Skip to content

Analyze Workflow

Start with Quickstart for a short walkthrough. This guide explains the controls that determine which data you are looking at.

Choose the lot and snapshot in a New analysis. A saved analysis pins its context; use Rebase latest deliberately or start new work to change lots. See Upload And Snapshots.

Analyze loads several datasets independently: test summaries, wafer summaries, exact die identity, bin totals, lot metadata, bin labels, and spec overrides. The small load indicators represent datasets, not individual wafers. During die preparation, the status text can also show decode/index work progress.

The counters below are examples; the numbers change as datasets load or fail.

Load status What you can infer
Preparing exact selection · 3/7 Core datasets are still loading. Available test summaries can be read, but exact brushing is not ready yet.
Die index ready · context loading Core data and die identity are ready; other context is still loading.
Lot data ready All lot-load phases completed. Opening a different test or expanded view can still require a chart-data request.
2 datasets failed At least one phase failed. Use Retry failed; a partially visible page is not evidence that every dataset succeeded.

The lot’s complete badge describes ingest, not this browser’s readiness.

The views share lot and snapshot context and a linked die selection. Test and wafer dropdowns can be Synced or Independent. Independent dropdowns let you inspect different views without moving every dropdown; they do not create separate die selections.

A wafer filter, a brush selection, and the Cpk table’s population mode are different controls. Do not assume that changing one silently narrows every calculation on the page.

Cards provide compact inspection. Expand scatter, histogram, or a wafer map for more space and its detailed interactions. Check the chart’s population notice: an expanded chart may still use sampled measurements. See Scatter And Histogram for chart-specific behavior.

In an editable, brushable view with die identity ready:

Action Result
Shift + drag Add the brushed dies to the existing selection.
Alt + Shift + drag Replace the selection with the brushed dies.
Selected-count button in the selection toolbar Clear the shared selection.
Inv or I Invert the selection against the lot’s die population.
Undo / redo arrow buttons Step through selection changes. Keyboard shortcuts are Ctrl/Cmd + Z and Ctrl/Cmd + Shift + Z.

Read-only saved work and share links do not permit selection edits.

Brushing highlights linked dies, but the Cpk table stays on Full lot until you choose its Selected count button. That mode requests selected-population statistics from the server and compares them with the full-lot baseline. Clearing the selection returns the table to Full lot.

The selected count measures dies. A test’s N measures its contributing measurements, so it need not equal the selected count. Capability is suppressed for selected tests with fewer than 30 measurements; a missing Cp/Cpk is not zero or a passing result.

Return to Full lot to edit what-if limits or use Review impact; both are disabled in Selected mode. Limit review evaluates the lot, not just the current brush; see What-If Limits.

Spatial views use observed die coordinates when available. If a lot has no observed X/Y coordinates, sequential review preserves die identity and ordering; its gallery positions are not physical wafer locations. Do not interpret that layout as a defect pattern. See Wafer And Spatial.

Use SQL Scope for a custom query using Analyze’s lot, selection, and scenario context. This flow is one-way: querying in SQL does not brush new dies in Analyze.

Give the investigation a title and use Create Analysis, then Save for later versions. Save at meaningful review checkpoints. Active what-if edits and the selected die population are part of that saved state, not changes to the ingested data. Saved Work And Sharing explains snapshot pinning, local recovery, conflicts, and sharing a version.