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 Mesa allocation — which signal lands on which card¶

9 September 2026. Proposed, for sign-off before any terminal list is generated.

Grant's requirement, and the only one:

"Arranging the landing onto the Mesa in a logical way so that the ribbon cables can be kept as neat as possible would be my only requirement. It will make accuracy of landing the right colours in the right spots easier."

✅ That is the right constraint to optimise for, because it is the one that prevents mistakes. Nothing else binds: after the 18.5 mA measurement, current is a non-issue, and there is spare capacity everywhere except outputs.

The shape of the problem¶


Each ribbon	24 signal ways, all of one type — three natural groups of eight, because that is how the modules chain
Each Mesa block	24 terminals: 16 inputs then 8 outputs

🔴 So the group of eight is the natural unit on both sides. A ribbon is three of them; a Mesa block takes two as inputs, or one as outputs.

That single fact decides the whole layout.

🔴 The consequence nobody can design around¶

24 output ways = three groups. Each card has only two output groups (16 outputs).

So no card can take a whole output ribbon. Every output ribbon must split across at least two cards. The best achievable is exactly two — and that is what this proposal does.

Input ribbons have no such problem: 24 input ways fit as 16 + 8 across two blocks on the same card, so each input ribbon stays entirely on one card.

The proposal¶

Ribbon	Ways	Card	Block	Terminals
`CN1` inputs	1-16	7I84U	`TB3`	1-16
	17-24	7I84U	`TB2`	1-8
`CN4` inputs	1-16	7I77U	`TB8`	1-16
	17-24	7I77U	`TB7`	1-8
`CN10` outputs	1-8	7I84U	`TB3`	17-24
	9-16	7I84U	`TB2`	17-24
	17-24	7I76EU	`TB6`	17-24
`CN13` outputs	1-8	7I76EU	`TB5`	17-24
	9-16	7I77U	`TB8`	17-24
	17-24	7I77U	`TB7`	17-24
Home switches ×3	—	7I76EU	`TB6`	5, 6, 7

Why this arrangement¶


✅ Every group of eight stays contiguous	Ways 1-8 land on eight consecutive terminals, 9-16 on the next eight. The wire bundle stays parallel and never crosses over
✅ Each input ribbon lives on one card	`CN1` is entirely 7I84U, `CN4` entirely 7I77U. Two connectors, two destinations
✅ Each output ribbon touches only two cards	The theoretical minimum. `CN10` is 7I84U ×2 groups plus 7I76EU ×1; `CN13` is 7I77U ×2 plus 7I76EU ×1
✅ Each card owns one input ribbon and pairs of output groups	The 7I84U is the `CN1`/`CN10` card, the 7I77U is the `CN4`/`CN13` card, and the 7I76EU takes the one leftover group from each output ribbon
✅ Way number = terminal number, within each group	Way 1 to terminal 1, way 9 to terminal 9, way 17 to terminal 17 of its block. Nothing to look up while wiring

🔴 Land the unused ways positionally — leave the terminal empty¶

Seven input and seven output addresses carry no wire at the module end: IF.5, 6, 19, 41, 42, 47, 48 and OF.4, 5, 31-34, 39.

Do not close the gap. Skip the wire, leave the Mesa terminal empty, and let the next way land on the next terminal as if nothing were missing.


✅ The rule stays "way n to terminal n"	With no exceptions to remember and nothing to consult mid-job
✅ The empty terminals become spare outputs	Seven of them, available later for anything new
⚠️ The cost is seven unused terminals	With 96 inputs available for 44 needed, and outputs landing at exactly 48 of 48, this costs nothing that is needed

The axis signals — unchanged, and on the 7I77U¶

These come from the four D-subs, not the ribbons, and were already fixed by the hardware:

	Card	Block	Terminals
Encoders X, Y, Z	7I77U	`TB3`	channels 0, 1, 2 — all 24 terminals
Velocity commands X, Y, Z	7I77U	`TB5`	`AOUT0`, `AOUT1`, `AOUT2` + adjacent `GND`
Servo enables X, Y, Z	7I77U	`TB5`	`ENA0±`, `ENA1±`, `ENA2±`
Spindle speed command	7I77U	`TB5`	`AOUT3` + `GND`

⚠️ Note the 7I77U therefore carries the most: an entire input ribbon, two thirds of an output ribbon, all three encoders, all four analog outputs and all three enables. It is the busiest card by a wide margin. If physical crowding turns out to be the problem rather than logical tidiness, this is the card to reconsider.

⭐ Spindle analog on `AOUT3`, not the 7I76EU's `TB4`¶

The 7I76EU has a purpose-built spindle connector, but its analog output needs an external 5-20 V reference supply on SPINDLE+/SPINDLE- and scales to it. The VFD wants 0-10 V.

AOUT3 on the 7I77U is a true ±10 V output needing no reference, and three of the six analog channels are spare. ✅ Simpler, and it keeps every analog signal on one block.

⭐ Optional — spindle speed feedback for free¶

7i76euman gives TB6 inputs 0-3 as analog-capable, 8-bit. The VFD's AM-AC pair outputs a voltage proportional to actual output frequency at unity gain, and the Tecnos never used it.

🔴 So reserve TB6 terminal 1 (INPUT0/ANA0) and run one pair from the VFD. That gives LinuxCNC real spindle speed rather than the commanded value.

⚠️ It is 8-bit over a 36.3 V full scale, so roughly 70 counts across the VFD's 0-10 V — coarse, but enough to answer "is the spindle actually turning at about the right speed". The home switches are on terminals 5, 6 and 7 in this proposal specifically to keep the analog-capable terminals free.

🔴 Two things to check before wiring a single conductor¶


🔴 The 7I77U board revision	Mesa's own errata: on Rev A boards the `TB7` and `TB8` silkscreen labels are swapped. This proposal puts an entire input ribbon on those two blocks, so getting it backwards misplaces 24 conductors. Check the revision, then label both blocks with tape from the manual's tables before starting
⚠️ `VFIELDA` and `VFIELDB` feed opposite blocks on the two cards	On the 7I77U, `VFIELDA`→`TB7` and `VFIELDB`→`TB8`. On the 7I84U it is reversed: `VFIELDB`→`TB2`, `VFIELDA`→`TB3`. Both must be fed 24 V on both cards — the 7I77U will not start otherwise, even with one block unused

On card diagrams¶

The manuals have them and they are better than photographs — every Mesa manual carries a connector-location diagram with each block labelled.

✅ Print one page from each manual (7i77uman.pdf, 7i84uman.pdf, 7i76euman.pdf, all on Drive) and keep them with the landing list. A photograph shows what is there; the diagram names it.

⚠️ With the one exception above: on a Rev A 7I77U the board is wrong and the manual is right.

mesa-card-reference.md — every terminal on all three cards
housing-connectors.md — the ribbon way map and the formula
io-map.md — every machine signal to its wire number