How IC are designed by engineers

[RITESH KAKKAR]

Hello,
How the IC are designed i dont understand less than 1mm is very less visible how engineer design less than 1mm wires circuits?

 

[betwixt]

They are not designed at real size, everything is done by CAD systems then scaled down to required size to make photo masks for production. In the world of IC's 1mm is HUGE!

Brian.

 

[RITESH KAKKAR]

ok, which computer aided design is used?
and if 1mm is huge then how they work in less than 1 mm in real we make transmission line of meters easily how engineer pic less than 1 mm wires?

 

[Arya Kumar]

Seems like you are new to this world. Every1 knows they are not hand made. They are mde by costly machines which can go down to molecular level to make chip designs

 

[RITESH KAKKAR]

i know they are manufacture from machine but how uin small size.

 

[udhay_cit]

https://www.youtube.com/watch?v=oa3DZUj9Q-s
https://www.youtube.com/watch?v=AIjGRzNIWC4
https://www.youtube.com/watch?v=TOZ9CsOi-xc

This may help you...


Udhay

 

[RITESH KAKKAR]

how we get silicon wafer who invented that silicon is good for designing's electronics?
from where silicon is collected for chip
i have seen people use silicon for panel avoiding rain water.

 

[dick_freebird]

One nice thing all of those mysterious chips provide you is
access to very dense information like Wikipedia, where you
would find the answers to each of these questions without
the wait.

I once made a chip "by hand" back in the early '80s. Real
hand drawn pencil schematics, green ASCII terminal SPICE
netlist typed in text editor, rubylith and X-acto, projection
aligned by hand through optical microscope, dip-n-drip
photoresist and wet bath etches. An NMOS-only process,
and the N+ dep and diffusion oven I wasn't allowed to touch.

Worked not entirely but somewhat. Lousy yield and 2" wafers.

Good enough for an A-. Gotta love the curve. And maybe it
was all the fab's fault anyway.

Just let the robots do it. Serves them right for taking over.

 

[RITESH KAKKAR]

Hello,
Can you show that?

 

[KlausST]

Hi,

Back in the early 80s the chip structure was so huge (compared to nowadays designs) that you could see it with a usual microscope.
Chips were copied by "reading them out"with a microscope.

The structure were many micrometers in size.
Nowadays the structurea are a few (14) nanometers.

Wikipedia talks about 9 million transistors per square mm.

Btw: 1000 nm (nanometers) = 1 um (micrometer)
1000 um = 1 mm (millimeter)

If one wire is 14nm in with and you expect the same as isolation (don't know if this is realistic) then this means more than 35000 wires per mm. It really is much, much smaller than a millimeter.

Klaus

 

[RITESH KAKKAR]

9 million transistors per square mm.

Hello,
it is realy huge what do so transistor do?

one wire is 14nm in with

How these wire are made so small they will be damage in hands

how invented this technology?

 

[betwixt]

Please read the articles in the previous links, they are not wires and transistors like you buy in a store, they are optically projected on to sensitized silicon then chemically etched. Some newer ICs use connections and transistors so small they have to be electron scanned because their size is smaller than can be achieved using optical methods.

Brian.

 

[RITESH KAKKAR]

ok, but how we see in so small nano meters transistor ?

can we also build them?

 

[betwixt]

I'm not sure I understand your question.

Using electron microscopes we can see much smaller than that. Even my old (probably 50 years old!) optical microscope can see individual structures in most ICs.

Of course we can build them, they exist! But, it takes very expensive precision machinery to do it.

There was a factory close to where I used to live that made 'Transputers', a kind of RISC processor. They had difficulty making them acurately enough because vibration from passing cars along a highway over 200m away was shaking the machines!

Brian.

 

[RETGT]

Using electron microscopes we can see much smaller than that. Even my old (probably 50 years old!) optical microscope can see individual structures in most ICs.

Of course we can build them, they exist! But, it takes very expensive precision machinery to do it.

There was a factory close to where I used to live that made 'Transputers', a kind of RISC processor. They had difficulty making them acurately enough because vibration from passing cars along a highway over 200m away was shaking the machines!

Brian.

Can you post the picture of it?

 

[dpaul]

This is an old post: Last reply was 4th July 2016

RETGT should learn so search and filter out results himself/herself.

 

[betwixt]

Agreed dpaul.

For completeness, here is a photograph of the factory. It now says "International Rectifier" but at the time I mentioned the vibration problems it was owned by "Inmos" and making Transputer ICs.



If you want more information on the IC itself: https://www.cs.bris.ac.uk/~dave/transputer.html is the designers web site.
Take a look at the top right corner of the silicon die image - even IC designers have a sense of humor!

Brian.

 

[RETGT]

The company is very very big..

 

[RETGT]

https://en.wikipedia.org/wiki/Transputer

what is transputer is different from other?

 

[betwixt]

You think that is big! Compared to some places I have worked that is VERY small! I have worked in places where there is a 24 hour bus service to carry staff around INSIDE the buildings!

If you want to see the whole factory, unzip and click on the attached Google Earth link. I'm glad I work in electronics and not plumbing!

There was nothing special about the Transputer compared to todays processors but in it's time (early 1980s) it was one of the first RISC designs to hit the market. The conventional microprocessor market was heading towards ever more complicated ICs which meant lower manufacturing yield and because of the silicon technology at the time didn't allow high speed, they were quite slow. The first IBM PC for example, used the most advanced processor at the time, the Intel 8088 and it ran at 4.77MHz!

What made the Transputer different was it used a much simpler 'modular' approach. Each Transputer was relatively uncomplicated but ran at high speed (as much as 20 times faster than the 8088!) and they were designed so they could be linked to together to share tasks. I worked on systems with maybe 100 Transputers linked together (I never counted the total) and they were MUCH faster than could be achieved with conventional ICs. Their downfall was that they were left behind when manufacturing improved and the spped of more complex ICs caught up with them. As a stepping stone along the way to todays ICs they were invaluable and many of their innovative ideas are now part of standard designs.

Brian.