Noise generators (noise sources) inside entropy sources for random bit generators · physical and non-physical noise sources, raw data and digitisation, min-entropy per sample, conditioning, NIST SP 800-90B health tests, and entropy source validation certificates
In a random bit generator, the noise generator (NIST calls it the noise source) is the part that contains the non-deterministic process, for example free-running oscillator outputs or CPU timing jitter in the certified examples below. Its output is digitised into raw bits, which an optional conditioning component processes and health tests check. This page is about that component, not about RF or EMC test instruments sold as noise generators. Ask three things first: whether the noise source is physical or non-physical, what min-entropy per sample has been assessed, and whether the entropy source holds a NIST validation certificate.
Noise sources and complete entropy sources (TRNG blocks, hardware RNG chips and IP) used to seed cryptographic random bit generators. Laboratory RF noise generators, EMC impulse noise simulators and acoustic noise generators are outside this page.
01
NIST SP 800-90B models an entropy source as a noise source, an optional conditioning component and a health-testing component. It calls the noise source the root of security for the entropy source and for the whole random bit generator. The noise source contains the non-deterministic process behind the uncertainty in the output; if that process produces something other than binary data, a digitisation step converts the samples to bits, and the digitised output is called raw data. NIST adds that if the noise source fails, no other component can make up for the missing entropy. [1]
The standard divides noise sources into two categories. Physical noise sources use dedicated hardware to generate randomness. Non-physical noise sources use system data, such as API function output, RAM data or system time, or human input such as mouse movements. [1]
Give the supplier the target standard (for example NIST SP 800-90B) and any certificate required, the noise source type, the output rate and sample size, the claimed min-entropy per sample, the conditioning function, the health tests and their false-positive probability, the operating temperature and voltage range, and the interface (on-chip block, IP core or separate device).
02
SP 800-90B requires a sequential dataset of at least 1,000,000 raw samples taken directly from the noise source for validation. If 1,000,000 consecutive samples cannot be generated, smaller sets of at least 1,000 consecutive samples from the same noise source may be concatenated, provided the combined dataset holds at least 1,000,000 samples. The standard defines two continuous health tests. The Repetition Count Test catches a noise source stuck on one value; for example, with a false-positive probability α = 2^-20 and 2.0 bits of entropy per sample, the cutoff is 11 identical samples. The Adaptive Proportion Test catches a large loss of entropy, using a window of 1,024 samples for binary sources and 512 for non-binary sources. NIST recommends a false-positive probability between 2^-20 and 2^-40 for these tests. [1]
Editorial calculation: 2^-20 is about 1 in 1.05 million and 2^-40 is about 1 in 1.1 trillion.
03
NIST's Cryptographic Module Validation Program publishes entropy source validation (ESV) certificates under SP 800-90B. The examples below are certificate entries, not CNFX test results. [2]
The certificate list also includes jitter-based entropy sources from CPU timing. Filter by physical or non-physical classification and by reuse status when you search it. [2]
| Certificate | Noise source as listed | Listed entropy data |
|---|---|---|
| E2 — WISeKey VaultIC408 | Physical; one noise source of five oscillator outputs XORed together. [3] | 0.52427 entropy per 1-bit sample. |
| E5 — Western Digital FE2-PG4 ASIC | Physical ASIC TRNG. [4] | 2.69 entropy per 32-bit sample. |
| E11 — STMicroelectronics STM32U5x TRNG | Physical; vetted AES-CMAC conditioning. [5] | Full entropy; 128-bit output. |
04
This is a functional reference built from the sources cited; it is not one maker's specification.
| Part | Role | Fix in the order |
|---|---|---|
| Noise source | Non-deterministic process (physical or non-physical). [1] | Type; claimed min-entropy. |
| Digitisation | Converts non-binary samples to bits (raw data). [1] | Sample size; rate. |
| Conditioning (optional) | Processes raw data; vetted functions such as AES-CMAC. [1] [5] | Function; output size. |
| Health tests | Repetition Count and Adaptive Proportion tests. [1] | Cutoffs; false-positive probability. |
| Interfaces | GetEntropy, GetNoise and HealthTest (conceptual). [1] | Raw data access for testing. |
05
A quotation for a noise source or entropy source needs values for your design. The table explains each field, gives published values where a reviewed source states them, and says what to ask the supplier for.
| Field | What it means and published values | Ask the supplier for |
|---|---|---|
| Noise source class | Physical or non-physical. [1] | Physical principle. |
| Min-entropy per sample | Examples: 0.52427 per 1-bit sample; 2.69 per 32-bit sample. [3] [4] | Assessment report. |
| Validation data | At least 1,000,000 raw samples. [1] | Raw data access. |
| Health tests | False-positive probability 2^-20 to 2^-40 recommended. [1] | Test cutoffs and startup tests. |
| Certificate | ESV certificate under SP 800-90B. [2] | Certificate number; reuse status. |
06
Paste it into your enquiry below or into an email to a maker, then replace every bracket.
Standard: [NIST SP 800-90B; certificate needed] Noise source: [physical or non-physical; principle] Entropy: [claimed min-entropy per sample; sample size] Conditioning: [function; output size] Health tests: [type; false-positive probability] Integration: [on-chip block; IP core; separate device; interface] Environment: [temperature; voltage]
No manufacturer record has been reviewed for this product yet. Send your specification and CNFX will match it against manufacturer records.
Explore company information and manufacturing capabilities, then discuss the scope of your project with a relevant supplier.
NIST SP 800-90B defines it as the component containing the non-deterministic process that is ultimately responsible for the uncertainty in the entropy source's output. [1]
SP 800-90B requires a sequential dataset of at least 1,000,000 raw samples from the noise source. [1]
Sources checked 2026-10-11. NIST SP 800-90B is the US federal recommendation for entropy sources. The certificate entries come from NIST's entropy source validation list; they record vendor submissions tested by accredited labs.
Ask for use case, specification boundaries, supplier type, and RFQ preparation information for this product.
Compare manufacturer profiles with relevant product and process capability.