LEDs and a low-powered microscope
March 18, 2012 8:34 PM   Subscribe

Several tablet and smartphone LEDs, at 80x magnification, including the new iPad (3), the Galaxy Nexus, and the Nintendo 3DS.
posted by ignignokt (36 comments total) 11 users marked this as a favorite

 
Here's a related post, linking to some technical analysis about this stuff. Pretty neat-o.
posted by Blazecock Pileon at 8:43 PM on March 18, 2012 [1 favorite]


From the link: "It’s a very cheap USB microscope from Bresser. I wouldn’t recommend it; as you can see, the images aren’t particularly good."

Yikes. A similar USB microscope scope that I have, from a different maker, doesn't allow the LED illumination to be turned off, and doesn't allow the auto-expose function to be disabled. That explains why the author has to make excuses anti-glare screen.

Screen resolution has dimensions in X, Y, and Z. The properties of the color depth are important to the actual clarity of the detail, not just the hues. Setting the magnification on those toys is not really repeatable, either. From the article: "...a picture of the Galaxy Nexus’s screen that should be roughly at the same scale."

This is starting to sound like an evaluation of MP3 bit rates made through $3 ear buds.
posted by StickyCarpet at 9:03 PM on March 18, 2012


I've been meaning to do basically the same thing, for a while. I had a couple shots already, and so I just took a few more now. I don't have a microscope, but I do have a ridiculous 5:1 macro lens. So these links all cover 7mm of horizontal screen space:

(warning: large self-linked images ahead)

iPhone 4S
PlayStation Vita
Nexus One
Desktop monitor (NEC 3090WQXi)

I don't have an iPad or new iPad or 3DS, and who cares about a PSP anymore.

Some of the blurry spots are touch-screen grease. I had to touch it to get the [+] in the right place. (Note the professional white background.)
posted by aubilenon at 9:36 PM on March 18, 2012 [3 favorites]


I have a Galaxy Nexus, and while the color's a little funny if you use the phone next to anything else, without a comparison point, your eyes adjust and it looks fine. And the resolution is outstanding. It's a good screen for a phone.

I wouldn't want a pentile monitor, but the tech works well in that small form factor.
posted by Malor at 9:48 PM on March 18, 2012


Not to be picky, but those are not pictures of LEDs, those are LCDs. (LCDs are lit from behind, by either LED's or fluorescent.)
posted by wanderingstan at 9:57 PM on March 18, 2012 [2 favorites]


"This is starting to sound like an evaluation of MP3 bit rates made through $3 ear buds."

Agreed, but at least he's not claiming any great insight from it - more just "here's some cool (if crap) pics - and hey, you can see the difference even at roughly equivalent scales."

And it got me to clean the screen on my (old) iPod Touch, which is always a good thing. Then I looked at it at 40x under my (ancient) Olympus stereo dissection microscope. I could even see the little lands where the terminals connect to each LED junction.

(On preview: wanderingstan, the R, G, & B dots visible in the pics are from the LED backlight - LCDs are B&W only. The resolution apparent to the eye is a combination of LCD dot pitch & LED dot pitch. Gross mismatches between the two lead to issues like colour fringing, so as the LCD dot pitch gets finer the LED pitch has to as well.)
posted by Pinback at 10:06 PM on March 18, 2012


Not to be picky, but those are not pictures of LEDs, those are LCDs.

Well, except for the picutres of the OLED screens.
posted by hippybear at 10:18 PM on March 18, 2012 [1 favorite]


Wanderingstan: They are not all LCDs. The PSVita and the Nexus displays are OLEDs. But yes, for the most part your correction is ... correct - the rest of them are in fact backlit LCDs.
posted by aubilenon at 10:18 PM on March 18, 2012


*pictures

If anyone has picutres of OLEDs, please post links. Because I have no idea what those may be.
posted by hippybear at 10:18 PM on March 18, 2012


hippybear: I'll see if I can find a caemra to tkae smoe wiht.
posted by aubilenon at 10:25 PM on March 18, 2012 [4 favorites]


(On preview: wanderingstan, the R, G, & B dots visible in the pics are from the LED backlight - LCDs are B&W only. The resolution apparent to the eye is a combination of LCD dot pitch & LED dot pitch. Gross mismatches between the two lead to issues like colour fringing, so as the LCD dot pitch gets finer the LED pitch has to as well.)

This is pretty wrong. All LCD backlights are pure white, whether they're fluorescent or LED. The liquid crystals let through red, green, and blue light. There's no dot pitch to speak of with the backlight. It's just a giant shining white thing. Some LED-backlit TVs do have rough control of brightness in zones around the image (this is called local dimming), but there are only usually a few hundred of those zones. You don't even need a backlight with some color LCD screens.
posted by zsazsa at 11:04 PM on March 18, 2012 [1 favorite]


Totally missed the OLED's. But the point is that this post is titled wrong: these are not pictures of LEDs. At most, LEDs were involved in generating the backlight. Better to follow the lead of the article itself, these are pictures of screens.

But a cool post. I've been working on an article about screens and pixels as the under-appreciated ubiquitous machine-human interface, so the timing is perfect.
posted by wanderingstan at 11:10 PM on March 18, 2012


All LCD backlights are pure white, whether they're fluorescent or LED

To nitpick again, the white is rarely totally pure. Fluorescent "white" is anything but, and LED backlighting typically uses blue LEDs coated with phosphor coatings to fill in the green and red bits of the visible spectrum. The purest white is comes with RGB LED LCDs, which are thicker and only appear in expensive professional equipment.
posted by wanderingstan at 11:32 PM on March 18, 2012


If we're doing nitpick, there's no such thing as "pure" white, because white is what's perceived when all three of the retina's cone cells are roughly equally strongly stimulated, and that can only happen with a mixture of lights of different wavelengths.

There are broad-spectrum mixtures that will do this, like sunlight and incandescent light; incandescent light typically looks yellower than sunlight, because its spectrum has more power at the red/green end than at the blue end, but both sunlight and incandescent light are mixtures of a very large number of wavelengths.

Mixtures containing only three wavelengths can also look white, provided those wavelengths are spread across the spectrum in such a way as to stimulate all three color receptors equally. When you're looking at white on a color CRT, that's exactly the kind of light you're seeing - the red, green and blue phosphors are all quite narrow-band emitters. In fact pretty much all phosphors are like this, including the ones randomly scattered throughout the fluorescent coating in a fluorescent tube light. With a suitable choice of phosphors, it's possible to create a spectrum full of narrow-band spikes that the eye can't tell apart from sunlight or incandescent lamp light.

However, these narrow-band spectra don't reflect from colored objects the same way wideband spectra do. Imagine illuminating a blue-green object with a very narrow reflection spectrum with both kinds of white light. The wideband white would contain enough energy within the object's reflection band to light it up nicely; narrow-band red+green+blue white light from a CRT or fluoro tube would probably do so much less strongly, making the object appear relatively darker under that illumination. This kind of effect is why replacing tungsten incandescent lamps with compact fluorescents can completely screw up an interior design; when you're picking indoor colors you should do so using the same illumination source they will end up lit with.

Given that the aim of a screen backlight is to push as much light as possible through the narrow-band red, green and blue filters in front of the LCD subpixels, it would be very wasteful to use a wide-spectrum source as a backlight. Whether the backlight is LED or fluorescent, for best efficiency the light it makes should be a peaky mixture of the same red, green and blue as the subpixel filters let through.

It seems to me that it ought to be possible to build an LCD display with green phosphor behind the green subpixels, red phosphor behind the red subpixels and blue phosphor behind the blue subpixels, dispense with the filters, and backlight the thing with ultraviolet to make the phosphors glow. I don't know if anybody is already doing this, but I would expect it to save quite a bit of battery power.
posted by flabdablet at 12:45 AM on March 19, 2012 [8 favorites]


And for comparison, a closeup of a Commodore 1702 Color Monitor.
posted by i_have_a_computer at 2:01 AM on March 19, 2012 [4 favorites]


And for comparison, a closeup of a Commodore 1702 Color Monitor.

I don't believe that's a closeup.
posted by dumbland at 2:59 AM on March 19, 2012 [1 favorite]


here is a closeup of an Etch-A-Sketch. here is a closeup of a test pattern. here is a closeup of a robin's eye. here is a closeup of a pail of sand. here is a closeup of a ghost leaving a small child's bedroom. here is a closeup of the wall in that bedroom. here is a closeup of a memory of the carousel going around and around. here is a closeup of a small hand in a older hand. here is a closeup of wind blowing through a curtain and a blanket being pulled up under a chin.
posted by Lipstick Thespian at 3:41 AM on March 19, 2012 [7 favorites]


This bit at the bottom kind of annoyed me:
Dr. Drang has some more pictures, taken using a better microscope. Also, on a completely unrelated note, I agree with him that iPhoto on iOS isn’t necessarily a beginner’s program, but I don’t think pro users are happy about having to read a manual or watch an Apple Keynote, merely to figure out how to do the most basic things in the application.
Come on really? You expect a program to be able to do everything you want, and you don't want to have to learn anything before you use it? How realistic is that, really? I suppose an app can 'guide' you through learning it with a kind of self-tutorial, but that takes a huge amount of work.
posted by delmoi at 5:59 AM on March 19, 2012


It seems to me that it ought to be possible to build an LCD display with green phosphor behind the green subpixels, red phosphor behind the red subpixels and blue phosphor behind the blue subpixels, dispense with the filters, and backlight the thing with ultraviolet to make the phosphors glow. I don't know if anybody is already doing this, but I would expect it to save quite a bit of battery power.
Interesting idea, but presumably prices on OELDs are going to keep going down, I suspect people probably expect that those are going to take over for LCDs over time.
posted by delmoi at 6:05 AM on March 19, 2012


You expect a program to be able to do everything you want, and you don't want to have to learn anything before you use it?

Does iPhoto for iOS even have a manual?
posted by smackfu at 7:19 AM on March 19, 2012


The thing I like about OLEDs is that black is black (y'all). OLED displays use no power to display black and contrast is infinite. It just makes a lot of sense to me.
posted by asok at 7:40 AM on March 19, 2012 [1 favorite]


...contrast is infinite.

I'm not sure that I want pixels with the brightness of a supernova, although it would be nice to know the power was there, in case I need it when merging.
posted by StickyCarpet at 7:50 AM on March 19, 2012 [2 favorites]


Interesting idea, but presumably prices on OLEDs are going to keep going down, I suspect people probably expect that those are going to take over for LCDs over time.

I imagine that this is the expected replacement path, though there appear to be many other new display technologies that could replace LCDs too.
posted by ZeusHumms at 7:53 AM on March 19, 2012


Don't OLEDs have the problem of oversaturating certain colours (reds and greens, IIRC), leading to unrealistic (though, some would argue, attention-grabbingly zingy) displays?
posted by acb at 7:58 AM on March 19, 2012


It seems to me that it ought to be possible to build an LCD display with green phosphor behind the green subpixels, red phosphor behind the red subpixels and blue phosphor behind the blue subpixels, dispense with the filters, and backlight the thing with ultraviolet to make the phosphors glow.

The problem is UV production isn't terribly energy efficient. Further, it's still a two-step process anyway, electricity->UV->visible light with losses in both steps. Double-ungood.

If you take the next step and remove the inefficiency of the UV->visible phosphorescence, and just directly stimulate visible light with electricity, you have an (O)LED screen. This is the problem than OLEDs solve. OLEDs are usually touted as about 50% to 75% more power efficient than LED LCDs.

Normal electrolumenscent/fluorescent OLEDs approach 20% (quantum) efficiency today (though blues are still in the 5% range), with a theoretical maximum of 25% for common chemical substrates. Phosphorescent LEDs, (PHOLEDs), however, can be really efficient, and are effectively 100% electric power-> light in the lab right now. The problem is that the chemistries I've seen for PHOLEDs are fairly exotic, Iridium organometallic complexes, which are hard to produce and would have high materials costs. I'm not certain either if a full range of colours has been demonstrated with PHOLEDS yet. I don't know about durability either, a big problem with OLEDs in general.

Still, there is amazing potential for increases in LED efficiency.
posted by bonehead at 8:52 AM on March 19, 2012


So why do I care about this? I just want more than two fucking bars so I can make my damn phone call.
posted by Fizz at 11:52 AM on March 19, 2012


So why do I care about this? I just want more than two fucking bars so I can make my damn phone call.

I have no idea why you care about this enough to be commenting here. Why don't you tell us?

I can tell you why I care about this. I do browse the web on my phone, and use it for lots of other things besides phone calls. That's not why - I can tell from reviews and from looking at it with my naked eye how good the screen looks. These I care about because of geeky technology "how does this stuff work?" curiosity. I don't really think that these are that useful otherwise. For most people there's not much tangible benefit to understanding the reasons different devices' displays look different, or how some differ even though they look similar.
posted by aubilenon at 12:20 PM on March 19, 2012


To add to the self link extravaganza -- here is a 200x image of the IBM T221, my 22" desktop "retina resolution" monitor from long before Retina displays were cool.

USB microscopes are lots of fun!
posted by autopilot at 4:07 PM on March 19, 2012


it's still a two-step process anyway, electricity->UV->visible light with losses in both steps. Double-ungood.

My thinking was that it should be about three times as efficient as a standard fluorescent backlight plus filters LCD display, because it's not wasting two thirds of the UV->visible conversion step on generating light that's doomed never to pass any filter. I would also have thought that laying down an RGB pattern of phosphors shouldn't cost any more than laying down an equivalent pattern of filters, so a panel built this way shouldn't be expensive. But if OLEDs are better still, bring 'em on.
posted by flabdablet at 4:23 PM on March 19, 2012


It's a good thought. Phosphorescence generally has higher QE than fluorescence, but I think the double conversion, in particular the UV generation, makes it problematic.
posted by bonehead at 4:40 PM on March 19, 2012


I agree that the double conversion is inherently wasteful. It's the same process that goes on inside a standard fluorescent backlight, and if I recall correctly, the overall conversion efficiency of a fluorescent lamp is around 20%.

My point is that doing a 20%-efficient conversion from electricity to a mix of red, green and blue wavelengths, then allowing only a third of that energy through each subpixel's color filter, is extra wasteful and cuts overall panel efficiency to under 7%.

Putting the UV-to-visible conversion phosphors exactly where they're needed instead of having them inside a backlight tube and then filtering out two thirds of what they emit should triple a panel's overall efficiency, putting it back into the same 20% ballpark as a standard fluoro tube.

I knew nothing about OLEDs before thinking about this, but if they can do better than 20% for less cost, they're obviously a better thing.
posted by flabdablet at 6:07 PM on March 19, 2012


Oh yeah, and it would make LCD panels suffer from burn-in just like CRTs do, and make screensavers pointful again.
posted by flabdablet at 6:11 PM on March 19, 2012


Screensavers would not be useful again, now that we've figured out how to just turn off the screen when nobody's looking at it.
posted by aubilenon at 11:54 PM on March 19, 2012


(or rather, after a few minutes of nobody moving the mouse)
posted by aubilenon at 11:54 PM on March 19, 2012


(I wrote the linked article.)

StickyCarpet wrote:

A similar USB microscope scope that I have, from a different maker, doesn't allow the LED illumination to be turned off, and doesn't allow the auto-expose function to be disabled. That explains why the author has to make excuses anti-glare screen.

LED illumination is turned off, but you're right that I can't disable auto-exposure.
posted by L_K_M at 11:26 AM on March 20, 2012 [1 favorite]


Incidentally, this stuff is basically a pathologically bad case for digital photography. Auto-exposure typically optimizes for the mid-tones, but these shots don't really have much in the way of mid-tones. USB microscopes probably use the main sensor for auto-exposure, but if it didn't (DSLRs don't), a lower resolution or poorly focused AE meter would, for these images, come up with "white, 1/3 intensity" when looking at full-power R,G,B pixels, leading to overexposure of all the actual "on" subpixels. And the dynamic range is super high to begin with - apple advertises a 800:1 contrast ratio on the 4S. Furthermore, because of the camera's mosaic, resolution suffers as well - full on red only exposes 1/4 of the pixels in a typical sensor (they're often RGBG).
posted by aubilenon at 4:27 PM on March 20, 2012


« Older Turkish football fans have probably kept many flar...  |  Neil Gaiman writes a poem abou... Newer »


This thread has been archived and is closed to new comments