q-extraction

Automated Q-factor extraction from resonator frequency sweeps (FEM/EM simulation output: HFSS, COMSOL, ADS, ...), with three interchangeable front ends:

Path	File	Use when
OriginLab, embedded Python	origin/extract_q_origin.py	You work in Origin 2021+ (no extra packages needed)
Standalone Python	standalone/extract_q.py	You have CSV exports; adds a Butterworth-Van Dyke fit cross-check
Pure LabTalk	labtalk/extract_q_diag.ogs	Origin older than 2021, or any Origin without working Python

Method

Half-power (-3 dB) bandwidth with linear interpolation between samples:

Q = f0 / (f2 - f1)

where the crossing level depends on what the trace physically is:

• motional conductance G = Re(Y): level = peak / 2 (G is already a power-like Lorentzian)
• admittance magnitude |Y|: level = peak / sqrt(2)

Using the wrong level biases Q by ~1.55x, so the two Python front ends auto-detect it per trace: a column of complex literals with a trailing i (e.g. 4.656e-7+0.0141i) is real Y data, so conductance = Re(Y) is used directly — unambiguous, since that's exactly what conductance mode already needs. A column of plain real numbers can't self-identify as G vs |Y|, so it falls back to MODE = 'conductance' | 'magnitude' (top of the Origin script) / --conductance/--magnitude (standalone script, mutually exclusive; also force the mode for complex traces if given — e.g. --magnitude on a complex column takes abs(Y) instead of Re(Y)). The pure LabTalk front end doesn't implement this detection (parsing a+bi reliably isn't worth the risk in that interpreter) — it always uses the plain 3dB level from its MODE setting.

Extracted parameters

Parameter	3dB method	BVD fit (standalone only)
f0 (resonance frequency)	peak sample + parabolic refinement	fit — use this one to track softening/hardening
peak value	raw maximum	—
BW, Q	interpolated -3 dB crossings	fit
baseline	median of trace (off-resonance floor)	fit offset
asymmetry	(f2-f0)/(f0-f1), 1.00 = symmetric	phi (conductance) / C0 (magnitude)
Rm, Lm, Cm	Rm = 1/(peak-baseline), conductance only	fit

Butterworth-Van Dyke fit (standalone script)

The standalone script also fits a BVD model near resonance: motional branch Ym = (1/Rm) / (1 + 2jQ(f-f0)/f0), plus a Fano skew angle phi (conductance mode) or a complex feedthrough ~ j*2*pi*f0*C0 (magnitude mode). On a symmetric peak it agrees with the 3dB method; on an asymmetric (Fano) peak the 3dB Q is biased (the asym column flags this — 1.00 means symmetric) and the BVD values are the ones to trust. For softening/hardening assessment across drive levels, track the BVD f0: it stays exact even when feedthrough shifts the raw peak position by several kHz.

Data layout

Column A = frequency; every other numeric column = one trace. Column X/Y designations in Origin are ignored (multi-trace ASCII imports often mislabel data columns as X2, X3, ...). Columns whose values fall inside the frequency range are skipped as duplicate axes. Trace names are taken from Origin's Comments row (falls back to Long Name).

Sweep parameter columns (Origin script)

If a trace's Comments/Long Name is a name=value list — e.g. R_l=0.1 Ohm, motional conductance — origin/extract_q_origin.py splits it and adds one QResults column per distinct parameter found (R_l in this example). Sweeping more than one parameter (e.g. internal_ring=2.5 um, external_ring=4 um, ...) adds one column each, so a multi-parameter sweep unstacks into separate, sortable/filterable columns instead of one combined text label. Tokens without an = (like the trailing motional conductance) are ignored. f0/peak/BW columns get their units from the source worksheet's Units row (falls back to (x units)/(y units) if that row is empty).

Usage

Origin 2021 and newer (embedded Python)

Origin 2021 introduced the embedded Python environment with the originpro module — that is all origin/extract_q_origin.py needs (deliberately no numpy/scipy, so a stock install works):

1. Activate the data workbook (click it so it is the active window).
2. Connectivity -> Open Untitled.py, paste origin/extract_q_origin.py.
3. Set MODE = 'conductance' | 'magnitude' at the top, press Run (F5).
4. Results print to the message log and land in a new book named <source book> - QResults, with units on f0/BW/peak/baseline copied from the source columns.

Tested on Origin 2025b; anything 2021+ should behave the same.

Pasting into Open Untitled.py every time is fine for occasional use but gets old fast if you run this often. Two less copy-paste-y ways to launch the same script, both documented by OriginLab:

• One-click button: bind a button (worksheet button, or Format -> Toolbars custom toolbar button) to a LabTalk command that runs the saved file directly (the trailing 2 tells run.python() its argument is a file path, not inline code — see Run(object)):

  run.python("C:\path\to\origin\extract_q_origin.py", 2);
  

  No paste step; you edit the .py file on disk and the button always runs the current version.

• Origin App (origin/app/QExtraction/): a real Apps Gallery icon, see below.

Building the Origin App

origin/app/QExtraction/ in this repo has the two files Origin's App packaging system needs — launch.ogs (entry point) and AppIcon.png (32x32 icon) — following the Create and Update Apps convention (Python itself isn't one of the officially listed App content types — X-Function/.ogs/.c/.h/DLL are — but .ogs calling run.python() is a documented, working way to run a Python script from one). To build and install it:

1. Copy origin/extract_q_origin.py into your Origin User Files Folder (%Y in LabTalk — find it via Tools -> User Files Folder). launch.ogs runs %Yextract_q_origin.py, so re-copy it there whenever the script changes; Package Manager bundles the launcher and icon, not an arbitrary file-copy-on-install step, so the .py can't ride inside the .opx itself.
2. In Origin, open Code Builder, find origin/app/QExtraction/ (or copy it into your Apps workspace folder), right-click it, and choose Generate... to open Package Manager. Point it at launch.ogs as the entry script and AppIcon.png as the icon, set a name/version, and generate the .opx.
3. Double-click the resulting .opx to install; the icon appears in the Apps Gallery. With a data workbook active, click it to run — same output as the manual flow (<source book> - QResults book).

This last step (running Package Manager and testing the installed App) has to happen inside Origin itself — I can't drive that GUI or verify it end-to-end from here, so treat step 2/3 as the part to sanity-check first if the icon doesn't behave as expected.

Origin older than 2021 (no originpro)

Pre-2021 Origin has no embedded Python (Connectivity menu and originpro do not exist; 2017-2020 only offer the external PyOrigin bridge, which needs its own Python install). Two options, in order of preference:

1. Export + standalone script (full feature set incl. BVD fit): File -> Export -> ASCII, comma or tab separated, then run standalone/extract_q.py on the exported file (see below).

2. Pure LabTalk (3dB method only, results in the Script Window): open the Script Window (Window -> Script Window) and run

   run.section(<path>\labtalk\extract_q_diag.ogs, main);
   

   with the data worksheet active. If main fails silently (old LabTalk interpreters reject different constructs), run the staged sections t1..t5 one by one — the first section that prints its header but not its == tN OK == line names the offending construct. Note: the LabTalk path uses the peak/sqrt(2) (magnitude) level and X/Y column designations; check both match your data.

Standalone script

python standalone/extract_q.py data.csv --funit GHz --conductance

Tests (synthetic Lorentzian with known Q; both modes must recover it):

python tests/test_synthetic.py

Accuracy notes

• Make sure the sweep has at least ~10 points inside the 3 dB bandwidth.
• A visibly asymmetric (Fano-like) peak from capacitive feedthrough biases the 3 dB method; watch the asym column and trust the BVD-fit values (standalone script) when it deviates from 1.00.
• Q is unitless: frequency units (GHz vs Hz) do not affect it.