MeerK40t mock-controller MOPA spike

Date: 2026-10-18
Upstream: meerk40t/meerk40t
Commit: d511944e08788e88077727f682957759884ff5ca
Scope: offline command-list encoding only

Result

The pinned BalorMK controller selected its MockConnection and encoded a synthetic square with explicit power, speed, 40 kHz frequency and 200 ns MOPA pulse width. Two consecutive runs produced identical parsed command traces. The harness supplies a synthetic 110 mm field solely for BalorMK's conversion from mm/s to galvo units; it is not evidence that the physical lens is 110 mm.

The trace contains:

listQSwitchPeriod value 0x01f4 (500), matching BalorMK's internal fiber-source conversion 20000 / 40 kHz;
listFiberYLPMPulseWidth value 0x00c8 (200);
jump and mark commands for the synthetic square; and
a one millisecond listLaserOnPoint dwell represented as 100 ten-microsecond controller units; and
an explicit end-of-list command.

The harness replaces PyUSB with a trap before BalorMK is imported. Any call to USB discovery or resource disposal fails the run. The controller is also constructed with force_mock=True. No USB enumeration or hardware access was attempted.

Reproduce

Check out the pinned MeerK40t commit outside this repository, then run:

python scripts/spike_meerk40t_mock_mopa.py C:\path\to\meerk40t

The script refuses to run against any other commit. Its JSON output includes the upstream revision, inputs, selected transport, USB status and parsed trace. The checked result is stored as tests/fixtures/meerk40t-mock-mopa-trace.json for later golden comparisons.

What this proves

BalorMK's mock path can be forced without controller discovery.
Operation settings preserve frequency in kHz and pulse width as the integer value emitted by listFiberYLPMPulseWidth through command-list encoding.
The tested square produces a deterministic mock trace.

What this does not prove

It does not validate the units or safe range expected by the makerspace's specific JPT E2/M7 source.
It does not exercise MeerK40t's graphical UI, hatch planner, correction file, job spooler, framing, cancellation or status behavior.
It does not establish compatibility with the physical controller, mounted lens or correction profile.
It does not authorize USB access, galvo motion or laser output.

Decision impact

This is positive evidence for the BalorMK command encoder and mock transport, but it is not enough to select stock MeerK40t as the workshop application. The next useful extension of Spike A is a full cut-plan/spooler test containing synthetic hatch geometry. Spike B, the external Rayforge no-device driver registration proof, can proceed independently.

2.7 KiB Raw Blame History