Hacker News Reader

”So say you want a 4kx2k display, well, that is 8 million LED's”

24 million, if you want RGB. On the positive side: OLEDs have the same number of LEDs, and we can manufacture them economically and at large sizes (LG showed a 88 inch 8k one in January (http://bgr.com/2018/01/01/lg-8k-oled-display-announced-88-in...)

That'd be true for large displays, but not for phone OLED displays.

Phone OLED displays seem to have only 2 LEDs per pixel.

See:

https://en.wikipedia.org/wiki/PenTile_matrix_family#PenTile_...

http://bgr.com/2017/11/07/iphone-x-display-microscope-oled-v...

LG OLED is not true color OLED, but a hack, having a white OLED emitting panel and a color filter on top. It has many downsides like lower efficiency, maximum brightness and color accuracy.

Sort of like an LCD?

Yes, though it still improves on the many problems of IPS and VA LCD technologies.

does it have true blacks at least?

Since it doesn't rely on the panel trying to block a very bright backlight, yes.

However, because they're all the same color all of the OLEDs degrade at the same rate so the color balance doesn't shift over time.

>Seems like the initial deployment of the microLED's into a smartwatch makes some sense.. But unclear how this would make any sense for a large size phone display. The sheer quantity of LED's that must be placed properly on a glass substrate would be quite high.

How would it be different than producing screens for e.g. 10 watches (or what the analogous real estate is)?

I guess it would be about the ratio of the price for the screen compared to the device. Smartwatch at 600$ with a small screen vs. a smartphone at 1000$ but with a 10x bigger screen. With 10 smartwatches I would have 6000$ to spread the additional cost over.

[edit] typo

Assuming the same (as current) prices for iPhone and smartwatch, that would be the same ratio though if we divided by the current models, so it doesn't tell us much regarding how might the new kind of screens affect price.

Only something that took into account the relative cost of each type of screen would (and, if screen area is a non 1x linear factor of cost, that too).

Every pixel on the panel must work, otherwise the whole panel is useless. Now imagine you have one broken pixel on average for every million pixels. Apple Watch screens have 100k pixels so on average 9 out of 10 screens test OK and can be used to make a watch. But the iPhone has about a million pixels, so you end up throwing away about half of all the screens you make.

More pixels = lower yield.

Probabilities works out even worse. For million pixels roughly 37% of screens would turn out OK.

Could smaller adjacent panels be coordinated to appear imperceptibly as if they’re a single panel?

Potentially, that's how led billboards are assembled.

Somethings off. It should work like this:

  p = 1e-6
  n = 10^(1:7)

  frac_accepted = (1-p)^n

Results:

  num_pixels frac_accepted
      1e+01     0.9999900000
      1e+02     0.9999000049
      1e+03     0.9990004993
      1e+04     0.9900498288
      1e+05     0.9048373728
      1e+06     0.3678792572
      1e+07     0.0000453997

This assumes each failure is independent. Often a single failure will cause many bad pixels on the same screen. So you might have 98 good screens 1 with one bad pixel and one with 5,000 bad pixels when the per pixel odds say you are going to end up with 1 good screen if you are lucky and 99 bad ones.

I simulated 100 screens with different numbers of pixels where the probability the first pixel is bad in a screen is p0 = 1e-6 and the presence of each bad pixel increases the probability of the next one by the same amount (ie probability for a second bad pixel is 2e-6 and for last pixel is p0*number_already_bad_px). A screen was accepted if it had 0 bad pixels. The results look pretty much the same:

  num_pixels frac_accepted
       1e+01          1.00
       1e+02          1.00
       1e+03          1.00
       1e+04          0.98
       1e+05          0.89
       1e+06          0.32
       1e+07          0.00

Of course this was just a convenient scheme to simulate. The greatest number of bad pixels was 86k for the 10 million px screen (so chance the last pixel was bad was ~9%). It does show that the independent pixel model isn't necessarily very far off though.

You messed up the math. You need to count (# number of pixels that where bad) / (total number of pixels produced) not simply call the odds for the first bad pixels as the rate of bad pixels.

You are saying to adjust p0 so I get that the frequency of bad pixels = 1e-6? That makes sense but it would probably now depend on what size screens are being considered... not sure if I'll go back to it.

I didn't actually calculate the probabilities, it was just my off-the-cuff estimates. 37% could be 'about half' :P

>Every pixel on the panel must work, otherwise the whole panel is useless

And that is not so far fetched. The most high yield semi processes can give you defects in single digits for a fully exposed 300mm wafer.

The challenge here is that they will have to be grown on a substrate different than silicon for anything bigger than apple watch screen.

Growing inorganic semiconductors on glass, organics, non-crystalline materials is what is hard

lol these things aren't going to be priced per LED. Just like you don't pay per transistor when you buy a chip, you're not going to pay per diode when you buy a microLED display.

The costs scale with device area, not device count. This is because these devices are made with lithography on a wafer. A single silicon wafer would produce 5000 diodes, or it may produce 5,000,000, depending on how small you make each one.

Of course, so let’s assume a 5” display, and assume that the fill factor per LED color is 25% (rather than expected 33%) then 25% of the area of a 5” display would be required to populate one color. So assuming a small dicing kerf let’s assume for a 12 inch wafer you get 8 displays from that for one color. Which feels pretty generous, normally that same vendor would have yielded millions of LEDs from that same wafer and they made say 10000 per wafer, then if you asked them to sell for far less than that, what is their incentive? If they did that they would essentially be crashing the entire LED market in one fell swoop and their own profits. So the die area argument makes sense from apples perspective but doesn’t from the entrenched LED makers perspective. So seems like apple essentially would need to own the actual LED fab as well.

> If they did that they would essentially be crashing the entire LED market in one fell swoop

Would they? A one-subpixel microLED isn't going to be able to be used as a regular LED. It's not large or bright enough.

interesting point but for a given luminance in nits the area doesn’t matter since it is cd / m^2 so assuming that there is a “cheap variety” of LED wafer material that outputs say 2000 nits for a low current, then a small area — say 20x20um for the subpixel size would save some cost for sure. I would estimate about the area of the subpixel would be about 30 times smaller than most indicator type low cost LEDs — which are in the sub 1mm^2 area range. So we can find the 30X factor but still need around 300X cost reduction still to get a 10000X reduction in price. Ultimately the Display should be around $100 or less, so if you can buy an entire wafer for $800 or so then you have something, but as far as I know most 12 in wafers with zero processing done to them are more like 1k. Then put them thru a fab and add value and run the fab and the cost goes up a ton. And we need 3 different wafers.. so anyway, I’m worried cost is just going to really hurt this approach for a long time.

> the area doesn’t matter

Producing the same luminance using a smaller LED is equivalent to running the same current through a thinner wire. At some point you're just going to be burning out the LEDs instead of powering them.

Given that you respond to that problem by lowering the current to within the smaller LEDs' reduced tolerance, then you get something too dim to use as an indicator LED.

The cost of the equipment to produce the wafers grows with the density as well as the area.

Yeah, but it can be reused. We're talking about large scale fabrication here.

Oh shit, I've been doing it wrong. It's been really expensive.

It depends on whether the display is assembled with a monolithic microLED wafer or a microLED transfer process. I believe Luxvue which Apple acquired was focused on the latter. Transfer cost and rework cost do scale per LED. Additionally, defect rate most likely does scale per LED to some extent in addition to per area.

There are probably manufacturing processes where LED transfer costs don't scale per LED. For example, they could use some type of expanding substrate material as part of their transfer process.

Great job describing some of the real advantages and manufactuting challenges of microLED displays. The 320x320 pixel wearble displays seem like a place to start with only 100k pixels (300k sub-pix). With that number of pixels they'll be able to have decent yield at 1ppm defect rates. The power, brightness, and lifetime arguments are even more compelling for a watch.

I don't think the cost side is so bad. LED wafers are only a few hundred $ at 8in for over 3M 50x50um2 die so ignoring other processing and placement costs they are in the right ballpark. Assuming they use some PWM or passive matrix method that keeps the power control substrate cost/yield reasonable they could have a very high performance square inch sized display for a reasonable price... soon?

For any larger displays the dicing and KGD placement yield become a huge issue. Ultra-reliable reworm and/or uncorrelated defects become critical when you need to approach 10ppb for reasonable yield at WQ.

Didnt Sony demonstrate such display in a TV once? Yep in 2012, so did Samsung recently http://www.zdnet.com/article/ces-samsung-unveils-modular-146...

There's a substantial difference in DPI, which probably makes it easier to make this big screen than an iphone screen with near the same resolution.

Yea for the big TV's I think its basically like pick and place for the LED's.. the TV is essentially a giant PCB covered in around a million or so LED's.. which is pretty ridiculous that they tried that, but pretty awesome anyway..

>The sheer quantity of LED's that must be placed properly on a glass substrate would be quite high.

They will be simply grown on the substrate in-situ. A very different epitaxial pricess from bulk leds, but nevertheless very doable.

Apple is known to have bought a Taiwanese company that was making monochrome microled screens for night vision goggles.