Storage you can watch
It occurred to me the other day that the Kindle is really nothing more than a very tiny storage device. I don't mean the actual SSD chip inside, although that's physically very small; K3 and KK have 4GB of storage with 1GB-ish masked off for a wee Java kernel and the scratchpad for the OS and experimental browsers, and text takes up so few bytes that the only reasons to give it that much are for the Audible audiobook option and because demand for standard USB storage below that is probably so underwhelming that it would cost them more to order a lesser capacity. Those chips are basically nothing -- I have a 4GB flashdrive attached to my keys which is exactly big enough to house the USB-A connector. If computers commonly had micro-USB ports, you could easily manufacture a 32GB portable drive small enough to accidentally insufflate, never mind lose.[1]
I'm actually talking about the screen.
Using a display as temporary storage is nothing new. Any kind of persistent display will do. Nowadays we use RAM whose bits are flipped electromagnetically, but in yon early days of computers, most memory was a kind called delay line. The idea is that you have a thingamajig that has an input at A, an output at B, and an extra line running from B back to A, and that the distance from A to B takes some time to traverse. You squirt in some pattern of bits at A, and if it takes those bits 30ms to get all the way to the end from there, you've just stored your bits for 30ms. The loopback wire is for reading those bits off at B and shunting them back to A for another 30ms of storage, should you so desire. In the very earliest delay line memory, the titular delay was physical; the bits were composed of pressure waves in a sealed tube of mercury[2]. Later on, though, the delay was sort of electro-optic, which is to say that bits were displayed as a pattern of dots on a cathode ray tube with a long decay[3], and read back with a small camera. All the camera had to do was take what it saw and convert it into electrical impulses for outputting or looping back -- so, pretty much any video camera.
e-ink displays, however, are a more permanent storage device. The thing about CRTs is that the illuminated phosphors eventually fade, and need to be refreshed in order to retain the picture. e-ink does not. The advantages that make e-ink so indispensable for a small electronic reader are that the picture is reflective, not transmissive -- you read by light that heads toward the page and bounces off it, not by light shining through the picture from behind, as with TFT displays (this means no backlights, which are a huge power sink in laptops, handheld game consoles and tablets) -- and that e-ink displays only require power to change what they show. Each pixel is a little capsule containing opaque black and opaque white; to change from one to the other, a tiny electrical charge zaps the capsule and upends everything, making the other color rush to the top. Once you've written something there, it stays there. Like, for instance, the bits in a solid-state drive chip, or the gates in an EPROM.
I note that e-ink is not just black and white anymore, it's also available in grayscale and now in a somewhat rudimentary range of color. (Displays can be manufactured, so far as I know, using any two colors for the pigment beads. You could have a neon orange and burgundy reader if you wanted. They picked black and white for obvious reasons of readability.) If I read things a-right, e-ink has progressed to a state where the medium itself is effectively analog[4], though the controllers are still simple digital things. The reader is digital; the display is not. A similar situation used to exist with SVGA CRT computer monitors -- VGA is an analogy technology, which is why no, you can't get a cheap VGA-HDMI dongle from Best Buy, but the PC controllers were very much digital, which is why you could only have "32 million colors" from the monitor when technically the CRT's per-pixel brightness and color mix was infinitely variable.
The pixel resolution of a K3 screen is 800 x 600 -- comparable to a low-end SVGA display from the good ol' days, but shrunk down to a 6" diagonal, so it looks nicer. (And e-ink does genuinely have a much sharper image than VGA. I don't know if you've noticed this, but light goes everywhere, to the point where we had to advance quite a ways in science before we worked out how to keep everything in lockstep long enough for lasers. The e-ink pixels, not having the halo that surrounds every glowing phosphor, don't bleed into their neighbors no matter how white they are.) That's a possible 480,000 bits stored on the screen, which is chump change compared to the SSD inside in terms of raw data. You can encode more information in a grayscale display, depending on how you choose to decode multiple levels of brightness, but it'll still only be 480,000 bits of whatever. The only reason we have such an odd fondness for it is because, unlike the SSD, we can decode what's on the e-ink with our naked eyeballs. Well, presuming that we're literate.
As the late great George Carlin used to say, "These are the thoughts that kept me out of the really good schools..."
[1] By far the largest portion of an e-reader's mass, if you were wondering, is the battery. Batteries by their nature are dense, manufactured as they are with heavy metals -- they have names like Li-on [lithium ion] and NiCad [nickel-cadmium] for a reason. The second largest part is the case, which needs to be stiff enough to not flop around or bend when you press buttons, and impact-resistant enough to not shatter when you inevitably drop it. The top layer of the screen is plastic rather than the Gorilla Glass used in phones, I think, hence the popularity of protective cases.
[2] The filling didn't need to be conductive -- in fact, when proposing this technique, Alan Turing's first suggestion was gin, which was plentiful, easy to get a hold of, and whose purchase and shipment would not raise any red flags. Turing thought this was hilarious; unfortunately, the British government did not.
[3] Cathode ray tube displays work by hitting phosphors behind a fine metal mesh with electrons to make them glow. The length of time it takes the glow to fade completely is the decay of that particular sort of phosphor. The ideal decay profile for television was short and non-linear, glowing at as steady a brightness as possible until just before that phosphor is hit again, then dropping off precipitously just as that phosphor is hit again for the next frame. Other applications, like oscilloscopes, medical equipment, and radar screens, used phosphors with a much longer, smoother decay.
[4] Technically, the e-ink would have a very large number of possible valid states, down to the limits of the size and shape of the pigment particles in the capsules. Before anyone says anything wise-ass about this being light-years away from "infinity", I will also point out that the signal resolution of cassette tapes is limited by the speed at which the recording apparatus can vary its strength and the size and permittivity of the magnetized particles on the mylar, and that the resolution of records is limited by the response of the tone arm and the hardness and density of the vinyl. In short, any analog thing (smooth flow from state to state) will turn into a digital thing (falls out of one state straight into another) if you stare at it from the right level of granularity. Even holograms, whose spatial resolution is dependent on the wavelength of the laser used to record the image and the area available for encoding.
I'm actually talking about the screen.
Using a display as temporary storage is nothing new. Any kind of persistent display will do. Nowadays we use RAM whose bits are flipped electromagnetically, but in yon early days of computers, most memory was a kind called delay line. The idea is that you have a thingamajig that has an input at A, an output at B, and an extra line running from B back to A, and that the distance from A to B takes some time to traverse. You squirt in some pattern of bits at A, and if it takes those bits 30ms to get all the way to the end from there, you've just stored your bits for 30ms. The loopback wire is for reading those bits off at B and shunting them back to A for another 30ms of storage, should you so desire. In the very earliest delay line memory, the titular delay was physical; the bits were composed of pressure waves in a sealed tube of mercury[2]. Later on, though, the delay was sort of electro-optic, which is to say that bits were displayed as a pattern of dots on a cathode ray tube with a long decay[3], and read back with a small camera. All the camera had to do was take what it saw and convert it into electrical impulses for outputting or looping back -- so, pretty much any video camera.
e-ink displays, however, are a more permanent storage device. The thing about CRTs is that the illuminated phosphors eventually fade, and need to be refreshed in order to retain the picture. e-ink does not. The advantages that make e-ink so indispensable for a small electronic reader are that the picture is reflective, not transmissive -- you read by light that heads toward the page and bounces off it, not by light shining through the picture from behind, as with TFT displays (this means no backlights, which are a huge power sink in laptops, handheld game consoles and tablets) -- and that e-ink displays only require power to change what they show. Each pixel is a little capsule containing opaque black and opaque white; to change from one to the other, a tiny electrical charge zaps the capsule and upends everything, making the other color rush to the top. Once you've written something there, it stays there. Like, for instance, the bits in a solid-state drive chip, or the gates in an EPROM.
I note that e-ink is not just black and white anymore, it's also available in grayscale and now in a somewhat rudimentary range of color. (Displays can be manufactured, so far as I know, using any two colors for the pigment beads. You could have a neon orange and burgundy reader if you wanted. They picked black and white for obvious reasons of readability.) If I read things a-right, e-ink has progressed to a state where the medium itself is effectively analog[4], though the controllers are still simple digital things. The reader is digital; the display is not. A similar situation used to exist with SVGA CRT computer monitors -- VGA is an analogy technology, which is why no, you can't get a cheap VGA-HDMI dongle from Best Buy, but the PC controllers were very much digital, which is why you could only have "32 million colors" from the monitor when technically the CRT's per-pixel brightness and color mix was infinitely variable.
The pixel resolution of a K3 screen is 800 x 600 -- comparable to a low-end SVGA display from the good ol' days, but shrunk down to a 6" diagonal, so it looks nicer. (And e-ink does genuinely have a much sharper image than VGA. I don't know if you've noticed this, but light goes everywhere, to the point where we had to advance quite a ways in science before we worked out how to keep everything in lockstep long enough for lasers. The e-ink pixels, not having the halo that surrounds every glowing phosphor, don't bleed into their neighbors no matter how white they are.) That's a possible 480,000 bits stored on the screen, which is chump change compared to the SSD inside in terms of raw data. You can encode more information in a grayscale display, depending on how you choose to decode multiple levels of brightness, but it'll still only be 480,000 bits of whatever. The only reason we have such an odd fondness for it is because, unlike the SSD, we can decode what's on the e-ink with our naked eyeballs. Well, presuming that we're literate.
As the late great George Carlin used to say, "These are the thoughts that kept me out of the really good schools..."
[1] By far the largest portion of an e-reader's mass, if you were wondering, is the battery. Batteries by their nature are dense, manufactured as they are with heavy metals -- they have names like Li-on [lithium ion] and NiCad [nickel-cadmium] for a reason. The second largest part is the case, which needs to be stiff enough to not flop around or bend when you press buttons, and impact-resistant enough to not shatter when you inevitably drop it. The top layer of the screen is plastic rather than the Gorilla Glass used in phones, I think, hence the popularity of protective cases.
[2] The filling didn't need to be conductive -- in fact, when proposing this technique, Alan Turing's first suggestion was gin, which was plentiful, easy to get a hold of, and whose purchase and shipment would not raise any red flags. Turing thought this was hilarious; unfortunately, the British government did not.
[3] Cathode ray tube displays work by hitting phosphors behind a fine metal mesh with electrons to make them glow. The length of time it takes the glow to fade completely is the decay of that particular sort of phosphor. The ideal decay profile for television was short and non-linear, glowing at as steady a brightness as possible until just before that phosphor is hit again, then dropping off precipitously just as that phosphor is hit again for the next frame. Other applications, like oscilloscopes, medical equipment, and radar screens, used phosphors with a much longer, smoother decay.
[4] Technically, the e-ink would have a very large number of possible valid states, down to the limits of the size and shape of the pigment particles in the capsules. Before anyone says anything wise-ass about this being light-years away from "infinity", I will also point out that the signal resolution of cassette tapes is limited by the speed at which the recording apparatus can vary its strength and the size and permittivity of the magnetized particles on the mylar, and that the resolution of records is limited by the response of the tone arm and the hardness and density of the vinyl. In short, any analog thing (smooth flow from state to state) will turn into a digital thing (falls out of one state straight into another) if you stare at it from the right level of granularity. Even holograms, whose spatial resolution is dependent on the wavelength of the laser used to record the image and the area available for encoding.
Comments
Post a Comment