Comments by "Grak70" (@Grak70) on "Asianometry" channel.

  1. The idea that someone once thought combining mustard gas and anthrax was a step forward boggles the mind.
    607
  2. I lived and worked in Japan from 2010-2015 and visited Sharp sometimes for customer interaction. One of the most depressing and bloated organizations I had to interact with. And for Japan, that’s saying something.
    554
  3. Whoever marketed this as drinkable is a goddamn genius. It tastes like someone set dirty dishwater on fire.
    311
  4. They will, but Americans are not going to live in the fab for peanuts. If TSMC intends to expand further into the US, they need to get on board with the US office/fab culture. The Arizona fab is already having problems with this: they don’t want to compete hard on salaries and expect you to die in your bunny suit.
    200
  5. Fun story: the first full field functioning scanner ever made for 157 was manufactured by SVG (later acquired by ASML for their catadioptric lens designs). At the time, it was thought the entire lens train, including the condenser optics, would need to be made from ultra-pure calcium fluoride. It was later discovered that fluorine doped quartz would do the trick for all but the few final lens elements. At the time, that one scanner contained the largest concentration of lithography suitable CaF2 in existence. The projection lens was literally worth its weight in gold.
    141
  6. I’m sure the discomfort with automation that would eliminate labor was not lost on the management with a resident Party member standing behind them at all times either.
    87
  7. It’s incredible that for as many people as Motorola employed throughout its existence, I’ve never once heard a positive story about working there.
    82
  8. It’s moments like this that make me laugh when people claim laissez faire capitalism will always properly allocate resources. It may make progress in fits and starts, but it can also be obscenely wasteful.
    74
  9. Also, half the tools I work with on a daily basis are older than I am. And I’m over 40. Compound semiconductor makers almost exclusively inherited equipment from the 1970s-early 1990s silicon world because that’s all that’s available in many cases. What’s left are often 200mm tools down-converted to 150mm. With those kind of capital restrictions, an industry that charges cents per unit for its product and can fit 5,000-15,000 chips on one 6” wafer is never going to transition to 200 or 300mm.
    67
  10.  @devondetroit2529  living in Japan is awesome. Having to work there, especially in a corporate job, sucks.
    52
  11. You can just say “saw” and “baw”. Everyone else does. Source: work in this specific industry. Also 50M parts for $1.40 is a whopper untruth. I think you confused bulk pricing per unit with order pricing. A product run capable of covering an entire generation of smart phone wouldn’t cover the cost of a single photomask for said filter if $1.40 bought 50 million units.
    51
  12. Yup. I work in GaAs. Hard to make a Cz ingot when half the atoms in your melt sublime into vapor at atmospheric pressure.
    48
  13. ASML was and is huge in all previous UV step and repeat systems. They just have competition in that arena. There are ASML I-line steppers where I work that were built in the early 1990s with Zeiss optics stamped “made in West Germany”. ASML is famous now because they’re the only game in town for EUV and that makes them a strategic resource at a national security level. But they’ve been doing business in this space for well over 30 years.
    34
  14.  @roc7880 did you watch the video? This was done in the interwar period, well before nuclear weapons were a thing. And it was only 4 years after Hiroshima that the Soviets got the bomb so not sure what you’re on about other than trying to stir the pot.
    32
  15.  @ryandick9649  I agree completely. No lack of talent or good ideas, just smothered by excess bureaucracy. In a lot of ways, reminds me of the problems that plagued and ultimately killed Motorola.
    30
  16.  @Braydan789  yes, by design. Almost all of it is IP protected. And the process integration matters much more than raw capital capability. Most RF fabs are using equipment that was already considered old in the early 1990’s by silicon CMOS standards.
    30
  17.  @wernerhermann9120  jfc some people will start a pissing fight about literally anything.
    28
  18.  @Taygetea  a bit different today because demand is still ludicrously strong. To their credit however, this time most semi manufacturers resisted the siren call of overbuilding capacity. And now with interest rates the way they are, there’s no way more capacity that wasn’t already under construction is going to start.
    26
  19.  @PaulFisher ere are several sources of flare in DMD. 1) the material below the mirror itself does show through a bit since it has to be exposed for a mirror to be free of its adjacent companions. While this is typically coated in a material selected to be as black as possible over the desired wavelength range, it’s not perfect. So even if 100% of light directed at the mirror is diverted, there will be some reflected light from the substrate. 2) the mirror can only tilt so far. Ideally you’d direct all the light onto a light dump orthogonal to the exit pupil, but the mirrors can’t flex 90 degrees, only about 10-12. So some light will be scattered from the mirror edges and the anchor via structure, letting a little light through even in the static full off position. 3) ideally, your light source would have total spatial coherence (or equivalently, be a point source at infinity). Real light sources are only partially coherent and have finite size, so some photons impinge on the mirror at a small angle. When these are reflected in the off state, some small amount of light will still go through the exit pupil. 4) the light dump material is chosen to be as black as possible and actively cooled from behind. However in practice, a true black body does not exist, so there will be a tiny amount of reflection which will bounce around the DMD cavity. Some of this will escape through the pupil.
    25
  20. I actually work directly for one of the former Intel guys who killed 157. 😆 it’s a fascinating story.
    25
  21. In addition to the patent issues already mentioned, at least one of the two Japanese makers (I can’t remember which) bet the farm on DPP source tech rather than LPP. That was the wrong choice and they paid the price.
    25
  22. Yeah I laughed out loud at that. Based on that price, enough SAW to cover an entire generation of iPhone wouldn’t net enough profit to cover a single photomask.
    24
  23.  @ArawnOfAnnwn  correct, but they also ship dozens of twinscan platforms a year at all wavelengths. Most of them are I-line or KrF. Characterizing ASML as a one trick pony with EUV is just not true.
    23
  24.  @wernerhermann9120  Japan-chan is not going to sleep with you. Please find a creative hobby.
    22
  25. Micromirror and microaperture arrays are fascinating. One of the NIR spectrometers on JWST has a particularly cool one. And it’s the basis for all of Texas Instruments’ IP on DLP.
    22
  26. This is exactly correct. More cowboy engineering in action. The engineers who invented plasma etch used 13.56MHz because it combined RF generators that were commercially available off the shelf with license-free use.
    18
  27. This idea that you can run a company with hyperrationalism and no office politics is a pipe dream of folks who have never managed anything more complex than their next white paper.
    16
  28.  @ad2094  are you being pedantic or just not understanding the difference? I liked being there except for the work environment. Vacations were great. Weekends were always fun. If you’re independently wealthy, it’s a playground. If not, you spend a third of your life doing something soul crushing.
    15
  29.  @snorman1911  yeah and as someone else pointed out he says “more than what the bucket costs” then said he felt bad for the manufacturer’s profit margins. It’s just a dumpster fire series of sentences. lol
    15
  30.  @souvikrc4499  actually no. The power to control who ASML sells to through the EUV license is proving, at least so far, to benefit US national security interests, which was the plan all along. There wasn’t and still isn’t any company in the US with anything like the skills or capital to build out EUV infrastructure and produce tools.
    14
  31. Damn you are CRANKING the videos out. This is one I’ve been looking forward to.
    14
  32. An enormous amount of III-V production has never scaled beyond 0.5um. There’s simply no need in many cases, and the tools and processes used to create them aren’t capable of doing much better.
    13
  33. If a supply of raw materials is more expensive to sort for authenticity than to cosmetically reprocess and there is no downside to counterfeiting, the bad actor has no incentive to ensure quality. The same mindset that cuts baby formula with poison sure as hell isn’t going to care if you get a bad Atari sound chip.
    13
  34.  @Adamrc98  people have been smoking for thousands of years and that’s nasty as shit. Gross Chinese mud booze is gross.
    13
  35.  @TamagoHead  SRAM is definitely still in use; its applications are just different to DRAM. SRAM cells are significantly larger, complex (they use latching to store data instead of simple charge like DRAM) and must be continuously energized to retain data. But it’s stupid fast, so it’s usually only used for onboard cache.
    12
  36.  @akatsuki6371  you can definitely do 7nm and to some extent even 5nm nodes without EUV. The problem is your yield is crap and your cost is through the roof. What kazedcat is saying is you only use EUV on the layers you absolutely have to. Fabs still have a ton of I-line scanners because they can spit out like 250 wafers an hour with much lower unit cost and preventative maintenance cost if resolution isn’t an issue. And this is what they do; most lithography machines in a fab by count are still excimer KrF/ArF and I-line. Fab mantra is: “always use the cheapest thing that works.”
    12
  37.  @dxelson  yet somehow Intel manages to make billions a year? The fab I work for (in the US, not Intel) is decades old and still bonkers profitable. The idea that a cutting edge fab (or even one built in the 80s) can’t make money in the US because our engineers are paid significantly more than an engineer in Taiwan is laughably wrong.
    11
  38. It’s quite difficult to build a late 90s/early 2k PC because so many motherboards surviving from that era have Capacitor Plague.
    11
  39.  @curtlundgren6867  sounds about right. Off state black levels are a major reason DLP became obsolete as LC black levels improved and then LED off-current improved. Then came active LED, OLED, and then quantum dots with ever better black levels. There was a lot of room at the bottom for solid state light solutions, but DLP could never quite keep up. MEMS in general are much more finicky and lower yielding to fab, not to mention they face very different challenges to CMOS. TI repurposed a lot of silicon equipment to make DMDs and I think the ROI of using apples to make orange juice was difficult to justify in the long term, especially with such stiff competition from Japan, Korea and later China. Once China started pumping out decent quality LED displays that kicked even the best DLP’s ass, the margins in display evaporated.
    11
  40. You need bigger problems.
    11
  41.  @connormauk3210  I literally don’t give a shit what you think. I’ve been forced to drink this slop at multiple dinners in Taiwan and Shanghai. Moutai is fucking gross.
    11
  42. EUV had already surpassed anything they hoped to achieve with proximity X-Ray and e-beam is direct write and was never going to meet throughput requirements.
    10
  43. I assure you, stepper optics makers knew about immersion as a concept from the beginning. The problems were 1) why solve a bunch of novel engineering issues before you absolutely have to and 2) once you have solved them, how to convince fabs that having liquid in contact with their wafers during exposure wouldn’t cause killer defects.
    10
  44.  @ArawnOfAnnwn  ASML has had more market share in exposure tools than both Canon and Nikon (taken individually) since 2005. And it’s been beating Canon longer than that: the 1990s were mostly a duopoly between Nikon and ASML. You’re just making shit up at this point.
    10
  45.  @ChimpyChamp  if someone asked me to name a technology they thought came from aliens, it would be scanner stage positioning. You look at the requirements and can only conclude it shouldn’t work but it does. At a very simplified level, they use bands of closely spaced electromagnets to move the stage (the entire stage is levitating over a near perfectly smooth block of granite). The closed loop control uses an interferometer and powerful signal processing to feedback an electrical signal to those magnets that results in oscillation damping and acceleration control. The part of every stepper/scanner you don’t see is the rows of 2m tall control racks supporting it. They’re usually tucked away in the sub-fab, quietly and frantically sending out a flood of information to the control systems. It is truly mind boggling.
    10
  46. The fact that photo rework works at all is almost as much of a miracle as lithography itself.
    9
  47.  @rrai1999  “more than what the bucket costs” then saying “I feel bad for the saw manufacturers” then makes ZERO sense.
    9
  48. You can brute force a lot of things. We still don’t know what their yield or cost is. I expect it’s pretty poor. Not particularly problematic for SMIC, they can just dip into public money to make up for it until those things improve.
    9
  49.  @Daz912  I ran the entire division. But thanks for playing.
    8
  50. What are you even talking about. “Micron” is standard diction in the entire semiconductor and electronics industry.
    7