FidStorageBase

FidStorageBase extends DataStorageBase to manage free-induction decay (FID) waveforms for CP-FTMW acquisitions. It holds a fixed number of FID records per segment (d_numRecords) and implements the DataStorageBase lifecycle methods. The constructor reads any existing fid/fidparams.csv into an internal template list so that on-disk data can be accessed without re-reading the full experiment.

Three concrete subclasses cover the standard acquisition modes:

FidSingleStorage — a single persistent FID segment with optional backup snapshots (single-segment and target-shots modes).
FidMultiStorage — multiple FID segments for LO-scan and DR-scan acquisitions.
FidPeakUpStorage — a transient rolling-average store for peak-up mode (d_number is -1; no disk I/O).

These subclasses live in their own headers and are not documented on this page; they each implement the pure-virtual loadDifferentialFidList and getCurrentIndex methods.

FID I/O surface

addFids() co-averages an incoming Fid list into the current segment, applying an optional time-domain sample shift. setFidsData() replaces the current list unconditionally (used for background-subtraction updates where the new list is not simply additive). getCurrentFidList() returns a mutex-protected copy of the in-progress accumulation. loadFidList() reads segment i from the in-memory cache if available, or from fid/<i>.csv on disk; it is safe to call from any thread. The pure-virtual loadDifferentialFidList() returns the background-subtracted list and is implemented by each concrete subclass.

Cache

Loaded FidList objects are stored in d_cache, keyed by segment index, up to d_maxCacheSize bytes (approximately 200 MB by default). The insertion order of segment indices is tracked in d_cacheKeys; when the cache is full, the oldest entry is evicted before inserting a new one. Cache updates are serialized by an internal pu_baseMutex (separate from pu_mutex in the base class, which guards the current FID list).

Backup interface

The virtual backup() method saves a point-in-time snapshot of the current FID list. The default implementation is a no-op; FidSingleStorage overrides it. numBackups() returns the number of snapshots available.

Processing settings

writeProcessingSettings() serializes a FtWorker::FidProcessingSettings struct to fid/processing.csv using the keys in BC::Key::FidStorage:

Key	Meaning
`fidStart`	Processing window start (μs)
`fidEnd`	Processing window end (μs)
`fidExp`	Exponential apodization time constant (μs)
`zpf`	Zero-padding multiplier
`rdc`	DC removal flag
`units`	Output magnitude units
`autoscaleIgnore`	LO exclusion half-width (MHz)
`winf`	Apodization window function

readProcessingSettings() deserializes the same file back into a FidProcessingSettings struct. getLORange() returns the [min, max] probe-frequency range across all stored FID segments, which is used by LO-scan deconvolution in FtWorker.

The FID data format and the broader FTMW acquisition workflow are described in Data Storage.

API Reference

class FidStorageBase : public DataStorageBase 

Abstract base class for FID data storage in FTMW acquisitions.

Extends DataStorageBase with the storage model for free-induction decay (FID) waveforms. Holds a fixed number of FID records per segment (d_numRecords) and implements the lifecycle methods: start() and finish() toggle an internal acquiring flag, save() writes the current FID list to disk (skipped for peak-up instances where d_number < 1), and advance() calls save() then delegates to the virtual _advance() hook for subclass-specific segment transitions.

Three concrete subclasses cover the standard acquisition modes: FidSingleStorage (single-segment), FidMultiStorage (multi-segment / LO-scan), and FidPeakUpStorage (peak-up / rolling-average mode).