1947

Transistor

Bitwise builds AND and OR out of relay contacts in 1937 and says plainly that NOT cannot be made from series and parallel alone: you need something that inverts, and a contact does not. Front Panel is a working machine in 1975. What happened in between starts here, at Bell Labs in December 1947, with a device that does two things a relay cannot. A small current controls a large one, so one output can drive several inputs. And wired to a supply through a resistor, the output falls when the input rises. That is the missing gate, and it is free.

One stage, and the two facts that made everything else possible

One stage: a resistor into the base, a resistor from the collector to a 5 volt supply. Move the input and watch what comes out.

Input voltage: 0.0 V

Current gain: 100

current into the base: 0.0000 mA
current through the collector: 0.00 mA
which is: nothing is flowing
output voltage: 5.00 V
which region it is in: cut off

the base is below 0.7 V, so nothing conducts

The same stage across the whole input range. The output falls as the input rises, and it never does otherwise.

Input voltage, base current, collector current, output voltage and which region the stage is in
in	base	collector	out	region
0.0 V	0.0000 mA	0.00 mA	5.00 V	cut off
0.5 V	0.0000 mA	0.00 mA	5.00 V	cut off
1.0 V	0.0064 mA	0.64 mA	4.36 V	active
1.5 V	0.0170 mA	1.70 mA	3.30 V	active
2.0 V	0.0277 mA	2.77 mA	2.23 V	active
2.5 V	0.0383 mA	3.83 mA	1.17 V	active
3.0 V	0.0489 mA	4.80 mA	0.20 V	saturated
3.5 V	0.0596 mA	4.80 mA	0.20 V	saturated
4.0 V	0.0702 mA	4.80 mA	0.20 V	saturated
4.5 V	0.0809 mA	4.80 mA	0.20 V	saturated
5.0 V	0.0915 mA	4.80 mA	0.20 V	saturated

Read as logic, the same stage is a NOT gate.

Each input level, the voltage it is, the voltage out, and the level that comes out
in	volts in	volts out	out
0	0.0 V	5.00 V	1
1	5.0 V	0.20 V	0

the input, as a logic level: 0
the output, as a logic level: 1
two stages in series: 0.0 V in, 5.00 V between them, 0.20 V out

yes, the output is the opposite of the input

One stage turns 0.0000 mA into 0.00 mA and turns the signal upside down doing it. The gain is why one output can drive several inputs; the inversion is the gate relay contacts could not make.

What this page checked when it loaded.

Each claim, whether it held, and the values behind it
claim	held	measured
a low input gives a high output	yes	0.0 V in, 5.00 V out
and a high input gives a low output	yes	5.0 V in, 0.20 V out
collector current is a hundred times base current	yes	0.0170 mA in, 1.70 mA out
the output never rises when the input does, across the sweep	yes	51 points from 0 to 5 V
an inverter driving an inverter is a buffer	yes	5 V in, 5.00 V out

What is real here, and what is not

This is a silicon textbook stage, and 1947 was neither

The device Bardeen and Brattain built in December 1947 was a point-contact transistor: two gold contacts pressed on a germanium slab, with a current gain under two and a habit of dying. Everything on this page — 0.7 volts across the base junction, 0.2 volts when saturated, a gain of a hundred — describes the silicon junction transistor that came later and is what a person means today. The date is right for the invention and wrong for the numbers, and it seemed better to say so than to publish a page nobody could compare with a datasheet.

The model is three straight lines, not semiconductor physics

Cut off below 0.7 volts, collector current equal to beta times base current above it, and a ceiling where the collector resistor cannot pass any more. That is enough to show the two facts this page is about and it is not enough for anything else: it has no temperature, no early effect, no leakage, and the transition between regions is a corner rather than a curve. A real device rounds every corner here.

Volts, milliamps and kilohms, so the arithmetic has no hidden thousand

Ohm's law works with no conversion factor in those units, which is deliberate: a stray factor of a thousand is exactly the kind of error that hides on a page like this and still looks plausible. Every number shown is computed from the three constants and the two resistor values, and the base current is never rounded before the collector current is derived from it.

Nothing here is measured from a real transistor

No device was put on a bench for this. The numbers come out of the model, and the model comes out of a textbook. Where the page says the output never rises when the input does, that is a property of the model checked across the sweep when the page loads, not a measurement.

Sources

Logical Art builds interactive explanations of difficult machines.

The machines:

1935-1991 RangekeeperThe answer is on both sides of the equation, so it had to be a machine.
1964-1990s AbendThe loop you did not write, and the day it cost to get it wrong.
1975 Front PanelA working 1975 Altair 8800 you program by flipping switches.
1980s 300 BaudA real dial-up modem and the BBS on the other end of it.
1994 WinsockThe two seconds after the modem stopped screeching, before anything worked.
2020s LLM LabTrain a tiny language model and watch every number as it learns.

Smaller pieces, on the studio itself: Morse, Baudot, Punch Cards, Interlace, Bitwise, Stored Program, Trigger, Bit, Nibble, Byte, Transistor, Hamming, Huffman, Core Memory, The Interrupt, Score Reel, CRC, Page Fault, ASCII, CPU Cache, Race, Fifteen Planes, B-tree, Backoff, Checkout, Diffie-Hellman, Fringes, LZ77, Zero Address, Signature, Z-Characters, Merkle Proof, Branch Prediction, Interleave, Eight Sprites, DNS, Floating Point, Slow Start, UTF-8, Masking, PageRank, Merge Collapse, Count Distinct.

A Logical Art machine by RoamingPigs, Inc. Built by Cisco Caceres in Las Vegas, NV.