Note : this post is just something from the physicist trying to sort out his ideas on a subject, it’s not about OS development
Screens are a very important component of computers. You spend most of the time you use the computer staring at it. So you want it to be easy on the eyes, and offer excellent color rendering that doesn’t depend on the direction from which you’re looking at it. Recently, there have been lot of technological developments in the screen area, so I thought it’d be interesting to summarize how the various screen technologies we have nowadays work and conclude by an explanation of why the PixelQi technology, which sounded very promising, turned out to work incredibly badly when tested by Engadget in a review of Notion Ink’s Adam tablet (one of the first consumer device using this technology).
A long time ago, there used to be only kind of screen, the Cathodic Ray Tube or CRT. It worked by sending a stream of electrons in vacuum, and deviating them using high voltages to a point of the scren where they’d get a chance of reacting with a chemical that emits light. At the end of their evolution, CRTs gave high-quality images for a low price, but still had some fundamental issues, noticeably that they were large, heavy, extremely energy-inefficient, and very sensitive to magnets and electromagnetic fields (unless they had proper electromagnetic shielding, which made them even heavier). CRTs have also this issue that they send electrons all over the place, which although not dangerous for the user can be a serious problem in some environments.
The portability issues in particular led to the development of large Liquid Crystal Display (LCD) screens. Originating from the calculator and watch world, LCD screens work by acting on the polarization of light. To picture yourself this, imagine that light particles, called photons, oscillate in a direction that is defined by a vector called polarization. In light emitted by everyday sources (fluorescent lamps, LEDs…), the polarization associated to each photon is random, but perpendicular to the direction in which photons travels.
To work with polarization, LCDs mainly use two tools. The first is the polarizer, a chemical device that projects the polarization of light on a given axis. To say it simply, it damps all oscillations in the orthogonal direction and dissipate the energy in the form of heat. The second tool used by LCD screens is liquid crystals themselves. These rod-shaped molecules have the interesting property of rotating the polarization of light in a controlled way: when you take a solution of liquid crystals at room temperature and do nothing with it, the polarization of light crossing it will be left almost unchanged. But if you start to apply a voltage on the solution, it will start to rotate the polarization of light in a way that’s directly dependent from the voltage being applied. To summarize, liquid crystals allow one to rotate the polarization of light between 0 and 90°, and that’s exactly what we want in order to build a screen.
All LCD displays work in the following ways, with only implementation variations : a white light source on the back emits light, that goes through a first polarizer and becomes polarized light. Then liquid crystals are used to control the polarization of light and put it in the desired direction. Finally, a second polarizer is used to control the intensity of outgoing light : if the light coming from the liquid crystals has the same polarization as the second polarizer, then it will get through, but the more its polarization differs, the more it will be damped, to the point where if the polarization of light ends up being perpendicular to the direction of the polarizer, one will get maximal damping : very few light gets through, the result is near black.
It is more complicated to build a high-quality LCD than an high-quality CRT, so LCDs tend to either give lower-quality image or cost much more. LCDs are thin, light, more energy-efficient, do not care about the electromagnetic environment, and emit nothing but heat and light, so at first look they seem to be the exact contrary of CRTs. However, LCDs have several extra issues worth pointing out :
- LCDs are inefficient by design. Given a sufficient amount of engineering work and a sufficiently high price, nothing fundamentally prevents a CRT from reaching 100% efficiency. But in an LCD, the first polarizer used to make polarized light out of normal light eats up exactly half of the incoming optical power (the part which had a polarization orthogonal to that of the polarizer) and dissipate it in the form of heat. So the ideal LCD has 50% power efficiency when displaying a pure white image. When displaying black images, the situation is even worse : optical power is generated only to be fully dissipated in the form of heat shortly thereafter. In short, depending on the image displayed, and no matter how far technology goes, an LCD screen may either waste all or half of the energy you put in it for producing heat. Like the thermal engines found in most modern cars, LCDs are fundamentally inefficient. It is a matter of opinion, of course, but I don’t think it’s healthy for engineers to work on technology that will always waste power no matter how much effort they put in developing it.
- Current LCDs need white light to operate. Now, this defect is not a fundamental defect of the LCD technology, but all current consumer LCD screens share this issue. What I’ve described above only gets you a monochromatic LCD screens. To produce colors, modern LCDs use the following trick : they put red, green, and blue filters in front of each red, green, and blue pixel, so that only the red, green, and blue portion of the incident light may get through. There are two issues with this. First, it’s much more difficult to make efficient white light sources than to make efficient color light sources. Second, if we assume that each color takes one-third of the white light’s energy, then the ideal LCD, when displaying a pure white image, will have an efficiency of one third of 50%, so 17%. Add up the efficiency of the rest of the screen and get out of the ideal model, and you get this inconvenient truth : a modern LCD, when displaying pure white (which is the situation where its efficiency is optimal), is less efficient than a car.
- LCDs produce poor black. Once we get out of the ideal model, the first thing which we have to consider is that polarizers are not perfect. A part of the incident photons that do not have the correct polarization will always get through. That’s why all LCD screens have this remarkable feature that if you put yourself in a dark room, they will be bright even in dark regions. While this effect exists for nearly all kind of screens to some extent, the LCD’s “please waste my battery power” subtractive color synthesis described above makes it significantly worse, because production of light on black pixels is not accidental, it’s a feature of the core design.
- LCDs have moving parts and liquid inside. When you apply a voltage to the liquid crystal solution, the molecules inside of it rotate to take a new internal organization. This process takes time, like everything that involves moving parts in computer hardware. This is why on early LCDs, scrolling looked so painfully laggy. Modern LCDs have thankfully improved a lot, but one can’t help but wonder how much longer screens would last and how much easier manufacturing would get without the need for moving parts. Speaking about it, the need to have a liquid crystal solution in the screen at all is in itself an annoyance : it causes LCD screens to be vulnerable to pressure, bending, and sharp objects, and it puts additional constraints on the manufacturing process (again, if it was simpler, we’d get better color rendering for a cheaper price).
- LCDs only look right in a single direction. Ever noticed how the look of the image on your screen changes when you move your head ? Ever wondered how we can still have screens like this in the 21th century ? Vision angles have always been a big issue with LCD screens, and even though things have improved a lot, it would still be much better and natural to have a screen which looks the same no matter from which point you look at it. Why do we have to deal with this, by the way ? Because LCD screens are made of several layers, and use polarizers. Multiple layers are problematic because if you look in a direction that goes too far from the horizontal one, chances are that you’ll end up looking at the liquid crystal of one pixel through the polarized of another. Polarizers are another problem, because they work in a single direction while light propagates in 3D, which means that they’ll always end up damping legitimate light. See schematic below.
LED displays (including OLED, AMOLED, etc…)
To address the multiple shortcomings of LCD, the display industry has been working for a pretty long time now on yet another evolution of traditional watch/calculator screens : LED displays. In this technology, one uses lots of small red, green, and blue Light Emitting Diodes (LEDs) as sources of light, and simply puts more or less power in them in order to make them emit more or less light. You probably know LEDs as the small round lights which are frequently used to indicate an appliance’s status (power on/off/sleep, etc…) and as the red and green lights frequently used to display digits on alarm clocks.
Color LEDs are the easiest to manufacture and most efficient source of light produced as of today. They can be made absurdly small, and can last very long. So this solution is quite attractive for screen in the future. On the other hand, manufacturing screens based on them is an engineering challenge, due to the need to produce huge lots of LEDs (a usual screen would contain millions of them, so they must be really, really cheap) and, perhaps most important, to control each of these LEDs individually through individual wires. To make LEDs cheaper, the industry has decided to manufacture them using extremely cheap organic materials, calling the resulting device OLED (O stands for Organic). But the wiring complexity problem is much harder to solve, which is why as of today, OLED screens are still mostly used on small devices like phones, though some 12″ prototype screens have been demonstrated.
OLED displays are very complex to manufacture as of today, so they are typically only seen in expensive devices like high-end phones while less expensive ones prefer to stick with traditional LCD. OLEDs are thinner than LCDs, lighter, do not care about the electromagnetic environment, do not perturbate it, can reach extremely high theoretical efficiency as they are not inefficient by design, do not require complex white leds, produce excellent black (you turn the LED off, so it doesn’t emit light anymore), have no moving part or liquid as part of their design, more generally can last much longer, and produce an image that looks right from a very large set of vision angles (that is only limited by the LEDs’ ability to emit light in all directions). If the manufacturing problems can be solved, these screen will probably end up fully replacing the LCD in the future, since they are better in every other way and are probably as good as any portable screen based on the emission of colored light will ever get.
Care must be taken, though : recently, many LCD screens manufacturers have tried to capitalize on the OLED’s increasingly well-known intrinsic superiority and mislead people into thinking that some LCD screens are LED displays. As an example, so-called “LED TV” products are in fact LCD screens which use an array of white LEDs as their backlight instead of a fluorescent lamp, which leads to slightly higher power efficiency but still remains under the very restrictive limits of the theoretical LCD efficiency. If large OLED screens become sufficiently common and easy to manufacture to compete with LCD on wide screen areas, it is to be expected that the number of scams like this will grow significantly as a reaction.
And now, for something totally different…
Electronic ink and iMoD/Mirasol : the electro-mechanical reflective screen
As said before, based on what we know of physics today, OLED is the best portable light-emitting screen which can be envisioned. But thinking of it, do we really need to waste power emitting light all the time ? There is this thing around the Earth called the Sun which provides light to us during most of the day. We use either this light or other electric sources of light as part of our light, and it is known to be medically a bad idea to stare at a backlit computer screen in the dark. So what if computer screens stopped using our precious battery power emitting light and used the wildly available ambient light instead ? Potential result would be a screen which looks equally well from all directions, is much easier on the eyes, and eats so little power that we’d get computers and phones which last weeks (months ?) on battery.
Well, this is not a technological dream. There are commercially available products which work this way (Amazon’s Kindle and Sony’s Reader spontaneously comes to mind) and R&D work on such screens has been going on for a long time and is still ongoing. There are two main approaches which are being studied today : electro-mechanical approaches, which use moving part commanded by electrical currents, which I’ll describe now, and LCD-based approaches, which I’ll describe later.
The first kind of electro-mechanical approach is known as “electronic ink” or “e-ink” and works as follows : a solution of small spheres is placed between two electrodes, the topmost of which is transparent. These spheres are made of two small hemispheres in contact with each other : one is black and has a negative charge, while the other is white and has a positive charge. By putting a positive or a negative voltage between the electrodes, it’s possible to either put the white or the black side of the spheres on top, and as such to create monochromatic graphics. Filters may also be used for colors here, as we’re not wasting extra power by using subtractive color synthesis : it’s only the sunlight which is absorbed, not our battery’s precious power.
This approach offers very good visual appearance and has a very low energy consumption when displaying idle graphics (a pulse of current is required from time to time to make sure that the spheres keep their orientation), which is why it is widely used for displaying books, but it also has its issues. E-ink is very expensive to manufacture, uses liquids on the inside (which we know is bad), and has extremely slow refresh times (much slower than early LCDs) due to the large amount of mechanical motion involved. The durability of such screen is also to be questioned : a lot of mechanical motion is involved, and it is well possible that the colored spheres will lose their electrical charge as time passes, as an example when hitting each other. Another problem is that only “pure colors” can be displayed, while nothing like gray is available. We’ll see how this big shortcoming can be addressed when studying the second kind of electromechanical screen.
Second approach explored nowadays is the iMoD (interferometric modulator display) screen, marketed by Qualcomm as “Mirasol”. It works as follows : a semitransparent layer is put on top of a mirror, the space between both being very thin and controlled electrically. Optical interference phenomena occur in this configuration, so that only one color of the incident light is reflected. So if we switch between one visible color (say, red) and one invisible color in the UV spectrum, we effectively have one pixel that can switch back and forth between red and black. So we can use red-green-blue pixels, but the problem is that we do not control the intensity of the reflected light. iMoD partially addresses this problem by having several tiny red pixels side by side and turning only part of them on and off. It is clear, however, that this approach will not scale up to the 256 levels of color intensity per primary color allowed by modern screens.
The high complexity of iMoD is going to make it quite expensive to manufacture too, compared to other approaches, and its mechanical nature is going to impair the lifetime of iMoD-equipped devices. However, refresh rates are much better than with e-ink, the performance of first demonstrated models being on par with early LCD screens especially being very encouraging. No liquids are involves, which means that iMoD screens should be more robust than e-ink and LCD and that making foldable iMoD screens should be doable. On the other hand, the limited color depth still impairs the possible range of applications. The switching process is so that energy is only consumed when what’s displayed on screen in changing, so that iMoD screens have, like e-ink, a very small power consumption.
Transflective and reflective LCDs
It is pretty clear that LCD technology, as it stands today, has high chances of being outperformed by newer screen technologies in almost every possible way in the future. On the other hand, it would be sad to just throw away all this LCD display technology we’ve developed over the years, so if recycling is possible it should definitely be considered. One first approach, embraced by many manufacturers, is the transflective LCD, which is like a regular LCD screen but has also be designed to use the ambient light as an additional backlight. Coupled with light sensors or a very slight LCD backlight intensity, this technology can greatly reduce the power consumption of LCD screens, since the only thing which consumes power on such a screen is the optional slight backlight and liquid crystals themselves. Transflective screens also have these advantage over electromechanical technologies that they could theoretically have the full color depth of an LCD screen, though on most currently sold devices color rendering is still very bad and transflective operation remains best when used on roughly black and white graphics.
To quote Engadget’s review of the Notion Ink’s Adam state-of-the-art transflective screen… “(…) the Adam’s viewing angles are terrible. Approach it from any angle but head-on and either the whites or blacks wash out, and if you tilt it to the left everything begins to turn a sickly yellow. The colors are also a bit washed out, and if you’re a fan of deep, inky blacks you’d best look somewhere else, as the best the Adam can do is a shade of noisy purple.”
These are essentially classical defects of LCD screens, they are just made worse by the facts that transflective technologies are still relatively young and that ambient light has to travel twice as much in the LCD layer when the screen operates in reflective mode (from the exterior to the core of the screen to the exterior). Still, this technology has lots of potential and it’d be a very nice thing to see it reach the maturity of current LCD technology and competing with other reflective technologies and OLED. Considering how far standard LCD has went and how fundamentally flawed it was from the very beginning, it is not totally weird to think that transflective LCD technologies have their chance.