Where the chip meets everything else
Forty pins, and what the silicon says about each one.
This is the whole of the chip's contact with the world: a package, two rows of twenty, and a notch so you can tell which end is which. The numbering is a fact and is the one thing here taken from a datasheet. Everything beside it is read back out of the switch network, including the part a datasheet gives you as an arrow.
Loading the chip…
Click a pin. The label's colour is which functional block blocks.rs filed that signal under, the same palette the exploded view uses, so a pin here is the same colour as the region it belongs to there. Three of the forty are not connected to anything, and they are drawn rather than skipped.
Measured, not written
Which way a pin points, asked of the transistors
A datasheet gives you an arrow. An arrow is somebody's summary, and the silicon will tell you itself if you ask it the right way: a pin is an output if a gate drives it, an input if it feeds gates, and both if both. The data pins come out bidirectional without anybody saying so, which is the answer you want, because a byte really does travel both ways through them.
There is one subtlety, and it is the reason this is derived rather than transcribed. A gate whose every pulldown is wired to vss can never pull anything down: it is a pullup wearing a gate's clothes. Two pins have exactly that, and both are inputs, held high by a permanently-off transistor when nothing outside is driving them. Counting those as drivers reports two input pins as outputs, which is what the first version of this page did before the transistor was looked at.
The chip is warned about this elsewhere too. The die has a handful of transistors gated by ground, which are physically correct and permanently off, and they are exactly this kind of structure.
Measured
Every pin, resolved
The reach column is how many places a signal actually goes: gates that read it, plus pass transistors with a channel on it. It is why the address pins and the data pins do not look alike here even though a package drawing gives them the same weight, and it is worth comparing against the graph, where the same signals are drawn where they physically sit.
| Pin | Package | On this die | Direction | Functional block | Reach |
|---|
Two ways of counting
Forty pins is not the same as forty signals
The package has forty pins. Three are not connected to anything, and ground arrives on two of them, so the number of distinct signals the die actually names is smaller. Neither figure is wrong and they are not in disagreement: one counts legs on a package and the other counts wires on a chip.
It is worth saying plainly because the difference is the sort of thing that quietly turns into an error somewhere downstream. The published block diagram counts the same boundary a third way again, since it draws a family rather than one part, and shows a pin this die does not have.
Credit
Sources
The pin numbering is the one published in the MOS Technology hardware manual for this processor family. That assignment is a fact about the part; the descriptions beside it are ours, and every other column is measured here.
The die trace those measurements come from is by Greg James, Brian Silverman, Barry Silverman and Ed Spittles, and is licensed CC BY-NC-SA 3.0. Those terms carry to everything on this site.