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.
The retrofit boundary - how much of the machine gets touched¶
2 September 2026. Consolidates what the wiring diagram proved about CLAUDE.md §5.4, revised the same day against Grant's review. The short version: the retrofit is unplugging eight connectors on the front of the Housing. No field wiring is disturbed, and no field wire is re-terminated.
Naming (Grant, 2 Sep 2026)¶
| Term | Means |
|---|---|
| Housing | The rack enclosure holding the Monoscheda and its Syncro daughterboard. Its front panel carries the four 26-way ribbon headers and the four AXIS D-subs. Becomes the Mesa Housing |
| Cabinet | The outer electrical cabinet (Q.E.) the Housing sits in, with the drives, contactors, breakers, transformers, the DIN-rail I/O modules and the terminal strips |
| Field | Everything past the cabinet's Harting connectors - machine junction boxes, pendant, sensors, valves, switches |
⚠️ The IF1-IF6 / UF1-UF6 I/O modules are in the cabinet on DIN rail, not in the Housing (sheet 38 layout). The four ribbons run from the Housing's front panel out to them.
Where "the controller" is, exactly¶
Whenever these notes say a signal "runs back to the controller", the destination is the Tecnos POSIZIONATORE M68K Monoscheda in the Housing - the card, not a terminal strip. For the axis signals it is more specific still: the AXIS 1-AXIS 4 15-pin D-sub sockets on the Monoscheda front panel (sheet 35). The looms run D-sub straight through to the drive's CN1 with no intermediate terminal strip. The encoder pairs, the analog reference, the servo enable and the home switch all terminate there, on a plug.
That is why the boundary is so clean: the machine's designers already put a connector exactly where the retrofit needs one.
The whole retrofit, as a list of plugs¶
| # | Connector | Carries |
|---|---|---|
| 1-4 | CN1, CN4, CN10, CN13 |
26-way ribbons out to the DIN-rail I/O modules: all 96 I/O points, one wire per point |
| 5-7 | AXIS 1, AXIS 2, AXIS 3 |
Analog command, encoder feedback, servo enable, home switch, X/Y/Z |
| 8 | AXIS 4 (CONN.S) |
Spindle analog reference, into the -R6 override network |
Plus one change of port rather than a disconnection:
COM1 and COM2 |
Both serial ports go. They are replaced by a panel-mount Ethernet port on the Housing, carrying the Mesa link to the PC |
Unplug the eight, lift the Monoscheda out, fit Mesa and the Ethernet port, and the machine is ready. Nothing else in the cabinet is disconnected and nothing on the machine is opened.
❌ What is NOT on this list¶
There is no carousel encoder connector in the Housing. An earlier draft listed -SIG2 as a ninth plug. That was wrong - see the correction below. The Tecnos never reads a carousel encoder.
Why the I/O half needs no field work either¶
This is the part worth being explicit about, because it is the half that could have gone the other way.
The field wiring does not land on the Monoscheda. It lands on the I/O modules - IF1-IF6 for inputs, UF1-UF6 for outputs - which sit on DIN rail in the cabinet. Every tagged field wire (E101-E148, A201-A248) terminates on a numbered screw terminal on one of those modules. The modules do the 24 V conditioning and the opto-isolation. Four 26-way ribbons then carry the logic side of those optos to the Housing.
Mesa connects at the ribbons. The modules stay, the field terminations stay, and not one of the 82 field conductors is undone.
Two things make this sound rather than hopeful:
- ✅ One ribbon wire per point, tested at the boards (Grant, 31 Aug 2026). No multiplexing, no scanned bus, no strobed addressing. 4 ribbons x 24 points = 96 = 48 in + 48 out, matching the
EIN1-48/AUS1-48numbering on sheets 16-28. - ✅ The output cards use
TD62783APsource drivers (read off the chips in the rack photos). An ordinary 8-channel high-voltage source driver with logic-level inputs - exactly what "the ribbon is the logic side" predicts.
What still needs new wire¶
| Item | Work |
|---|---|
| Mesa cards | Mount on DIN rail in the Housing, feed 24 V DC from the existing -G3 rail (wires 014 / 12) |
| Panel-mount Ethernet | Replaces COM1 / COM2 on the Housing front panel. Cable to the PC |
| Ribbon interface | Four 26-way IDC to Mesa terminals. Adapter cables to make up. This is the bulk of the build |
| Axis mating plugs | Four DA-15 plugs and hoods (originals: Zaniboni 1402005 + cover 1413014) so no loom is ever cut |
| I/O module logic supply | If the modules take their logic power from the Monoscheda, feed it from the Housing instead. There is 24 V in the Housing for Mesa anyway, so this is a short run alongside the ribbons - a wiring detail, not a risk (Grant, 2 Sep) |
That is the whole list. Three items that appeared in an earlier draft have come off it, below.
✅ Three things that came off the list¶
Drive fault back to Mesa - not needed¶
An earlier draft called for taking ALM off the spare MY4 poles. Grant is right that this is unnecessary. The fault already stops the machine in hardware: ALM drops KAX/KAY/KAZ, which sit on rail 8A, and those relays feed the summary-fault chain. Nothing depends on Mesa noticing.
Better than that, Mesa gets "a drive faulted" for free. IF.8 is the summary fault input, and the three drive alarms are already among the eleven sources feeding it (sheet 13). It arrives through the ribbon like every other input, with no new wire at all.
What Mesa does not get is which of the three drives. That is a diagnostics nicety, not a requirement:
| If you want per-axis fault | Two spare poles on each MY4 (11-7, 12-8), three wires plus a common, all inside the cabinet |
| Do it | Later, only if hunting faults by drive panel gets annoying |
The carousel encoder - it does not report to the Housing¶
Grant's instinct was right and an earlier note here was wrong. The device list settles it: -E.561 is part 025AMP0001, described as "Amphenol, Panasonic encoder", listed immediately after -M560 "Amphenol, Panasonic motor". Sheet 32's header reads CONNETTORE AZIONAMENTO ASSE 561 - drive connector.
So -SIG2 is the Panasonic MSS083A1XP's own motor encoder input. It closes the indexer's loop and never goes near the Tecnos. That is exactly why it is not one of the four AXIS D-subs on the Monoscheda panel.
The carousel tells the controller where it is over binary I/O, precisely as io-map.md records: COIN in-position plus five position bits P1OUT-P16OUT, on IF.35-IF.40. That is the machine's only carousel position sense as far as any controller is concerned. The Panasonic also homes itself, from proximity switch -SQ560 via relay -KA560 into its own ORGL input.
Consequence: the retrofit needs no encoder input for the ATC, and there is no ninth connector.
The OF.1 watchdog - it is the same wire it always was¶
Grant's understanding is correct: OF.1 is Mesa saying "I am alive and ready", and it is the start of the safety chain - without it nothing unlocks.
On the wiring side there is nothing new to do. OF.1 is UF1 channel 1, wire A201, and it energises rail 8A, which powers KAX/KAY/KAZ and the KAF Z-brake relay. Mesa drives that output through CN10 like any other output. Same module, same terminal, same wire, same rail.
The only thing worth noting for the build - and it is a board-configuration matter, not a wiring one - is that this output should sit under Mesa's own card-level watchdog, so it drops when the PC hangs or the Ethernet link dies, not only when software chooses to clear it. That is what makes a dead controller drop the drives and set the Z brake, exactly as a dead Tecnos does today.
🔴 The two things left to check¶
1. Where the I/O modules get their logic power¶
Not a risk, but it has to be known before an adapter cable is made up. The Tecnos -CNC. takes 24 V AC on CN5P (sheet 35). What powers the I/O modules is not established, and the 26-way ribbons carry 24 points in 26 ways - leaving exactly two spare, which looks like a +5 V and 0 V pair.
If the modules are fed from the Monoscheda through the ribbon, the Mesa adapters must feed that rail instead. There is 24 V in the Housing for Mesa regardless, so it is a short run, but the voltage and which two ways need establishing. Ring out the two non-signal ways against a module, or find the module supply on sheets 33-35.
2. Ribbon logic level and polarity¶
Still unmeasured. One-wire-per-point is confirmed and Mesa's ranges (7I84U field power 5-28 V, sourcing or sinking selectable; 7I76E inputs 5-32 V, outputs 5-28 V) cover 5 V TTL comfortably, so it is a configuration detail. Measure it at the same time as the supply above - one trip, both answers.
✅ The §5.4 exception is closed¶
CLAUDE.md §5.4 carried "One exception: the comms to the servo drive controllers. That connector may need changing." It does not. The axis looms are standard 15-pin D-subs with a pin-by-pin schedule (SGM pages 6-8). Mesa mates them with ordinary DA-15 plugs at the Monoscheda panel. Nothing is cut, nothing is re-pinned, and the Tecnos can be plugged straight back in if the retrofit is ever abandoned.
Related¶
homing.md- who homes the axes, and why the drives cannotservo-drive-interface.md- the axis pin map and what Mesa takes from each loomtecnos-io-rack.md- the twelve ribbons and the three I/O cardsio-map.md- all 82 points, address to module terminal to field devicesafety-and-enable-chain.md-OF.1, rail8A, and what must stay hardwiredmesa-hardware.md- the three-card chain and the order record