Guides
Experimental features
Everything in this repo falls into one of two groups:
- Core is acquisition, contact checks, sample-rate measurement, recording, playback and BIDS export. It follows established practice, or states plainly where it doesn't.
- Experimental is analysis we find useful that is not yet an established method for this kind of hardware. It may change or disappear, and it has not been validated against a reference amplifier.
Experimental features carry a beta — experimental badge in the workbench, and every response from an experimental bridge route includes "experimental": true and a note. Their server code lives in apps/server/experimental.py. Nothing in core depends on them.
Using them
They are off unless the optional companion analysis tools are installed. Point the bridge at them:
export OLIMEX_TOOLS=/path/to/tools
python3 apps/web/serve.pyWithout them, those routes answer 503 not installed and the workbench hides what depends on them.
What counts as experimental
| Feature | Why it is experimental |
|---|---|
| Lock-in extraction, integration check, noise budget, detection curves | Novel use on a 10-bit, uncalibrated single channel |
| Auditory response atlas | Predictions from the literature, not measurements on this rig |
Rig noise floors in nV (experimental.MEASURED_RIG) | Depend on the uncalibrated µV-per-count scale; spread unexplained |
| Alpha-SNR part of the signal-plausibility check (EEG only) | A heuristic, not an impedance measurement |
Client-side bands() and iaf() (eeg-client.js) | Deprecated; the server computes these |
If you use any of these in a write-up, say that it is experimental and cite the version you ran (see versioning.md).
Rendered from docs/experimental.md.