Die shot legend image database

Is there any image database in existence that teaches how to decipher die shots? Something that shows photo snippets of what different transistors you might encounter will look like on the die, and explains what kind they are, and where the base, collector, and emitter are.

Ken Shirriff’s blog posts are very detailed, illustrated, and contain footnotes and cross-references. Not a tutorial, but highly recommended. He’s analysed the 6502, Z80, 8008, 8086 and many more.

He covers many different types of chip: you might want to start with NMOS. (PMOS, CMOS, TTL, I2L, are all different.)

Here’s a presentation of his, 95 pages of PDF or 30 mins of video:

(also here - see video description for more links)

1 Like

Yeah, something like what appears for the NPN and PNP transistors immediately after the “What bipolar transistors really look like” page of the supercon PDF, but enough images to RE an entire chip. I suppose those two transistors for those slides are from the TMS 0805. Since both PNP and NPN transistors appears there, can I conclude that it’s a CMOS chip?

For the 65c816, I would need to know how to RE both NMOS and PMOS.

Not quite, those are from a TTL chip, the 555, so that’s bipolar.

The 0805 looks like a MOS chip - I think it’s the earlier metal gate kind of process, whereas the '816 will be a polysilicon gate process. So the layout might look a bit different.

CMOS logic will generally have complementary pull downs and pullups for each logic gate, so for example a 2 input NAND has 4 transistors, as does a 2 input NOR.

I don’t think I’ve ever seen high resolution photos of the '816 die.

For low density CMOS logic, perhaps try

Ah, ok. The '816 die shots we have are nowhere near that pink, so it sounds like I have to rely on recognizing shapes. How did it get so pink?

Funnily enough the colours come from the thickness of the layers - for some processes polysilicon is green and active area (diffusion) is red, and for other processes it’s the other way around. But the characteristic shapes you find on those two layers are quite distinctive so you can generally figure out what you’re looking at.

1 Like

I feel like the next step is for me to read the Baker CMOS textbook. I’m not going to be sure I RE’d something correctly unless I have a way to simulate it.

1 Like