Analog Reverse Engineering

Analog Reversing

I was just thinking of reversing the old 339 comparators (yes the datasheet gives full schematic) to practice on a production IC. Anyone ever done these, or other simple IC's?

The discussions with rambus in the other analog topic made me think quite a lot over this - IC copyright is dedicated to the mask, and the copyright law contains a section explaining that reversing your own IC is perfectly legal as long as your structure and layout are unique.

Anyone interested in a reverse-engineering topic? I start with die photos too, so anyone who can write a c program to unpack photos to gds is very needed. I took one class in digital image processing, and finding shiny metal 1 edges is very easy in comparison to reading the alphabet!

Anyone care to lend their help in a standardized reverse-engineer library? Natonal Semi vs. Intersil vs TI vs Linear Tech vs. Maxim vs. Micrel vs. anyone have some reversed schematics to start a category?

We are a small group, but with a lot of talent - maybe elektroda should start a fabless IC company!

Re: Analog Reversing

Some time ago I saw a paper that talked about layout reconstruction from an image. I'll try to find it again and handle it to you.

By the way, have a look at www.chipworks.com.

What you say about creating a design house is interesting

Re: Analog Reversing

Yes, Chipworks is one of them. Another one is Semiconductor Insights www.semiconductor.com also in Canada. They both do reverse engineering. Some 5-6 years ago it was done like this. A chip package was opened, then they were making photo pictures of every layer - metal 5, then they strip it and make a picture of metal 4, then strip it again and make a picture of metal 3 and so on till they finish with the difusion layer. Then those pictures were printed on very large photo paper - covered considerable portion of the floor. And after that the poor engineers had to crawl on the difusion picture, identify the devices and restore the connections as they appear on the pictures of the upper layers. After that they extract the schematic. Of course, since those times they have developed image processing software, so now they don't have to work on the floor, but on the screen of a workstation. But I think the procedure of extracting the schematics is the same. You may find this work interesting, especially in the part when you have to organize and identify the function of the extracted schematic.
I also think that some of the big companies also support a small group doing reverse engineering.

Analog Reversing

This is great stuff - I didn't know you could make such a high-profile company out of attacking the new intel prescott! I thought it was still grey-area!

Yes - I know many companies that have RE groups, but usually it's just a group of designers that tear parts down from time to time when a fast product is needed to compete.

Re: Analog Reversing

I think that rverse a large logic circuit is not a good method for designer. But reverse a analog circuit is a good method., because analog circuit often is only less than 100 transistors.

Re: Analog Reversing

It may not be even 100 transistors. It can well be 10, or even just couple or so. Extracting the circuit is not a problem; understanding how it works, sometimes is and this is what is meant by reverse engineering.

Re: Analog Reversing

I think,
it's need too much efforts to reverse modern IC from 0.25 um process and higher (TSMC, UMC provide 0.18um RF process).
You will need very advanced optical microscope or even electron microscope and care about many other things.

So, complete reversing of modern IC almost impossible.
But you can reverse some critical parts of IC.

Re: Analog Reversing

yes i suggest top down