Ford is halting Mustang production

[Gregs24]

Many of you may also remember the news about Huawei, who were using semiconductor devices to spy.

Never actually either proven nor demonstrated to be happening. It is easy to accuse of 'spying' but when asked to explain what that actually means in practical terms there is no smoking gun. The real concerns lay in the company ownership and how it uses IP and data that many other companies also have and collect. Data collection is part of our normal lives and without it many useful devices couldn't function. It isn't all automatically bad.

Sponsored

 

[Gregs24]

You have a lot to learn about JIT manufacturing. We used it at Panasonic back in the late eighties to reduce high inventory levels and reduce the cost of selling it off.

https://en.wikipedia.org/wiki/Just-in-time_manufacturing

It is amusing how people who constantly crave cheaper products bemoan when JIT supplies occasionally fail. The cost savings are huge and have directly allowed products to get cheaper in real terms. Think of the storage space alone for parts for a day's production at a single car plant - and if you want to avoid JIT delays then you would need several days stock - the storage space would be bigger than the manufacturing plant !

 

[Gregs24]

While I agree with you (shocking!), didn't the UK also ban Huawei equipment

Politicians are dumb, no matter what their affiliation or location.

Yep. Moronic politicians here as well, making decisions based on...... well I have no idea to be honest !

 

[RazzaRossa]

There is a lot of incredibly complex equipment for manufacturuing chips. I worked for TRE Semiconductors back in the 80's. We manufactured most of the equipment on the photolithography end. Pattern Generators up to 15" x 15" reticles (glass like), Image rRepeaters that made master patterns for manufacturing the "wafers" on Wafer Steppers. From there the wafer is diced into individual chips for placing into the package you see on circuit boards. Mind you this is 1985, we were producing machines that had sub-micron trace widths. Then there are the light sources and lenses to achieve these accuracies. Zeiss lenses that ran 500k, just the lens. Now the positioning stages, drives and electronics plus a class 10 self contained clean room for each piece of equipment. We did not manufacture any of the silicon ingots, or sputtering (coating) equipment or the final packaging equipment. Add that all up, plus highly competent operators, AND chip circuit designers, you are pushing 10's if not 100's of millions for one line, again circa 1980's. I've lost track of the advancements in the equipment over the years and can only image were the technology is now. I believe there are at angstrom level trace widths and chips pushing multiple layers, perhaps as many as 10. Can you imagine the cost for a "scrap" wafer when something goes wrong in the process at layer 9 or 10? I still blown away at what we were doing some 40 years ago. The PHD's who thought this $hit up were/are technology gods, geeks to the nth degree.

 

[Hack]

There is a lot of incredibly complex equipment for manufacturuing chips. I worked for TRE Semiconductors back in the 80's. We manufactured most of the equipment on the photolithography end. Pattern Generators up to 15" x 15" reticles (glass like), Image rRepeaters that made master patterns for manufacturing the "wafers" on Wafer Steppers. From there the wafer is diced into individual chips for placing into the package you see on circuit boards. Mind you this is 1985, we were producing machines that had sub-micron trace widths. Then there are the light sources and lenses to achieve these accuracies. Zeiss lenses that ran 500k, just the lens. Now the positioning stages, drives and electronics plus a class 10 self contained clean room for each piece of equipment. We did not manufacture any of the silicon ingots, or sputtering (coating) equipment or the final packaging equipment. Add that all up, plus highly competent operators, AND chip circuit designers, you are pushing 10's if not 100's of millions for one line, again circa 1980's. I've lost track of the advancements in the equipment over the years and can only image were the technology is now. I believe there are at angstrom level trace widths and chips pushing multiple layers, perhaps as many as 10. Can you imagine the cost for a "scrap" wafer when something goes wrong in the process at layer 9 or 10? I still blown away at what we were doing some 40 years ago. The PHD's who thought this $hit up were/are technology gods, geeks to the nth degree.

It's amazing how discoveries and technology keeps building on itself. Every day people are coming up with new ideas and advancements are made.

 

[Hack]

While I agree with you (shocking!), didn't the UK also ban Huawei equipment

Politicians are dumb, no matter what their affiliation or location.

I believe that you are correct about the UK doing something with regards to Huawei.

However on your second remark, I doubt that bans on Huawei originated from any politician. A politician would take the donations and look the other way. Someone with serious pull had to be involved to get this kind of thing to happen. And I would bet there are smart people involved.

I'll just put it this way, modern lithography can no longer rely on the width of the laser to create features. The laser beam is too thick! EUV relies heavily on complex sets of masks, with the full silicon feature being completed in multiple stages. Surprisingly, they've gotten very good at keeping chip yields high. The number of defective chips on each wafer (for a mature node) is in the single-digit percentage, and many defective chips wind up being salvageable as lower tier products. I don't think any whole wafers ever wind up getting scrapped.

Just an FYI - there are different kinds of lasers. Depending on the lasing medium the laser light will be a different light frequency and therefore be capable of producing different feature sizes. You probably realize this already but just making the point that it's not as simple as "laser beam is too thick".

 

[Gregs24]

Cisco says computer chip shortage to last six months - BBC News

Bit more info

 

[RazzaRossa]

And for the most part our mainstream pattern generators used a mechanical means to create the R C networks, 2 sets of miniature carriages 90deg from each other. Vary the carriages widths and you can create squares and lines. Mind you this was done 500x , reduced down to 10x on the image repeater, and another down on the stepper 1:1. Whole wafers can, and are scrapped if the coating process is hosed up in a number of ways. Contamination comes in a lot of different forms.
I did a lot of work on their LPG (Laser Pattern Generator). Again initial pattern is done 100x and reduced thru process. It took a single HeCd beam laser (UV spectrum) and we spilt it 12x creating a ganged offset pattern, then the as stage was stepped X, the beams were rastered as it moved, stepped down Y and repeated until complete. One problem was at the end of each X move the 12 beam pattern left what we called "rat bites". Not good for electron flow. Took a while but they figured out how to vibrate the beams at that point to effectively smooth out the edges there. A lot less moving parts with lasers. Anyhow, I'll let this get back to Mustangs not being produced...

 

[RazzaRossa]

Sorry, one last point. TRE was bought by a group of Silicon Valley Venture Capitaists, put together by a young guy named Greg Reyes, ran it for 2 or 3 years. Company was called ASET, American Semiconductor Equipment Technologies. During that time a group of representatives from a Japanese company called Canon toured the facility taking pictures of everything (who ok'd this I don't know) 2 years later Canon introduced their own equipment line, and cheaper too since, well, they surprisingly make their own lenses. We survived maybe 2 more years and kaput. The rest is history.

 

[Bikeman315]

And for the most part our mainstream pattern generators used a mechanical means to create the R C networks, 2 sets of miniature carriages 90deg from each other. Vary the carriages widths and you can create squares and lines. Mind you this was done 500x , reduced down to 10x on the image repeater, and another down on the stepper 1:1. Whole wafers can, and are scrapped if the coating process is hosed up in a number of ways. Contamination comes in a lot of different forms.
I did a lot of work on their LPG (Laser Pattern Generator). Again initial pattern is done 100x and reduced thru process. It took a single HeCd beam laser (UV spectrum) and we spilt it 12x creating a ganged offset pattern, then the as stage was stepped X, the beams were rastered as it moved, stepped down Y and repeated until complete. One problem was at the end of each X move the 12 beam pattern left what we called "rat bites". Not good for electron flow. Took a while but they figured out how to vibrate the beams at that point to effectively smooth out the edges there. A lot less moving parts with lasers. Anyhow, I'll let this get back to Mustangs not being produced...

 

[RazzaRossa]

Yeah , all that experience in technology, and it took me 2 hours to change my horn

 

[Matthewstorm]

It's amazing how discoveries and technology keeps building on itself. Every day people are coming up with new ideas and advancements are made.

You can thank reverse engenieered Alien technology for that.

 

[Hack]

To put the mergers you mention in perspective, TSMC has a market cap of $568 billion. The industry is also consolidating because of the huge cost barriers to remain competitive. Smaller chipmakers simply can't afford to keep up with changes in technology. Their only choices are to keep making chips on older, increasingly obsolete fabs, or to merge and remain in the competitive space.

It's always been funny to me that the "brains" of the operation - the companies that design the semiconductors - have ended up being smaller than the "burger flippers" the big fab companies who grind out chip after chip repeating the same processes over and over. I'm sure it's my lack of understanding of how things are working.

You can thank reverse engenieered Alien technology for that.

I do. Every day.

 

[WaltA]

My employer had to shut down our semiconductor line, years ago.

The problem is that even if you were able to run at 100% capacity, 24/7, that only gets you to the break-even point. The root problem is that equipment goes obsolete very quickly, and you need to run the heck out of it, before it does go obsolete.

Add to that, that it is a very expensive investment. Not many companies can go it alone anymore (even IBM falls into that).

When my employer shut down our fab, we offered it for free to any university who would want it, with the caution that just routine maintenance alone is around $1M a month. There were no takers, and they ended up tearing it down.

 

[Bandit2277]

Ok. Alright. We are gonna – we’re gonna go out there during this break and we are gonna come back with a plan. We’re gonna come back with a plan for you. It’s a 45 day plan. 45 days to get us back on track. 45 points. It’s a 45 day/45 point. One point per day. We get the 45 points we are back in business. - Michael Scott - Boom problem solved

Sponsored