io.github.dfieser/thermocouple-its90
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@io.github.dfieser/thermocouple-its90Convert 12.209 mV from a 25C reference junction at 0°C ice bath. Convert to type K 'K' type and temperature C. From the type K thermocouple"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
thermocouple-its90
NIST ITS-90 thermocouple conversion for Python: voltage to temperature and temperature to voltage for all eight letter-designated types (B, E, J, K, N, R, S, T), with cold-junction compensation and Seebeck coefficients. The test suite verifies every one of the 12,026 tabulated points of the NIST reference tables on every run.
Pure standard library, no dependencies, fully typed. A browser version of this engine runs as a free online thermocouple calculator, vendored from this repository on every release.
Install
pip install thermocouple-its90Related MCP server: mcp-units
Convert thermocouple millivolts to temperature
from thermocouple_its90 import TypeK
TypeK.emf(300.0) # 12.209 (mV at 300 C, 0 C reference: the table value)
TypeK.temperature(12.209) # 300.0 (C, ice-bath reference)
TypeK.seebeck(300.0) # 0.0416 (mV per C)
TypeK.range # (-270.0, 1372.0)Cold-junction compensation
A thermocouple measures the difference between its junctions, and the reference tables assume the cold junction sits at 0 C. If your meter's terminals are at room temperature, a direct table lookup is wrong:
from thermocouple_its90 import TypeK
# Meter reads 4.096 mV, terminals at 25 C:
TypeK.temperature(4.096) # 100.0 C <- naive lookup, wrong
TypeK.temperature(4.096, reference=25.0) # 124.3 C <- the actual answerAll eight types work the same way:
from thermocouple_its90 import get, letters
letters() # ['B', 'E', 'J', 'K', 'N', 'R', 'S', 'T']
get("s").emf(1400.0) # 14.373 mV
get("t").temperature(-4.648) # -149.98 C (cryogenic ranges included)Accuracy, verified rather than asserted
The forward reference functions reproduce all 12,026 one-degree points of the NIST reference tables within their 0.001 mV print rounding. The full comparison runs in CI on every push (
tests/test_full_tables.py).Inversion seeds with the published ITS-90 inverse polynomials and refines by Newton iteration on the exact forward function. Round trips agree to microdegrees instead of the inverse polynomials' 0.02 to 0.06 C error bands.
Types K, E and T invert over their full cryogenic ranges (to -270 C), below where the published inverse polynomials stop.
Type B is handled with its physics: its EMF is non-monotonic near room temperature, so inversion below 0.291 mV (about 250 C) is refused with an explanation instead of returning one of two possible answers.
MCP server for AI agents
Language models mis-remember thermocouple polynomials. The package ships a Model Context Protocol server so agents call the verified implementation instead:
pip install "thermocouple-its90[mcp]"
claude mcp add thermocouple -- thermocouple-its90-mcpTools: thermocouple_to_temperature, thermocouple_to_emf,
thermocouple_types. All three are annotated read-only and idempotent with
no outside effects, so a host can tell a user what a call will do before it
runs.
The extra needs Python 3.10 or newer and works with either line of the MCP SDK, 1.7 and later or 2.x. CI builds the server against both on every push.
Data provenance and licensing
Coefficients were machine-parsed, never hand-typed, from the NIST ITS-90
Thermocouple Database (SRD 60, https://its90.nist.gov/). The database reproduces
NIST Monograph 175 (Burns, Scroger, Strouse, Croarkin, and Guthrie, 1993),
a United States government publication not subject to copyright. The
canonical dataset lives in data/its90.json; _data.py is generated from
it and CI fails if they drift. Code is MIT licensed.
Related
Live calculator, this engine in the browser, with worked examples and FAQ
More verified engineering calculators by the same author, including reference-electrode, alloy-composition, diffusion and XRD tools
lcf-strain-life, the same library-plus-MCP pattern for low-cycle fatigue analysis
Citation
If this library is useful in published work, please cite it via the
concept DOI https://doi.org/10.5281/zenodo.22036393, which always resolves to
the latest release (see CITATION.cff). Please also cite the underlying reference: Burns, G. W., Scroger, M. G., Strouse, G. F.,
Croarkin, M. C., & Guthrie, W. F. (1993). Temperature-electromotive force
reference functions and tables for the letter-designated thermocouple types
based on the ITS-90 (NIST Monograph 175). NIST.
This server cannot be deployed
Maintenance
Related MCP Connectors
MCP server for aerospace calculations: orbital mechanics, ephemeris, DSN operations, ...
QuLab MCP remote server (Streamable HTTP) for computational science and lab tools.
Engineering calculation MCP server for oil and gas engineering applications.
MCP server with quote and live cryptocurrency price tools, local and cloud-deployed transports.
Related MCP Servers
- AlicenseAqualityFmaintenanceA universal MCP server that exposes all UTCP-registered tools to MCP clients while providing a web interface for tool management.712 npm203MIT
- AlicenseAqualityDmaintenanceMCP server for converting cooking measurements (volume, weight, temperature) between common units like ml, cup, g, oz, and Celsius/Fahrenheit.31MIT
- AlicenseBqualityBmaintenanceAn MCP server for controlling lab instruments (oscilloscopes, signal generators, etc.) via standardized interfaces like USBTMC, RS-232, and LAN.100MIT
- AlicenseAqualityCmaintenanceThis MCP server enables AI clients to calculate the boiling temperature of water at a given absolute pressure using the IAPWS-IF97 standard, separating natural-language reasoning from authoritative scientific computation.1MIT