h a l f b a k e r yGetting blown into traffic is never fun.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
This flat panel display works vaguely like a CRT in that the active pixel is raster scanned in an analog way while three "guns" are modulated to create an image. However, the raster scanning is done using sound waves in water while the "guns" are LEDs feeding light into a total internal reflection light
chamber.
The display is built in layers. From back to front, the layers are:
1. Black paint backing.
2. Clear reflection layer.
3. Main plastic panel. This plastic panel has a mirror coating on all edges and has three LEDs feeding light into the edge. Total internal reflection means that the light from the LEDs fill the chamber without escaping. This plastic panel is made of a hydrophobic material.
4. Perforated clear mask. This is shaped a bit like a CRT honeycomb mask. It is made out of a hydrophobic material, and it's purpose is to keep the water from touching the main optical panel.
5. Water chamber. This is a chamber of water where sound waves travel. Where the sound waves constructively reinforce, bumps of water pushing into the perforations of the mask can touch the optical panel. This contact with high index of refraction water spoils total internal reflection, causing light to leak out in the active pixel. This light reflects against the curved surface of the bump to direct the light in a forward cone.
6. Clear plastic wall of the water chamber.
The only other equipment are a set of maybe a dozen or so sonic actuators along the bottom edge of the water chamber. This are fed sound waveforms so that they constructively reinforce to provide a raster scanning active pixel. These waveforms are synchronized with the incoming hsync and vsync signals (from either VGA or Component video signals).
While the waveforms appear complex, they are calculated in a simple way. For each pixel in the raster scan, the required waveform is a spike at time t-d/s, where t is the time the pixel needs to be displayed, d is the distance from the pixel to the actuator, and s is the speed of sound in water. The waveform is simply calculated by summing the spikes for all pixels in the raster scan.
This waveform calculation only needs to be done once per change in display resolution.
Compared to other display technologies, this display can be very thin and low power, and inexpensive. The control electronics are extremely simple, with analog amplifiers simply feeding VGA signals straight to the red, green, and blue LEDs. The sonic actuators simply move according to a repeating pattern, synced up to the vertical and horizontal sync signals.
The number of active components is not much larger than a CRT, with a modest number of sonic actuators instead of several focusing/aiming coils.
piezoelectric display
piezoelectric_20display
[xaviergisz, Mar 18 2007]
Speeds of sound
http://hyperphysics.../tables/soundv.html
[Smurfsahoy, Mar 18 2007]
cylinder design diagram
http://i68.photobuc...y/soniccylinder.jpg
[Smurfsahoy, Mar 20 2007]
Spinning mirrors concept
http://heim.ifi.uio.no/haakoh/avr/
(The second project on the page. Be sure to check out the .avi video.) [Smurfsahoy, Mar 20 2007]
Spinning sonic reflector diagram
http://i68.photobuc.../SonicReflector.jpg
[Smurfsahoy, Mar 20 2007]
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
Destination URL.
E.g., https://www.coffee.com/
Description (displayed with the short name and URL.)
|
| |
But with ultrasound, wouldn't your pet dog/bat/dolphin(?) start freaking out whenever you turned the TV on? |
|
| |
Anyway, I feel like the speed of sound may not be quite fast enough to scan every pixel 30 or 60+ times per second without any interfering between signals, considering that each wave is going to reverberate at least a few times, and so forth. |
|
| |
One thing that might help would be to scan in some complicated way that maximizes the distance of each pixel from the immediately next or previous pixel's row or column. Some sort of shimmying diagonal pattern. |
|
| |
At any given time, the limits of the speed of sound means that there will be waves for many future frames in the water at a given time. |
|
| |
Yes, there will be a lot of strange interference, but none of this will be visible because the water bumps only cause an optical leak when they reach the threshhold amplitude to touch the optical chamber. For example, the threshold bump height may be .2mm, while the "noise" level could only be .1mm. |
|
| |
But any normal image is going to have tons of regions of similar color and brightness right next to one another. Thus, all of the interference that hurts the most is happening at once - same frequency, nearby in the display,constructive amplification together... The bright patches are all going to start bleeding into their surroundings when 20 pixels of 0.1 combine together to be strong enough to trigger one or two otherwise darkened pixels up to 0.2 nearby. |
|
| |
Possible modification: Change the water to pyrex, and your speed of sound multiplies from 1,493m/s to 5,640m/s. (link) I'm not sure if solids are worse with the interference or not, though. If that doesn't work, at least change to glycerol for a +30% or so gain in speed. |
|
| |
And if synthetic diamond gets cheap enough to manufacture for stuff like this... yipes, you could go 6x faster than in water. Mmmm, HDTV, with solid diamonds! |
|
| |
Plus, if solid state stuff works, then you dont need any of that hydrophobic jazz. Though you would need to worry about some evil evil broadcasting station sending out an image over the airwaves that matches the precise resonant frequency of your TV screen, causing it to instantaneously shatter. Perhaps that's a reason to take those messages that say "Don't you dare change the channel!" seriously. |
|
| |
The sound waves don't form a complex image but rather just a single dot. This dot raster scans in a regular pattern, while the red/green/blue LEDs modulate brightness to form the image. |
|
| |
Using a solid material for the sound waves would certainly avoid the "messiness" of using water, but I'm not sure what the dynamics would be like. Your sonic material would be a thin flexible sheet preshaped into a mold with lots of little bumps. The vibrations are now transverse waves rather than sound waves (like vibrating a sheet of rubber). |
|
| |
My concern is that this system would be very noisy, since the transverse vibrating sheet is acting like a huge speaker. With a water chamber, the noise is mostly contained in the chamber itself. |
|
| |
Also, the slapping impacts of a solid sheet may be more damaging to the optical slab than water bumps. |
|
| |
Still...using a solid bumpy sheet for the vibration layer is potentially much lighter, simpler, and robust. Ideally, the sheet is pretensioned so that there's an evenly spread force trying to keep the sheet lightly pressed against the optical slab. In operation, the slapping effect of the raster scan keeps the vibration sheet "flying" above the optical slab. Thus, the vibration sheet will naturally be elevated to the maximum height of the vibrations. |
|
| |
1. Black rear coating.
2. Clear reflection coating (to prevent the black coating from spoiling total internal reflection).
3. Clear plastic optical slab.
4. air gap
5. Bumpy vibration sheet. |
|
| |
Hmm...actually, the more I think about it the more I like it. |
|
| |
[edit: Actually, a specific bumpy surface isn't needed, a rough frosty surface will do nicely. Unfortunately, using either preformed bumps or a frosty surface will harm black levels and render the rear clear/black coatings irrelevant. With water bumps, the water is transparent except for the currently active pixel.] |
|
| |
You couldnt just paint the back of the pyrex black and have it behind the optical sheet? Or does that mucks things up? (I dont understand the whole total reflection thing, really - how can you have black backgrounds that you can see, and also total reflection at the same time? That seems contradictory) |
|
| |
And yes, it would act like a speaker, but an ultrasonic one. No matter how loud it is, you couldn't hear it. It might still make you feel queasy or something though, not sure. If it isn't resonating, too, then it probably wouldn't be that loud anyway. |
|
| |
You would want to paint the back of the device black, but it doesn't prevent ambient light from reflecting off the frosted vibration surface. |
|
| |
The way total internal reflection works is that light within a higher index of refraction material will totally reflect if it hits at an angle shallower than a critical angle. Thus, the surface of a transparent material acts like a perfect mirror for light at a shallow angle, but it's transparent to light hitting it at steeper angles. |
|
| |
Without the frosting, ambient light would mostly pass through the transparent layers and get absorbed by the black coating in the back. Unfortunately, this also means that light escaping the optical chamber will do so at angles off to the sides--the image isn't visible from directly in front of the screen (which is what you'd want to be the ideal viewing position). Frosting deflects light forward, but it also reflects ambient light. |
|
| |
I've been trying to figure out how exactly to produce the desired raster scan with acceptable performance. Currently, I'm thinking that perhaps the simple approach is the best: |
|
| |
1. You stretch the vibration sheet sideways so the speed of vibrations is much faster horizontally than vertically. |
|
| |
2. You use one or more actuators to "ping" the left edge of the sheet. This produces vertical wavefronts from left to right at a 31.5+ kHz rate. |
|
| |
3. You use one or more actuators to "ping" the top edge of the sheet. This produces horizontal wavefronts from top to bottom at a 60+ Hz rate. |
|
| |
Where the two wavefronts intersect, there is sufficient amplitude for the sheet to touch the optical chamber. This provides a place for light to escape the optical chamber. The active dot raster scans left to right rapidly and up to down slowly. |
|
| |
Right, the scanning part is what I assumed you meant in the first place. I was suggesting that instead of scanning left to right, top to bottom, you skip to every 20th row/column or something instead, and then go back starting with column #2 and skip every 20 again, etc. That way, there is minimilized interference, since rows and columns right next to one another have the longest gap possible in between scans. |
|
| |
That might work, but still, it's ridiculously difficult to tune correctly. As in, your kid throws a foam ball at your TV, and it's out of sync. |
|
| |
I have been stuck thinking in terms of CRT style left-right raster scanning. This has caused me a lot of grief, and I now realize that for this display scanning along 45 degree diagonal scanlines is the best. This avoids the problems of trying to make the speed of vibration transmission different in one direction vs another. |
|
| |
As in my last post, the two vibration actuators are along two adjacent edges. In this case, those edges shall be the left edge and the bottom edge. The pinging periods are chosen to that they are relatively prime (when counting the period in pixels). |
|
| |
But there's a big unfortunate complication. The total length of scanline scanned per second is going to equal only 1.414 times the speed of sound in the material. For a 1280x720x60fps display, that translates to a display only 5 inches wide! So, while this approach may be suitable for a projector, it needs some more work for a larger flat panel screen. |
|
| |
Faster scanning rates are possible using wavefronts angled at shallower angles than 90 degrees. For instance, a 100 inch wide screen is acheivable using wave generators on the left and right angled with a trapezoidal shape with 1:20 sloped sides.
If you hide the funny shapes behind an outer bezel, you have 3 inch margins on the sizes and a 10 inch margin at the bottom. |
|
| |
The waves you generate from the left side are just one ping on screen at a time. The waves you generate from the right side is many pings on screen at a time. For a 16:9 screen, there are 17.8 pings on the screen at a time. Each of these wavefronts will encounter two wavefronts from the left by the time they reach the other side of the screen. |
|
| |
The raster scanning in this example is vertical. When two wavefronts interact, the pixel starts off at the bottom edge and it scans vertically upward. With a 1280x720 screen, each scanline is shifted 72+ pixels to the right of the previous scanline. |
|
| |
Wait... why have everything linearly installed? |
|
| |
Just have two cylinders inside the display on left and bottom. They have a few actuators on them, and they whirl around at 3600rpm. With each pass, each ultrasound source will sweep past the point where it is aimed at every single pixel in that row or column. Just turn it on and off really quickly at the right times for pinging, and it will hit every pixel in the row or column you want. Combine the two together, and you can hit individual pixels. |
|
| |
The advantage is that you can fire a number of pings at once without worrying about having to wait until the one before is done propagating, because by making a big enough cylinder, you can create steep enough angles to ensure that the beams wouldn't't cross paths until they got to the screen. |
|
| |
This would require a frequency for your ultrasound sources of 4MHz. I'm not exactly sure how the collimation for sound works, though. It might be that you need more than one cycle in order to get the focused properties necessary to collimate. If that is the case, add another 4MHz for every cycle needed? |
|
| |
And how expensive is one actuator? |
|
| |
I don't quite understand the cylinders suggestion. Anyway, I've figured out one way to solve the problem of limited speed of sound, which is to utilize the third dimension. |
|
| |
See, what I've been trying to do is get a wavefront to move across the screen at many times the fastest available speed of sound. Not possible, right? But imagine planar waves in a 3d block hitting the screen almost perpendicularly. It crosses the screen at a line, and this line scans across the screen much faster than the speed of sound. |
|
| |
I've calculated that by using a material with a LOW speed of sound, the block doesn't need to be very thick. For a 1080p display, the vibration chamber only needs to be about 12mm thick (interestingly, the required depth depends only on the scan rate and not the height/width of the display). |
|
| |
So, these planar waves are generated in the vertical direction by an actuator at the bottom edge. As the wave rebounds around inside the block, it rapidly scans up and down in a zig-zag. |
|
| |
On the left side, a simpler planar actuator generates a planar wave moving across the screen much slower (at the speed of sound in the material). This wave rebounds to zig-zag scan left and right. |
|
| |
So the total number of moving parts is now just three--the vertical wave transducer and the horizontal wave transducer, and the vibration block itself. Now, the thickest and heaviest layer of the display is the vibration block; the optical sheet is thinner. |
|
| |
I linked the cylinder idea as a diagram. In addition to what is drawn there, you would probably have to add some reflective surface to the actuator, so that it could handle dozens of rows of pixels with every pass. |
|
| |
As an alternative, you could have some sort of crazy rotating AND shaking reflection surface that every 60th of a second angled the incoming beam of a sonic actuator to every pixel location on the screen in turn. Then just turn the actuator on and off in a timed fashion based on which pixels are supposed to be on or off. (that is confusing too, I realize. i added a second link that uses this concept in a laser pointer projector. It could be just a little thimble sized object though, instead of a bank of mirrors, if it was professionally done.) |
|
| |
Since 150+MHz total are required to run the screen, you might need a handful of actuators assigned to different sectors of the screen, but same principle. The reflector would reflect beams down onto the screen much the same was as the cylinder does, thus not requiring any of this interference tweaking. |
|
| |
The problem with these ideas is that they would require a slightly thicker screen. But they could probably sell for about $400, so that might not matter so much. |
|
| |
I've probably missed something here, but can this screen be mounted vertically? |
|
| |
Yeah, once youve built any of the designs mentioned here, you could flip the tv around any way you wanted, and it would still work. Or do you mean in a different relation to the components? |
|
| |
I added another link showing the design diagram for the fixed actuator, moving reflector model. |
|
| |
Spinning for a ten facet drum (assuming the screen is divided into 4 sectors with an actuator for each) would be at a reasonable 1,100rpm. Rocking would be one full rock back and forth at about 30 times per second or so. Im not sure there are good motors out there that could handle the abuse of being shaken this much indefinitely, but if so, super. Otherwise, the rig could have more than one section, so that a single shake causes the laser to scan the whole distance several times. That's a more delicate piece of equipment, though, and hard to precisely machine and polish. |
|
| |
I am assuming here, of course, that it is possible to reflect pings of collimated ultrasound in an accurate and predictable manner, yes? |
|
| |
I thought there'd be a problem with water leaking out. |
|
| |
I've abandoned the idea of using water in favor of a block of solid material for the vibration medium. |
|
| |
Smurfasahoy, unfortunately ultrasound beams simply can't be collimated as tightly as you're thinking. It's a simple matter of wavelength and diffraction limits. Laser beams can be very narrow because the wavelength is very small. But sound waves have much larger wavelengths, so they'll spread a lot. This is why I've been restricting to "beam" geometries of expanding circles/spheres, and simple summations of them (a planar wave is the sum of a bunch of expanding sphere waves expanding from a planar surface). |
|
| |
Okay, so then use the exact same design, but with one blue, one red, one green laser, for the whole display. |
|
| |
The lasers for that would be about 1,000 dollars right now in bulk for a sufficiently high wattage to get a good picture, and the prices are going rapidly down every year (as in, the laser dot on the screen itself literally is the display. No LEDs or anything else, just a spinny mirror, a laser, and a piece of frosty plastic in a box.) |
|
| |
Or use one cheapo infrared laser, and some surface that mechanically reacts to laser light (like some sort of piezo with a cheap little photoresistor attached) A laser perfectly good enough to open a pixel for the whole screen at 60fps can be had for about $50, so this one would be about the costs of the screen parts. |
|
| |
Sound just is seeming too unnecessarily cumbersome here. And you need at least a few expensive actuators no matter what design you use, which makes it not that much cheaper than other models anyway. |
|
| |
The sort of laser projector you're thinking of does exist, but it's extremely expensive and the bandwidth is apparently currently only capable of low resolution images (like about 70 scanlines). |
|
| |
I've got this concept down to something relatively simple. I think I can make it even simpler and much less expensive using waves of electric charge instead of sound... |
|
| |
Ah, now the electric waves I could go for. You might have something there. |
|
| |
I still think the sound waves would be just too slow, at least for the 1280x720 full sized screen you want, maybe not something else. |
|
| |
But a jump from 5000m/s max or so up to 300,000,000m/s... you could pretty much make a monitor the size of a football field if you wanted to, I'm guessing. |
|
| |
The challenge with electric waves is that they're simply too fast. I have an idea to massively slow them down... |
|
| |
Imagine a zig-zag wire which runs left-and-right while stepping down in increments. The wire could step downward by 1mm while running horizontally 1000mm. A signal pumped into this wire will form a horizontal line of electric charge which creeps downward at only 1/1000 the speed of propogation through the wire (maybe 50% to 80% the speed of light in vacuum). |
|
| |
Okay, so you have one zig-zag screen going downward, and another zig-zag screen going rightward. The first screen has electric charge in a horizontal line; the second screen has electric charge in a vertical line. Where these lines intersect, the charges repel. That repulsion causes the one layer to bump outward and touch the optical chamber. |
|
| |
2. Horizontal zig-zag sheet (printed on structural backing). |
|
| |
3. Thin insulation coating. |
|
| |
4. Vertical zig-zag sheet (printed on flexi-layer). |
|
| |
5. Flexi-layer of flexible frosted material. |
|
| |
This results in diagonal scanlines. I think the energy of the electric waves can be mostly recycled if the waves are "bounced" at the bottom/right corner. Thus, the vertical wavefront bounces left and right and the horizontal wavefront bounces up and down. The resulting scanning is like a ball bouncing in Pong, in diamond patterns. |
|
| |