True Random Number Generation from a tiny, low cost chip (about $1)
Meets FIPS-140-1/2    

Made for fast, effective Ciphers and Encryption using an off-processor solution for rapid embedded designs.
A true random 32bit value available by serial communications: High speed SPI or fixed 9600 baud UART connection
THE R101 IC CREATES TRUE RANDOM 32BIT NUMBERS (TRNG)

Utilizing a proprietary use of multiple noise sources
such as Johnson, shot, drift, and jitter, results in true 
physical randomness without an algorithm that can be hacked.

A unique method of noise dispersion is employed on each sample
approaching a Bernoulli distribution of a -2 excess kurtosis.

AEC-R101 IS AVAILABLE IN 3 PACKAGE STYLES;

AEC-R101-01 = DIP 8 PACKAGE
AEC-R101-02 = SOT23-6 PACKAGE
AEC-R101-03 = DFN 8 (2X3mm) PACKAGE

SPECIFICATIONS;

TEMPERATURE; STORAGE -65C to +150C, OPERATING -40C to +125C
SUPPLY VOLTAGE MAX = 6.5VDC, MIN=-0.3VDC
SUPPLY CURRENT = TYPICAL 2.2mA @ 3.0V, 2.4mA @ 5.0V
PIN CURRENT = 20mA MAX IN or OUT
INPUTS INTERNAL PULLED-HIGH WITH 100uA TYPICAL

Transmit of 5000 samples of 32bit values with carriage return
in Hexadecimal (tab delimited) 5Ksamples.txt
Link to +23C versus -35C Hot and Cold graph.
.
Operation:
THE STATE OF "MODE" PIN DEFINES WHETHER
DATA IS AUTOMATICALLY CLOCKED OUT BY UART 
WHEN THE "ENABLE" PIN IS CYCLED FROM LOW TO HIGH
 OR 
CLOCKED OUT MANUALLY BY TOGGLING THE "CLOCKIN" PIN

--IF MODE IS HIGH--
ENABLE PIN, WHEN LOW, LOADS A NEW 32BIT RANDOM VALUE,
DATA OUT SENDS A 32BIT VALUE WHEN ENABLE IS HIGH,
STANDARD BAUD RATE IS 9600, NO PARITY.
DATA IS TERMINATED BY (13) CARRIAGE RETURN CHARACTER.

--IF MODE IS LOW-- 
IT LOADS A NEW 32BIT RANDOM VALUE;
CLOCKIN LOW LOADS AN INDEXED BIT OF THE 32BIT VALUE,
CLOCKIN HIGH SETS BIT OF 32BIT VALUE INTO "DATA OUT" 
AND READIES THE NEXT BIT TO BE CLOCKED OUT.
( THIS IS SIMILAR TO SPI MODE-0 )
WHEN 32 BITS ARE COMPLETE AND MODE REMAINS LOW, 
A NEW 32BIT RANDOM VALUE IS LOADED AND THE PROCESS REPEATS.
THE OPERATION IS ABORTED ANYTIME "MODE" REVERTS TO HIGH.

IF "HOLD" IS KEPT LOW, THE RANDOM NUMBER IS NOT UPDATED 
AND THE DATA REPEATS THE PREVIOUS RANDOM 32BIT VALUE.
---

John von Neumann cautioned about the misinterpretation of a Psuedo.Random.Number.Generator as a truly random generator,
"Anyone who considers arithmetical methods of producing random digits is, of course, in a state of sin."

Simplified to 16 bits for demonstration, 
~450 single fixed noise samples(left) is cycled
through 16 phases of "static" factors of dispersion  
(below). Just 2 of 19 noise injections used.
These false snapshots use a fixed noise source.
Actual operation applies multiple noise sources and
a new random exponent of dispersion upon each sample.
The dispersion factors are continually randomized
and not fixed for any group of samples as depicted.
This assures that the Skew cannot be predicted.

Depicted are the least randomizations applied.
Actual 32bit values are not shown as they were
deemed graphically too difficult to view.