecting surface; you can achieve this by displacing one end of the mylar by 0.19× its length. If the mylar is 1mm long, that’s 190μm, which should be doable. You need two scanning axes at right angles to scan the light in two dimensions. In theory, these could be two actuators angling the same piece of mylar on different axes; it might be easier to use two separate deflectors. So you just have to hold a 1mm chunk of mylar fixed at one end and flat while pushing the other end back and forth with the cone of a cheap headphone. Cheap reflective mylar is over 95% reflective, so multiple reflections shouldn’t be a big problem. Many import-store laser pointers come with trick holographic “lenses” that diffuse their light into some pattern; when distributed over a surface comparable in size to a computer monitor, the light is usually still quite visible. Rip out the laser diode and drive it with a bitstream. As long as the bits are long compared to the RC constant of the laser diode, this should work fine. Unfortunately, I have no idea what these guys’ RC constants is, but I suspect they’re well up in the MHz. Assume 10% efficiency and 5mW on 6V. That means 50mW input, or just over 10mA, so the effective R is under 1kΩ, although diodes’ E-I curves are very nonlinear. Since they’re not designed as capacitors, C is unlikely to be over 10pF. That would give us τ = 10ns, adequate for dot clocks up to nearly 100MHz. 1024×600 at 60Hz would need about a 37MHz dot clock. (A little looking for datasheets suggests that (a) I’m being a little too conservative with the R, (b) it is unusual for anyone to care about things like parasitic capacitance on cheap laser diodes, so the figures aren’t published, and (c) the pulse rise time on expensive laser diodes is in the hundreds of picoseconds, so I’m not being *too* conservative.) The total cost for the parts should be around US$10 or so. The display certainly won’t be as bright as a normal LCD, but it might be serviceable at least in the dark. And it will use a hell of a lot less power and be more portable. And, given a high-speed phototransistor, you can use it to scan things, too. It may be a good idea to drive the horizontal scan, at least, with a sine wave instead of the traditional sawtooth wave, in order to improve the maximum feasible horizontal scanning rate. You have to compensate for the resulting variations in brightness and resolution in software: not feasible, or at least not easy, when the NTSC and PAL standards were defined, but fairly straightforward these days. For optimal light efficiency, you could put the projector on your forehead and project the display on retroflective paint — that stuff with little glass balls in it to provide heligenschein, commonly used for street markings and street signs. I don’t know what the dBi “antenna gain” of retroreflective paint is, but subjectively it seems quite substantial. (Some random website claims 4 mrad cone angle reflections for Colorado license plate paint, which would be 16 microsteradians; that would be about 59 dBi. That seems implausibly accurate; it would require the projector to be within 2mm of your pupil at a distance of 1m.) I’m surprised I don’t seem to have written about this on kragen-tol before, because I’ve certainly been thinking about it for years. -- To unsubscribe: http://lists.canonical.org/mailman/listinfo/kragen-tol