Skip to content

Working documentation

This page is generated from the project's working notes. It records a real retrofit in progress, corrections included. Anything marked 🔴 or ⚠️ is unresolved or carries a caveat - read those before acting.

Homing - who does it, and what LinuxCNC has to reproduce

2 September 2026. Written to answer directly: are the Yaskawas self-homing, and if so can TEA's design be duplicated?

The answer

🔴 The drives are not self-homing. The Tecnos homes them, and the home switch signal does come back to the Housing.

The premise that the homing signals do not reach the Housing is understandable - they are absent from the entire PLC map - but they do reach it. They just arrive by a different route from every other machine signal.

Where the home switches actually go

Sheet 22, F.C. O ASSI (FINECORSA ZERO ASSE X / Y / Z, "zero axis micro"). One switch per axis:

Axis Switch Wire Route
X -BOX 137 BOX in =BM+X43 → X43:137 → Harting -B pin 5 → CONN.X pin 6
Y -BOY 138 BOY in =BM+X4 → X4:138 → Harting -C pin 19 → CONN.Y pin 6
Z -BOZ 139 BOZ in =BM+X4 → X4:139 → Harting -C pin 20 → CONN.Z pin 6

Pin 6 of each axis D-sub is the home switch, and pin 6 has no CN1 destination. That is the decisive detail, and it is explicit in the SGM connector schedule (pages 6-8): every other pin on that connector names a drive CN1 terminal, and pin 6's column is blank. The wire runs from the machine, through the cabinet, into the Housing, and stops at the Monoscheda.

The drives never see the home switches at all.

Why they are missing from MW310.EQU

Because they are not PLC I/O. They land on the Syncro axis card in the Housing, alongside the encoder feedback and the analog command, not on the IF1-IF6 modules. The PLC map only ever covered the module I/O, so the homing method could not have been found from the software - which is exactly what happened, and why this went unanswered until the drawing was read.

Why the drives could not home themselves anyway

The SGDH is in speed control (Pn000.1 = 0, proved in servo-control-mode.md). In speed mode the drive has no homing function - it takes a velocity and turns the motor. Homing belongs to whatever closes the position loop, and in this machine that is the Tecnos. The drive contributes exactly one thing to the sequence: the PCO index pulse on CN1-19/20, which it puts out regardless.

So there is no hidden drive-side homing routine to discover, and no drive parameter involved.

The sequence TEA built

Switch plus index, run by the Syncro card, using parameters captured from Syncro.prm:

Step Parameter
1 Rapid toward the switch until the zero micro trips VTAR rap = 100
2 Reverse off and creep VTAR len = 40
3 Latch on the first encoder index pulse (PCO) -
4 Apply the offset to set machine zero Q.TARAT (X: −10.00 mm; Y, Z: 0)

This is the standard, and best, way to home a machine of this size. The switch gets you within one motor revolution; the index pulse gets you repeatability far finer than the switch could ever give on its own.

What LinuxCNC does with it

Nothing needs inventing and nothing needs adding. It is a direct translation:

TEA / Tecnos LinuxCNC
VTAR rap 100 HOME_SEARCH_VEL
VTAR len 40 HOME_LATCH_VEL (opposite sign to search)
Index latch on PCO HOME_USE_INDEX = YES
Q.TARAT HOME_OFFSET
Per-axis order HOME_SEQUENCE - set Z first, as the machine does

Mesa already has both halves of what this needs, on connectors it is taking anyway:

  • The home switch arrives on AXIS n pin 6 - a digital input on the 7I77U
  • The index arrives on the same encoder pair Mesa is already reading, PCO//PCO on pins 5 and 12

So homing costs no extra wiring at all. It is one more input per axis on a plug already being mated.

⚠️ Two small checks before wiring

  1. What the home switch signal looks like electrically. -BOX is drawn with contacts 1-2 returning to X43:14, and rail 14 is a 24 V control rail, so it reads as a 24 V dry-contact or proximity switch pulling pin 6 to the 24 V rail. The 7I77U's field inputs handle 24 V, but confirm the sense (does the switch make or break on approach) and whether pin 6 needs a return, since pin 7 on the same connector is the 014 24 V feed. Measure it at the same time as the ribbon logic level.
  2. Q.TARAT on X is −10.00 mm, not zero. Carry that offset into HOME_OFFSET or X will home 10 mm out. Y and Z are zero.

The Panasonic MSS083A1XP does home itself: proximity switch -SQ560 drives relay -KA560 into the indexer's own ORGL home input. The Tecnos is merely told about it, on IF.46 (EIN46, O ASSE 560), through the ordinary I/O modules.

That behaviour survives the retrofit untouched. LinuxCNC does not home the carousel; it reads IF.46 to know that the carousel did.

  • io-map.md - the home switch routes and the full axis connector pinout
  • servo-control-mode.md - why the drives are in speed mode and cannot home themselves
  • servo-drive-interface.md - the Mesa pin map, including pin 6
  • mw310-recon.md - the Syncro.prm parameter table