Windows · general diffusion workspace
From spectrum to map, with every decision visible.
Prepare signals, encoding and algorithm before calculation; explore the map and its local results without conflating them.
Guide reviewed · 7 November 2026 · DiffAtOnce Prime 0.15.99 · Calculation times depend on data and hardware.
Prepare an identified copy
Open the project or batch and save a new copy. Review sample, preparation, folder, acquisition, pdata processing and title. Keep Bruker originals unedited.
Before continuing: Two pdata processings of one acquisition are processing alternatives, not independent replicates.
Choose the correct input route
For 1r series review detected acquisitions and ramps. For a pseudo-2D matrix open DOSY 2rr · input firewall and select the exact pdata/N/2rr file. This workspace is independent and does not replace your open project.
Before continuing: A 2rr can be an attenuation series, a calculated DOSY map or a spectroscopic 2D dataset; they cannot all be fitted as gradient ramps.
Review signal, standard and conditions
Select peaks in ppm and confirm extraction common to the ramp. Separately review internal standard, temperature, medium and viscosity where applicable. In a batch, check each factor’s source and each destination’s temperature.
Before continuing: A peak matching a catalogue is a proposal, not confirmation of identity or purity.
Choose algorithm and D support
Review the method and its requirements before calculating. Compatible native TRAIn offers automatic or manual D support; apply a justified proposal and read the effective limits. Keep original DRECT/DREAM, B23 and physical-refinement variants separate.
Before continuing: B23 and TabPFN support remain research with documented limitations. Do not select a method solely because it reduces residuals.
Calculate and retain a new review
Confirm applied recipe and calculation scope: ramp or batch. Read all input blockers before starting. Inspect convergence, prediction, residuals, support and axis endpoints when complete.
Before continuing: The 2rr route provides measured D: opening that workspace does not apply TTMS, thermal correction, Rh or molecular mass.
Explore the map and its local results
Compare Original, Model and residual in their corresponding representation. Open a specific peak or bin for its local attenuation and distribution. Under Summary and means select each sample and ramp: its detail does not replace the global average.
Before continuing: An observed maximum, a fitted-only centre and a band cut by the axis must be interpreted differently; do not invent a missing tail.
Attach the visible figure and check the report
Prepare the average, view and visual threshold. Attach that figure to the report and read its sample, ramps and review. Export summary and full report from the same saved result, without recalculation during export. Check tables, numbering, units and reasons for missing values.
Before continuing: The visual threshold changes visibility; it does not change b, inversion weights, factors or saved moments.
Save and reopen before sharing
Retain project or batch, reviews and reports together. Reopen the copy and confirm that every sample and ramp is identified and the selected result is retained. Share only data you are authorised to share.
Before continuing: Beatrix guides review and retains sources and limitations; generated text does not replace checking the data.
BEFORE CALCULATION · DOSY 2rr
Correct reading is not enough to calculate D.
Input review separates binary format, row correspondence and physical encoding. A matrix can be readable yet still not support a reliable diffusion calculation.
procs · proc2s · 2rr
Dimensions, XDIM blocks, numeric type, byte order and NC_proc scale. An axis displayed by procs does not replace a physical b vector.
acqus · acqu2s
Acquisition status, increment count, timings and GPNAM/GPZ. These arrays use zero-based indices. Pending acqu is not a silent substitute for acquisition-status acqus.
pulseprogram · gpnamN
Actual program instructions identify the active gradient; comments do not configure it. GPNAM supplies the declared shape, while gpnamN retains its samples and area factor when archived. There is no need to type an already-declared shape.
Difframp · difflist · gradient_calib
List, amplitude and shape are compared with the applicable calibration. Standard Bruker difflist includes waveform area; it is neither a percent nor to be shape-corrected twice. A custom list requires its documented convention.
- Retain available evidence: program, shapes, lists and audita/auditp. If history links 2rr to another acquisition, resolve provenance before assigning current parameters to it. An optional file’s absence alone does not establish that the experiment is invalid.
- If shape, amplitude or list conflict, read the finding’s code, file and reason. Do not automatically set GPZ to 100%, substitute SINE for SMSQ, or copy another sample’s calibration to remove it.
- A row confirmation does not resolve a physical contradiction. On Windows an explicit b list needs provenance and review for this exact input. The web laboratory does not offer that route here; do not substitute gradient fields for it.
- Changing files, path, processing, list or declaration invalidates the dependent review. Repeat it before calculation or export of a new calculation. Historical results remain historical: opening them does not reauthorise or recalculate them.
If the conflict cannot be resolved from the available data, retain the diagnosis and obtain a coherent export. Do not invent b or interpolate a missing acquisition; padding rows are never fitted.
WHEN COMPLETE
Your result should be reopenable and explainable.
Retain source, project, review, recipe, factors, findings and report. If a criterion fails, retain its reason too. Professional formatting and cross-application agreement do not replace scientific review.
Usage documentation; it neither runs calculations, changes projects nor experimentally validates a method. Historical screenshots in complete help retain their version and provenance.