Preparation time and conversion activity.

Question and rationale

Does a later reading depend on elapsed preparation time, on intervening conversion activity, or on their interaction? The preceding investigation found that reference 2 exceeded reference 1 in 58/64 cases. Small first-return differences depended on the chosen reference. This experiment keeps the source disabled, crosses two elapsed intervals with command or read-only activity, and retains every first and subsequent observation.

The wider objective is sovereignty over all silicon components and their interconnections: discovery and control through SEIGRASM, replacement of inherited firmware dependencies, and Hyphos/SeigrOS with native electrical Seigsit relations across clusters. Here the contribution is an observable distinction between elapsed time and deliberate action in one component. Neither its source-control setting nor its numerical reading is assigned as a Seigsit state.

The earlier per-case 1,000-microsecond wait preceded three references. Here the controlled 100- or 1,000-microsecond interval follows those references. This changes the intervention structure; it is not an unchanged repeat. References provide a local trajectory without an assumption of complete settling.

Four-case comparison

Deadlines are anchored to the end counter of an unchanged source-control write. The subsequent action occurs after the wait returns; the actual write, conditioning action and probe command are bounded by recorded counters. Scheduled offsets are not exact internal electrical event times.

Sixteen blocks use orders A B D C, B C A D, C D B A and D A C B, each four times. Successive four-block cycles traverse the list forwards, backwards, forwards and backwards. Every case occupies each position four times, and every directed predecessor pair within a block occurs four times. Allocation is deterministic.

Before each block, establish the disabled source, wait at least 1,000 microseconds and admit the stopped-ready converter on channel 4, divider zero, FIFO disabled and empty, and interrupts disabled. Common initialization briefly enables the source before the comparison baseline is established; all four comparison cases keep it disabled. The admission wait is not evidence of complete settling.

Each case contains 31 native accesses:

1. Command three consecutive reference observations, recording command, readiness, RESULT and CS for each.
2. Issue the unchanged, guarded disabled-source write. Its report end is the common timing origin.
3. Wait until 50 or 950 microseconds after that origin, as declared for the case.
4. For B and D, command one conditioning conversion; for A and C, read CS without a command. In either case poll readiness, read RESULT and read CS. The latter is explicitly an untriggered observation, not a new conversion.
5. Wait until the 100- or 1,000-microsecond probe deadline, from the same origin. If either anchored wait begins at or after its deadline, record the missed deadline, stop the group and enter guarded restoration.
6. Command three successive probe observations, retaining command, readiness, RESULT and CS for each. There is no added wait between these probes.

All four cases occupy one 124-access compact group, with no SD transaction between their accesses or cases. Recording and admission occur between blocks. A complete schedule contains 64 cases, 416 commanded conversions, 32 untriggered observations, 64 unchanged origin writes and 128 anchored waits. Every case contributes three preceding references and three probes. The final 50-microsecond scheduled gap leaves margin beyond the preceding run's recorded local access durations; the actual gap must still be assessed from this return.

The conditioning intervention includes its command write and guarded readbacks. Its control performs read-only register accesses. A difference therefore associates with that command path; this comparison cannot separate effects of the write or bus activity from the converter's internal response. Untriggered RESULT equality is recorded retention, not independent proof of conversion freshness.

ERR and ERR_STICKY remain observations without retry or replay into writes. No source activation, channel change, divider change or FIFO consumption occurs inside the comparison groups. The disabled-source potential is unspecified; raw counts are not calibrated voltage or temperature.

SEIGRASM execution and dependencies

The SEIGRASM instruction streams are byte-for-byte unchanged from preparation 23D9BA799F01, whose completed return established execution of anchored waits and larger compact groups on this apparatus. Descriptor scheduling and the independent evidence reader implement the new comparison. Existing access kind 7 binds each deadline to the earlier unchanged-write report; the executor's 128-access capacity contains each 124-access block.

The target path is entry instructions, SEIGRASM route and component controls, compact native access execution, senary journal encoding and SD write/readback. PC Python and PowerShell package authored instruction words and descriptors, validate supplied responses, install the image and recover evidence. They do not implement the target experiment in another target language. No foreign target compiler or target OS is used.

Acquisition calls no firmware service. Existing entry firmware, processor execution, SD transport and retained bridge configuration remain apparatus dependencies. Their replacement across the silicon cluster remains part of the objective. This test does not establish full sovereignty by retaining those dependencies.

Assessment and decisions fixed before return

1. Verify identity, completeness, source-disabled observations, command allocation, unchanged-write readbacks, deadline anchors and arithmetic, all error flags and restoration. A response sign, size or exact equality is not an admission requirement.
2. Retain every reference, untriggered reading and probe, including the first case, first-position cases, zero/equal values and error-bearing observations. Report the two reference analyses and all three probe ordinals; do not select a favourable reference or discard an early reading.
3. Record origin-to-conditioning, origin-to-probe and conditioning-action-to-probe intervals for every complete case. In each full block require common overlap of A/B origin-to-probe brackets, separately of C/D origin-to-probe brackets, and of B/D conditioning-command-to-probe brackets. The first two assess matched elapsed time within each conditioning comparison; the third assesses the matched final gap in the conditioned duration comparison. A timing-compatible block must meet all three. Report every intersection and all timing-incompatible blocks. Overlap is necessary here but does not prove identical internal timing.
4. Report all complete contrasts, then the timing-compatible subset separately. Keep errors visible and distinguish error-clear comparisons. Balance does not make the 16 blocks independent boots or eliminate all shared drift and carryover. Inspect block, position, reference trajectory and absolute raw observations alongside local differences.
5. A conditioning contrast recurring at matched elapsed time supports an acquisition-activity association. A duration contrast recurring under both conditioning settings supports elapsed-preparation dependence. An interaction would indicate that the conditioning association depends on preceding duration. Persistence only in the first probe would bound the effect by observation order. These are measured associations under this apparatus, not an identified internal physical mechanism.
6. If the contrasts remain near the recorded variation or depend on timing mismatch or reference choice, report that unresolved result. Do not select longer waits or discard initial observations merely to suppress deviations.

Independent analog freshness, calibrated source potential, six maintained electrical states, native physical Seigsit creation and transfer, replacement firmware and qualification on another cluster are not established by this preparation. The result must inform the wider component-discovery and sovereignty investigation.

Preparation and acquisition

Preparation CD5BF1356C00 freezes 32 source inputs; 71 image-bound supplied-response checks passed. Installation verified 4,096 chunks and two full image readbacks. The prescribed allowance was three minutes. Two read-only captures of the returned SD agree. The coordinator interval is 98.800518 seconds, excluding entry loading and the terminal record's write/readback. This interval does not measure total powered duration.

Results

All 2,528 records, 177 operations and 64 cases completed. The journal contains 416 commanded conversions and 32 untriggered observations. All 64 origin readbacks matched. No current or sticky conversion errors, source-enabled comparison observations, missed deadlines or admission failures were reported. All 16 blocks meet the three timing-overlap criteria; restoration passed.

Every untriggered observation equals reference 3. In C, the first commanded reading after the long interval is 1–2 counts below reference 3 in all 16 cases. Its next reading rises in 15 cases and is equal in one. In D, the conditioning conversion at the 950-microsecond deadline is also 1–2 counts below reference 3 in all 16 cases; the probe approximately 50 microseconds later rises in twelve and is equal in four.

Against reference 3, D−C for the first probe has median +1 count, with ten positive and six equal contrasts. C−A has median −1, with eleven negative and five equal contrasts. These patterns distinguish command activity from a read-only interval at a matched deadline. Their size and consistency depend on the reference: against reference 1, the same contrast medians are 0 and −0.5, and both contain positive and negative values.

Reference trajectory and observation order

Reference 2 exceeds reference 1 in 19/64 cases, equals it in 36 and is lower in nine. Thirteen of the nineteen positive increments occur in the first position of a block. The preceding experiment had a preparation wait before every case; this experiment has that admission wait and SD recording between blocks. The changed position pattern is consistent with a preparation/history relation, but the between-boot comparison does not isolate which changed activity causes it.

The first case has references 748, 746, 745 and probes 744, 746, 745. Its first probe is −4 against reference 1 and −1 against reference 3. Later cases also expose reference sensitivity: case 11 (B) records references 742, 742, 739 and all three probes at 742. The three differences are 0 against reference 1 and +3 against reference 3. Case 20 (B) has references 740, 739, 737 and probes 740, 739, 740. These readings remain included.

Reference-1 block medians fall from 746 to 731; reference-3 medians fall from 745.5 to 731. No temperature or calibrated source-potential measurement accompanies that drift. All four case types have second- and third-probe medians of 0 against reference 3; individual deviations remain.

Time and activity contrasts

For each case, subtract its local reference before comparing A–D inside the same block. All-block, timing-compatible and timing-compatible/error-clear sets contain the same 16 blocks. These are descriptive contrasts within one boot, not independent boot replications. Both reference choices and all three probe ordinals are retained.

The first-probe duration contrast is consistently negative only for the unconditioned comparison against reference 3. The conditioned duration contrast has both signs. Consequently, this result does not support elapsed time as a sufficient explanation independent of the intervening action. The interaction median is +0.5 against either reference, but individual interaction contrasts have both signs. Later observations reduce several pooled contrasts without making every individual contrast zero.

Recorded timing

The counter frequency is 54 MHz throughout. Every anchored wait enters before its deadline and exits at or after it. Across all short cases, the origin-write-to-probe brackets span 101.852–106.148 microseconds; long cases span 1,001.815–1,006.222 microseconds. The conditioning-command-to-probe brackets in B/D span 47.907–52.074 microseconds. Scheduled deadlines and exact internal event times remain distinct.

Overlap limits observed scheduling mismatch; it does not determine internal sampling instants or prove identical physical intervals. Per-case origin-to-conditioning, origin-to-probe and conditioning-action-to-probe brackets appear below. The read-only control also performs register accesses, so it is not an electrically inactive control.

Individual cases and contrasts

Every case is listed in acquisition order. References and probes are commanded. The conditioning column uses U for untriggered and C for commanded. All rows completed with source-disabled CS observations and clear current/sticky error flags. Values are raw counts; case numbers and positions are one-based.

Interpretation and next question

The repeated low first commanded observation after the long interval, followed by an increase with subsequent observations, narrows the question to recent acquisition and control history. The command at 950 microseconds in D produces that lower reading before the main probe; D then differs from C at the matched 1,000-microsecond deadline. This supports an association with the intervening command path. It does not identify the internal cause.

Two confounds remain central. The conditioning command writes CS and performs guarded readbacks, while its control only reads. Also, every interval begins with an unchanged CS write, and reference readings themselves vary. A candidate next comparison would separate a conversion command, an unchanged control write and a read-only action at the same conditioning deadline, while matching the surrounding access pattern and retaining the three references and three probes. At the 1,000-microsecond probe deadline, this would test whether a control rewrite alone reproduces the command-associated shift. A repeated difference unique to the conversion request would narrow the implicated action; a shared difference would implicate the write path or a coupled response. Neither outcome would by itself identify an analog mechanism.

This run characterizes one RP1 component relation through SEIGRASM. Source-disabled potential, independent analog freshness, six maintained electrical states, native physical Seigsit creation and transfer, replacement firmware and sovereignty across all cluster components remain open. The reusable capability is timed intervention with recoverable action history and explicit uncertainty, to support component discovery across future clusters.

Integrity, restoration and provenance

Independent reconstruction verifies all 2,528 senary records, their identities and integrity fields, all grouped command allocations, unchanged writes, deadline anchors and arithmetic, observations, interval bounds and derived contrasts. Unused journal reservations match preparation. ADC control and divider restore to zero, interrupts are disabled and FIFO empty; clock control restores to zero with divider/select retained at 1.

The partition table, geometry and boot files match preparation; FAT copies agree. Seven sectors outside the journal differ, including System Volume Information. Their writer and timing remain unestablished.

Capture SHA-256: aebb0ef3d3a0cfbd823797ab3130e1499e086371af9d6767a54cb8dba46ee335. Image SHA-256: 878cf8d58fbf51de93796294b98be853d78a8c8ca542a53e63b0a0c1f39f1707.