# Engraving quality and mechanics


**Draft, not yet verified on hardware.**

Get the deepest engraving the machine allows on stock firmware v1.4.3 and fix text that fades or wobbles. Wear and spare parts: [Hardware and spare parts](/doc/manual/hardware-and-spares#wear). Heat: [Heat and noise](/doc/manual/hardware-and-spares#heat-noise).

<p class="alert alert-warning">Use a public test wallet for test runs. Seed words are typed on the machine only and never pass through a phone, a computer or a chat assistant.</p>

## What you need {#requirements}

- The machine on firmware v1.4.3 ("Firmware: v1.4.3" at the bottom right of the start screen).
- A USB-C PD port with 20 V or 28 V at 5 A, preferably 28 V ([power](/doc/manual/power-and-first-start#power)).
- The hex keys from the box, a drop of light oil, and low-strength threadlocker if you have it.
- For test runs: a spare plate and a public test descriptor sent with [NFC Tools](/doc/manual/nfc-tools-transfer).

## How deep it engraves {#depth}

The needle engraves about 0.3 to 0.4 mm deep into the 3 mm steel plate; 1.5 mm is out of reach. No setting changes the depth, and it does not drift with use. 2 mm plates engrave but warp ([Plates](/doc/manual/plates)). Strike strength follows the supply voltage, the highest the charger offers between 20 V and 28 V. At 20 V each strike gets a slightly longer pulse. 28 V engraves deeper and crisper (unverified). Full procedure: [Load a plate and engrave](/doc/manual/power-and-first-start#plate).

- Plug the cable from the box into the charger port that offers 28 V.
- Put a drop of light oil on the plate.
- If steel has built up on the needle tip, scrape it off with a fingernail.
- Close the lock firmly.

![The same job engraved at 20 V (left) and 28 V (right)](/static/img/engraving-quality-01.webp)
<!-- capture: photo, macro at a low angle, the demo descriptor lab/fixtures/2of3-descriptor.txt engraved "TEXT ONLY" on two 3 mm plates, left at 20 V, right at 28 V, steel ruler in frame -->

## What the machine shows {#machine}

- Homing: at power-up, before every job and after a job that ends normally, the head drives into its corner until both axes reach their stops, then to the plate's top-left corner.
- "Engraving failed.": the head stays where it stopped. Next: [Retry and restart](/doc/manual/troubleshooting#retry).
- Any other error: [Engraving errors](/doc/manual/troubleshooting#engraving).

![Engrave Plate screen after a homing failure](/static/img/engraving-quality-02.webp)
<!-- capture: II screen, flow "Engrave Plate", state failed, "Error: stepper: homing timed out", firmware v1.4.3, research/device-screen-map.md 4.2 row 22 with the test engraver returning that error -->

## Test a job without engraving a plate {#dry-run}

v1.4.3 has no preview or dry-run mode (removed after v1.3.8). Test on the back of the test plate from the box, if yours came with one, or on a spare plate. Or unplug the solenoid.

### Unplug the engraving solenoid {#unplug}

- Unplug the USB-C power.
- Unplug the 2-pin engraving solenoid plug in the middle of the controller board.
- Plug the power back in.
- Start the job. The countdown shows how long the real job takes.
- Watch where the head travels. It should run the whole job without striking (unverified on v1.4.3; it may stop with an error).
- Unplug the power.
- Plug the solenoid back in.

![The 2-pin engraving solenoid plug on the controller board](/static/img/engraving-quality-03.webp)
<!-- capture: photo, controller board in the machine, 2-pin engraving solenoid plug in the middle of the board, pulled out, plug and socket both visible, arrow overlay -->

## Take the play out of the axes {#mechanics}

Each axis runs on a threaded shaft with a brass nut, driven by a stepper motor through a belt. Unplug the power first.

### Brass nuts: words fade toward the end {#brass-nut}

- Loosen the set screw in the brass nut with the supplied hex key.
- Turn the nut until the slack is gone.
- Tighten the set screw again. Low-strength threadlocker on it is optional.
- Do the same on the other axis.

![Brass nut on the threaded shaft with the hex key in its set screw](/static/img/engraving-quality-04.webp)
<!-- capture: photo, brass nut on the X-axis threaded shaft, set screw visible, supplied hex key inserted -->

### Pulleys: homing error, or an axis stands still while its motor turns {#pulleys}

- Move both axes by hand. They should turn freely.
- Remove the left motor cover.
- Plug in the power.
- Watch the homing. A motor that turns while its axis stays still has a slipping pulley.
- Unplug the power.
- Tighten both set screws on each pulley with the smallest hex key, pressing the pulley against the frame.
- If play remains, loosen the locking nut on the other side.
- Tighten the pulley.
- Tighten the locking nut.

![Pulleys behind the left motor cover with their set screws](/static/img/engraving-quality-05.webp)
<!-- capture: photo, left motor cover removed, motor pulley and shaft pulley, both set screws on each marked, smallest hex key beside them -->

### Belts and Y-axis holder: wavy or wobbly text {#belts}

- Tighten the set screw on the Y-axis holder.
- Secure that set screw with threadlocker.
- Loosen the four M3 screws that hold the belt's stepper motor.
- Slide the motor along its slot to tension the belt.
- Tighten the four screws.
- Repeat on the other axis.

![Stepper motor with its four M3 screws and slot](/static/img/engraving-quality-06.webp)
<!-- capture: photo, one stepper motor, four M3 screws marked, slot and belt visible; inset: Y-axis holder set screw with hex key -->

### Needle and cables: the head moves but stops striking {#needle}

- Press the engraving-head cable and the 2-wire cable fully into their sockets.
- Move the head to the right side by hand. Neither cable should pull.
- Unscrew the red head housing. The needle should move freely; spring resistance is normal.
- Oil the needle.
- Refit the housing.
- Run a test job from a port that offers 28 V at 5 A.
- If the head still stops, contact SeedHammer. It has sent replacement parts for this fault.

## If it does not work {#troubleshooting}

| Symptom | Cause | Fix |
| --- | --- | --- |
| Shallower than the test plate from the box | A 20 V supply; the test plate was engraved at 28 V and may have run twice | Use a 28 V port ([depth](#depth)) |
| Words fade toward the end | X-axis play, a loose brass-nut set screw or a loose lock | [Brass nuts](#brass-nut); close the lock firmly |
| Wavy or wobbly text | A loose Y-axis holder set screw or a slack belt | [Belts and Y-axis holder](#belts) |
| The head stops striking partway, often on the right or in the QR | A weak supply, a loose cable or a sticking needle | [Needle and cables](#needle) |
| "Error: stepper: homing timed out" (older firmware: "Mjolnir2 homing timed out"), or the head scrapes the frame | A loose pulley | [Pulleys](#pulleys) |
| The head starts over the orange frame at power-up | Normal | None |
| A new machine leaves grease on the plate | Normal, in small amounts | None; the grease liquefies as the machine warms up |

<!--
bench-checks:
  - [ ] Dry run with the solenoid unplugged on v1.4.3 completes without "Engraving failed." (watch "stepper: power loss or short circuit" and "engraver: not enough power available"; S_DIAG is watched, cmd/controller/engraver.go:278-287, :363-366); is anything harmed; how to reach the plug; the test plate that came with the machine still in the box of v1.4.3-era units
  - [ ] 60 W 20 V/3 A charger: passes power-up as ap33772s.go reads, then does a full job run or stop with "Error: engraver: not enough power available" (device-screen-map bench check 5, shared with power-and-first-start)
  - [ ] Photos of brass-nut and pulley set screws, belt tensioning, Y-axis holder, red housing (how it comes off) and needle, both head cables; two set screws per pulley, four in total (SYNTHESIS Q39); how tight a belt should be; brass-nut, belt and Y-holder fixes are community answers, confirm on a unit
  - [ ] Depth at 20 V versus 28 V on the same job, measured (shared with power-and-first-start); head stays put after a failed job; power-up homing failure is silent (statement and Check now on troubleshooting#engraving); cause and fix for "stepper: x-axis blocked" and "stepper: y-axis blocked". Retry, resume and restart on a partly engraved plate: owned by troubleshooting#retry (harmonized 2026-10-10)
  - [ ] Oil on the plate keeps steel from building up on the tip: a community answer only (HARVEST.md cluster "Does the engraving needle dull...", timeline 2026-08-07), no company statement or bench observation, so the page gives the oil step without the reason
  - [ ] "mjolnir2: buffer underrun" is the "buffer overrun" owners report (Q41); when "engraver unavailable" can appear on hardware (the error table moved to troubleshooting#engraving, the owner of message meanings)
  - [ ] Whether v1.4.3-era units ship a test plate (the "bacon" QA plate); every page now says "if yours came with one" plus this Check (harmonized 2026-10-10)
  - [ ] Power draw: the page no longer states "about 30 W on average"; the owner recollection and its bench item live on hardware-and-spares#heat-noise
  - [ ] Edit pass 2026-10-10: "28 V engraves deeper and crisper" rests on a company statement and one owner, not a measurement, so the body marks it unverified (see the depth item); threadlocker on the brass-nut set screw is one owner's practice, so the body calls it optional; moved out of the body: with the solenoid unplugged, firmware before v1.4.3 ran the whole job without striking. Body "Check:" sentences (test plate in the box, unplugged run on v1.4.3, how the housing comes off, belt tension) were already items above. Deleted as duplicate: the hammer hold, owned by power-and-first-start#plate
sources: manuals/REVIEW-wave1.md sections 2 to 4 (harmonized, see manuals/HARMONIZE-wave1.md); research/device-screen-map.md 1.1, 1.2, 1.5, 1.7, 1.8, 2, 5.1, 5 item 4, bench check 5; research/BRIEF.md (power 20 to 28 V at 3 A, no dry-run); firmware v1.4.3 ea4b65b (lab/check/upstream): cmd/controller/platform_sh2.go:154-161 (needle pulse 5 ms at 20 V, 4 ms at 28 V), :184-192 (origin, homing distance), :225-238 (power-up homing, error discarded), :428-441 (refused supply reboots into upgrade mode), :533-546 ("engraver unavailable"), :548-584 (homing before each job; Close skips the final homing after a write error), cmd/controller/engraver.go:115-118, :186-217, :278-287, :363-379, driver/ap33772s/ap33772s.go:100-139 (highest voltage in 20 to 28 V regardless of current, then "no suitable current found" if that offer is under 3 A), driver/mjolnir2/mjolnir2.go:240-248 ("mjolnir2: buffer underrun"), gui/gui.go:2173-2280, gui/engraver.go:76-97, :132-160, gui/ has no voltage display; lab/check: go run . check -v ../fixtures/2of3-descriptor.txt (lab/fixtures/2of3-descriptor.txt: TEXT ONLY 10m47s, QR ONLY 39m6s, planner time without homing per lab/check/README.md); harvest SYNTHESIS Q02 Q37 Q38 Q39 Q40 Q41 Q42, section D rows on fading, wobbly, striking and homing, section F row 10 (voltages, not watts); HARVEST.md clusters "Why is my engraving shallower than the sample plate that came with the machine?", "Does the engraving depth need adjusting after many plates?", "A brand-new SH II leaves grease on the plate...", "The engraving head moves but stops striking partway...", "Text comes out wobbly or wavy...", "Why are the last words on a plate almost illegible...", "Is there a test mode..."; HARVEST.md timeline 2025-10-14 (company: unscrew the red head housing, oil the needle)
-->