RP1 compact conversion traces.

Question and rationale

The single-command investigation recovered a readiness change and RESULT 861 after two untriggered zero readings. However, SD persistence separated command completion from the readiness check by 115.799 milliseconds and from the result read by 347.423 milliseconds. This experiment removes persistence from those local intervals and tests recurrence alongside neighboring untriggered controls.

The investigated capability is direct SEIGRASM component operation with observable consequences. It supports the control and measurement work required for physical Seigsit development; the present converter records do not establish six-state transport, physical Seigsit creation or Hyphos bootstrap.

Method

Retain the preceding route, identity, clock and ADC admission checks. Enable internal input 4 and the converter with CS 0x4003, require READY and CS 0x4103, and retain the disabled FIFO, zero divider and disabled interrupt controls.

Alternate thirteen compact traces: control 0, command 1, control 1, through command 6 and control 6. Each performs seven access operations: CS read; RESULT read; guarded command or untriggered CS read; immediate RESULT read; readiness poll; RESULT read after readiness; final CS read. A command requires immediate prior CS 0x4103 and writes 0x4007 once. The poll requires READY set and START_ONCE clear, with a ten-millisecond deadline.

There is zero added wait within a trace. Its seven reports remain in RAM until the trace returns; SD persistence then records the group. Instruction, counter and bus costs remain measured work. The control replaces a guarded command with one read, so their durations are not identical. After each persisted trace, a separate admission verifies CS, FCS, DIV, INTE and INTF. Failure stops further traces and selects staged cleanup.

The complete schedule contains 123 operations and at most 473 journal sectors. It permits six conversion requests and eight ADC writes including enable and restoration. Record six surrounding component controls before and after; restore ADC control first and clock enable second. No continuous conversion, FIFO read, divider change, external-pin action or interrupt enable is scheduled.

Assessment criteria

Assess every command and control, their timing, configuration checks and restoration independently. A valid zero remains an observation. Missing or incomplete reports remain unavailable. Retain equal results: a steady input can produce equal conversions, while an ignored command can also leave an old result. Neither equality nor variation alone establishes freshness.

The fixed alternating order provides local comparisons, without isolating all temporal drift or concurrent activity. Poll records contain first and last values and a read count, not a continuous record of every status. Counter brackets bound software accesses rather than the physical conversion instant.

Preparation and implementation

Candidate 749FB63F7081 appends a bounded compact executor to the existing SEIGRASM RP1 instruction source. It invokes the shared access operation, clears all subreports before execution and stops at the first failed access. The ten-cell group header records status, attempted and declared counts, frequency and start/end counters. Each of the seven ten-cell subreports retains raw values, status, read/write counts and counter brackets. Protocol 4 identifies these grouped reports; the other seven target instruction sources retain their preceding bytes.

Twenty-eight supplied-response execution checks passed, including complete native recording, no SD calls inside a trace, count bounds, failed commands, absent readiness, equal results, ignored commands, restoration and recording loss. Both full installation readbacks matched. The frozen protocol specifies a two-minute powered run. These preparation checks are separate from the returned observations below.

Existing ROM/EEPROM entry, AArch64 execution and counter, SD and PCIe transports, retained bridge setup and existing RP1 firmware remain apparatus dependencies. Target behavior consists of authored SEIGRASM instruction words; PC tooling packages and checks them.

Results

All 473 consecutive records and 123 operations were recovered with accepted completion. Every compact trace retained all seven access reports. Each command was issued once from CS 0x4103; its immediate response was 0x4003, with READY clear. The first poll read returned 0x4103 in all six cases. START_ONCE was already clear in every immediate readback; its asserted state was not sampled.

Four commands changed RESULT; commands 3 and 5 retained the preceding value. All seven controls retained their values, and immediate and post-readiness RESULT reads agreed in all thirteen traces. No missing result was classified as zero.

Every post-trace configuration check passed. The six surrounding controls were unchanged: root link 0x0003e0b0, root command 0x00100002, RP1 identity 0x20001927, clock CTRL 0x10000800, DIV=1 and SEL=1. ADC restoration at operation 111 and clock restoration at operation 117 each wrote zero once and read back zero; later observations and final checks confirmed both restored values.

Timing and comparison

With the reported 54 MHz counter frequency, the immediate RESULT operation began 1.500–1.778 microseconds after the command routine ended. The readiness poll began 4.074–4.444 microseconds afterward, and the final RESULT operation began 7.241–7.611 microseconds afterward.

The previous recording-separated poll began 6,253,157 ticks (115.799 ms) after command completion; these polls began 220–240 ticks afterward. This is a reduction in the observation gap, not a measured improvement in converter speed. The immediate RESULT read still precedes the poll, so this sequence does not locate the first possible READY-high observation.

Whole compact traces spanned 1,856–2,024 ticks (34.370–37.481 microseconds), including report clearing and all seven operations. The coordinator spanned 1,001,334,899 ticks, 18.543239 seconds, from entry to its terminal snapshot. Boot-loader time and final-record persistence are outside that interval. Exact powered duration was not independently measured.

Recording integrity

Two complete read-only 64 MiB captures matched. Independent reconstruction checks the raw sector identity, CRC and complement, senary fields, ordinals, group boundaries, command records and terminal counts. All 32,231 unused journal reservations matched preparation. The partition table, executable and boot configuration were unchanged.

Seven other boot-volume sectors differed and a System Volume Information directory was present; their writer and timing remain unestablished. No acquisition-tool SD write was performed. A terminal record cannot establish its own target readback. Partial group persistence would leave incomplete observations rather than an inferred completed command.

Interpretation and next investigation

The command-associated readiness response recurred in all six commanded traces, with no intervening SD persistence inside each trace. Controls retained READY and RESULT, while four commands changed RESULT. The two equal-result commands also retained the readiness change; this demonstrates why equality cannot be used alone to reject a command-associated response.

This result supports bounded command-controlled acquisition on the tested apparatus. It does not independently identify each analog conversion, establish calibrated temperature or voltage, exclude other component activity or qualify firmware replacement. The six commands refer to six requests of one internal input, not six electrical states. The earlier PMIC exceptions concern a different measurement path.

Next investigation: test bounded autonomous acquisition and explicit stopping, retaining no-command and single-command controls. Preserve acquired values, readiness, queue state, overflow and restoration separately. This would test whether progress continues after one initiating command, rather than merely repeating the present one-command sequence. Its exact control sequence and acceptance conditions require a new frozen protocol; no successor image is prepared in this record.

Evidence and provenance

Preparation identity: 749fb63f70810b10e065a047340f96e4eec8b9fa562e46eb460aa46afef40431. Returned capture SHA-256: 4d3c4fb30984f9208966565cbd2659271c5a912c2914c64d80311078be235404.

The independent return checks reconstruct the compact sectors, preserve equal values, verify timing and restoration, reject an altered command guard and distinguish a missing group from a completed command. The legacy section-3 trial count remains zero; the thirteen compact traces are represented separately.