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.

Mesa Housing internal wiring schedule

2 September 2026. Every connector inside the Housing, every pin, what it does, and which Mesa card and channel it lands on. This is the build document.

🔴 Read this before wiring anything

✅ Definitive	The Housing-side connectors and pins (from the SGM connector schedules, pages 6-8). Which Mesa card and channel every signal goes to. The I/O address to module terminal to wire number chain
⚠️ Structural, verify on the card	Mesa terminal-block letters (TB3, TB5, TB7…). Taken from Mesa documentation reached by search, not from the manuals themselves - egress to mesanet.com is blocked from this environment. Mesa silkscreens every signal name beside its terminal, so read the board and use the channel numbers below as the authority
🔴 Not yet known	Which way of each 26-way ribbon carries which I/O point. This is the one missing column, and it must be surveyed with the Monoscheda still in place - see the last section
🔴 Not yet known	The ribbon logic voltage, which sets the Mesa field-power domains - see "Field power" below

Download the three manuals before starting: 7i76eman.pdf, 7i77man.pdf, 7i84man.pdf from mesanet.com/pdf/parallel/. Check every terminal number below against them.

1. The card stack

PC  --Ethernet--> [panel-mount RJ45 on Housing]  --> 7I76EU
                                                      |
                                          P1 --DB25M-IDC26--> 7I77U   axes + I/O
                                                      |
                                       SSerial --CAT5E-1-TB6--> 7I84U  I/O only

Card	Used for	Not used
7I76EU	Host Ethernet. 24 inputs (CN1 ribbon), 16 outputs	Its 5 step/dir channels, its spindle analog, its spindle encoder
7I77U	3 analog axis commands + spindle analog, 3 encoders, 3 servo enables, 3 home switches, 20 inputs (CN4 ribbon), 16 outputs	Analog 4-5, encoders 3-5
7I84U	9 outputs (7 spare)	All 32 inputs

2. Power inside the Housing

Rail	Source	Feeds
24 V DC	Existing -G3 Cabur CS6, wires 014 (+) and 12 (−)	Mesa VIN on all three cards
Field power	See §7 - depends on the ribbon voltage	Mesa field I/O sections
5 V analog	⚠️ The 7I77's TB1 is a 2-pin supply input for its analog / encoder / RS-422 section. Confirm from the manual whether this is fed over the DB25 or needs its own supply - if it needs one, that is a part to buy
~~24 V AC~~	The old CN5P feed to the Monoscheda	Goes away. Nothing in the Mesa build wants 24 V AC

⚠️ Mesa field power must ramp in under 10 V/ms, so it cannot be switched by a relay or contactor. Take it from a permanently-on rail.

---

3. The axis connectors - CONN.X, CONN.Y, CONN.Z

Three identical 15-pin D-subs on the Monoscheda front panel, labelled AXIS 1, AXIS 2, AXIS 3. Never cut a loom - mate with the loom's existing connector.

🔴 What the connector is called

Proper name	DA-15 - D-subminiature, A shell size, 15 contacts in two rows (8 + 7)
What suppliers call it	"DB15". Technically wrong - DB is the 25-way shell - but it is what most catalogues list it under, so search both
⚠️ Do NOT confuse with	HD-15 / DE-15, the 15-pin three-row high-density VGA connector. Same pin count, completely different part, and it is what a naive "15 pin D-sub" search returns
Cable end (the Housing end of the drive loom)	🔴 MALE / pins - DA-15P. Verified physically by Grant 2 Sep 2026, and matching MASCHIO in the device list. An earlier entry here said female; that was wrong
Housing panel end	🔴 FEMALE / sockets - DA-15S. This is the half to buy
The plastic shell over it	A backshell or hood
Originals	Zaniboni 1402005 connector (CONNETTORE VASCHETTA 15PIN, MASCHIO) + 1413014 cover, 025ZAN0020 / 025ZAN0026

VASCHETTA is the Italian trade term for a D-sub shell - literally "little tray".

So Mesa needs the FEMALE half: DA-15S panel-mount, metal shell, IDC ribbon termination. ⚠️ Check the listing photo shows two rows, not three - most parts sold as "DB15" are the three-row HD-15/VGA connector.

D-sub pin	Conductor	Wire X	Wire Y	Wire Z	What it does	Drive CN1	→ Mesa
8	BN	69	83	97	Velocity command +, ±10 V	5 V-REF	7I77U TB5 AOUT0/1/2
15	WH	70	84	98	Velocity command − (analog reference ground)	6 SG	7I77U TB5 GND of the same 4-pin group
4	BL	63	77	91	Encoder A	33 PAO	7I77U TB3 ENC0/1/2 A+
11	GN	66	80	94	Encoder /A	34 /PAO	ENC0/1/2 A−
3	YE	64	78	92	Encoder B	35 PBO	ENC0/1/2 B+
10	PK	67	81	95	Encoder /B	36 /PBO	ENC0/1/2 B−
5	GR	65	79	93	Encoder Z / index	19 PCO	ENC0/1/2 IDX+
12	WH	68	82	96	Encoder /Z	20 /PCO	ENC0/1/2 IDX−
9	RD-BN	62	76	90	Encoder common	1 SG	ENC0/1/2 GND
13	-	71	85	99	Servo enable /S-ON	40 /S-ON	7I77U TB5 ENA0/1/2 +
6	-	137	138	139	Home switch BOX/BOY/BOZ	none	7I77U input - see §6
7	-	014	014	014	24 V out to the drive	47 +24V-IN	Leave connected. Do not land on Mesa
1, 2, 14	-	-	-	-	Unused	-	-

Channel assignment: X = analog 0 / encoder 0 / enable 0. Y = 1. Z = 2.

🔴 The servo enable is easier than expected

The 7I77's enable outputs are floating isolated switches - a dry contact pair ENAn+ / ENAn−, totally isolated from field power and logic power.

That kills the sourcing-versus-sinking problem raised earlier. The SGDH's /S-ON sinks to 0 V, and a dry contact does not care which way current flows.

Wire it: D-sub pin 13 (wire 71/85/99) → ENAn+; ENAn− → the drive's 0 V. The drive's own +24V-IN on CN1-47 still supplies the opto, exactly as now.

⚠️ ENAn− needs a 0 V return that is the drive's 0 V, not Mesa's. Wire 12 (the -G3 negative) is the same rail the drives sit on, so take it from there. Confirm the enable pair's current rating against the SGDH sequence input, and check whether the manual calls for a series resistor.

Three grounds, kept separate - as TEA built it

Ground	Pin	Goes to
Encoder common	9	7I77U encoder ground
Analog reference	15	7I77U analog ground on TB5
Enable return	(new)	Drive 0 V, wire 12

Do not common them at the Mesa end. TEA separated them for a reason and there is an 8 kW VFD in the cabinet.

⚠️ Screen termination - corrected 2 Sep. The encoder cable is SCH 8x0.25 with a single overall screen, not screened pairs. That is the known weak point.

🔴 An earlier version of this note said to land the screen at the Housing end. That was wrong - it is already landed at the drive. The SGM schedule puts the cable shield on CN1-50, the SGDH's frame-ground pin. Screens go down at one end only, so adding a second bond at the Housing would create a ground loop through the cabinet - exactly the fault it is meant to prevent.

What to do: find out what TEA did at the Housing end before changing anything. The AXIS 4 connector in the panel photo has a metal hood, so the screen may well be bonded to the shell there too - some industrial practice deliberately bonds both ends for high-frequency reasons. Check whether the screen is bonded to the hood when a loom is next off, and reproduce whatever is found. Do not improvise a change to a screening arrangement that has worked for 24 years.

✅ DECIDED - IDC D-sub to ribbon (Grant, 2 Sep)

Current was never the issue. Nothing on these connectors exceeds about 100 mA (pin 7's 24 V feed to the drive's sequence-input optos is the largest, at 30-50 mA) against roughly 1 A for 28 AWG ribbon. Ten times margin.

The screening objection is answered by the enclosure. The tails are 200 mm maximum inside a bonded metal Housing, which is the screen. And an IDC transition made by a machine is more reliable than 48 hand-soldered D-sub cups, where adjacent-cup bridges and dry joints are the classic failure.

✅ Bonding is straightforward here: the Housing is unpainted magnetic stainless steel (Grant, 2 Sep). No paint to scrape.

⚠️ Two small caveats that still apply. Stainless passivates - the chromium oxide layer that makes it stainless is mildly resistive - so use star washers to bite through it and get a real metal-to-metal bond. And stainless is a poor conductor next to plated steel or aluminium, so it is a competent screen but not a brilliant one; make sure the Housing itself is properly earthed to the cabinet rather than relying on incidental contact.

🎁 One free benefit: magnetic stainless is ferritic or martensitic (400 series) rather than austenitic 304/316, which gives it some low-frequency magnetic shielding that non-magnetic stainless does not have. Useful next to an 8 kW VFD.

🔴 Two things that will bite when ordering

1. An IDC D-sub takes ribbon with one MORE conductor than pins. DE-9 takes 10-way, DB-25 takes 26-way, and a DA-15 takes 16-way ribbon. Order 16-way, not 15.

2. Ribbon conductor n does NOT go to pin n. The two staggered rows mean the IDC contacts alternate top row, bottom row, all the way across. On a DA-15 (top row 1-8, bottom row 9-15) the standard mapping is:

Ribbon	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
Pin	1	9	2	10	3	11	4	12	5	13	6	14	7	15	8	–

🔴 The table above is the generic interleave. Get the map for the FEMALE part specifically.

For a male and a female IDC D-sub to mate pin-to-pin when joined by a straight-through ribbon, their internal maps must mirror each other. So the female part's ribbon-to-pin map is not the same as the male part's, and a map copied off a male datasheet will be wrong. Take the map from the datasheet of the exact DA-15S bought, and treat the table above as showing the shape of the problem rather than the answer.

⚠️ Related trap while terminating: pin numbering mirrors between genders. On a male mating face pin 1 is top-left; on a female mating face pin 1 is top-right. Easy to get backwards when checking a connector in the hand.

The resulting build table, per axis

Signal	D-sub pin	Ribbon conductor	Rainbow colour
Encoder B	3	5	Green
Encoder /B	10	4	Yellow
Encoder A	4	7	Violet
Encoder /A	11	6	Blue
Encoder Z	5	9	White
Encoder /Z	12	8	Grey
Encoder common	9	2	Red
Servo enable	13	10	Black
Home switch	6	11	Brown (2nd band)
24 V to drive	7	13	Orange (2nd band)
Analog −	15	14	Yellow (2nd band)
Analog +	8	15	Green (2nd band)
unused	1, 2, 14	1, 3, 12	–
no pin	–	16	Blue (2nd band)

⭐ The interleave is a bonus, not a compromise

Look at what it does to the pairs:

Pair	Pins	Ribbon conductors
Encoder B	3, 10	4, 5 - adjacent
Encoder A	4, 11	6, 7 - adjacent
Encoder Z	5, 12	8, 9 - adjacent
Analog	8, 15	14, 15 - adjacent

Every differential pair lands on two neighbouring ribbon conductors. Tightly coupled parallel conductors see near-identical pickup, which is most of what twisting buys for common-mode rejection. TEA's D-sub pinout puts each pair on facing pins across the two rows, which is exactly the arrangement the IDC interleave converts into adjacent conductors - very likely deliberate.

⚠️ One bad adjacency worth fixing for free. The analog pair sits on 14 and 15, right beside the 24 V feed on 13. It is steady DC to an opto supply so the risk is low, but if you want the analog command to have a quiet neighbour, leave conductor 13 out of the fan-out and feed pin 7's 24 V separately from the -G3 rail. Costs one wire.

⚠️ Where pin 7's 24 V comes from at the Housing end is still unconfirmed. The Monoscheda runs on 24 V AC and cannot be the source, so it is almost certainly a tap off the cabinet's -G3 DC rail joined at the panel. Confirm it - Mesa neither sources nor needs it.

⚠️ Screen termination is unchanged and still open. The loom's screen is landed at the drive on CN1-50. Whether TEA also bonded it to the hood at the Housing end is unknown; the AXIS 4 connector in the panel photo has a metal hood. Check when a loom is next off and reproduce what is found rather than improvising.

---

4. The spindle connector - CONN.S

15-pin D-sub, AXIS 4. Only two pins used.

D-sub pin	Conductor	What it does	→ Mesa
8	WH	Spindle speed command, into -R6:1	7I77U TB5 AOUT3
15	BN	Spindle command return, -R6:4	GND of the same 4-pin group

⚠️ Note the colours are reversed relative to CONN.X/Y/Z - WH on 8, BN on 15. Go by pin number, not colour.

The signal goes into the -R6 pot network, which then feeds the VS-606V7 inverter. R6 is a speed override in Mesa's path, the counterpart of FR1 on the feed. LinuxCNC's S word controls spindle speed; the pot trims it.

The inverter's run, direction and fault signals are ordinary I/O points on the ribbons - they are in the tables in §5, not here.

⚠️ There is no spindle encoder into the Housing. The HSD's zero/index sensor is an ordinary input (IF.21, wire E121). Spindle orientation for tool change will have to work from that single pulse, or from the inverter. Do not plan rigid tapping.

---

5. The four ribbons - the I/O

Four 26-way IDC headers on the Monoscheda front panel, running out to the DIN-rail modules in the cabinet.

Header	Carries	Module	Mesa destination
CN1	Inputs IF.1-IF.24	IF1, IF2, IF3	7I76EU inputs
CN4	Inputs IF.25-IF.48	IF4, IF5, IF6	7I77U inputs
CN10	Outputs OF.1-OF.24	UF1, UF2, UF3	7I76EU outputs, then 7I77U
CN13	Outputs OF.25-OF.48	UF4, UF5, UF6	7I77U outputs, then 7I84U

Only the 41 real inputs and 41 real outputs are wired. The seven unused input addresses (5, 6, 19, 41, 42, 47, 48) and seven unused output addresses (4, 5, 31, 32, 33, 34, 39) are skipped, so Mesa channels run consecutively and nothing is wasted. That matters on the output side: 48 ribbon ways against 48 Mesa outputs would have left no margin at all.

⚠️ The mapping below is by address, not by ribbon way. The ribbon way for each address is the missing column - see §8.

Inputs

Ribbon	IF.n	Wire	Module:term	Mesa card	Mesa channel	MW310.EQU symbol
CN1	IF.1	E101	IF1:2	7I76EU	IN0	IF_START_ZONA1
CN1	IF.2	E102	IF1:3	7I76EU	IN1	IF_START_ZONA2
CN1	IF.3	E103	IF1:4	7I76EU	IN2	IF_BLOCCO_AEI
CN1	IF.4	E104	IF1:5	7I76EU	IN3	IF_BLOCCO_DHL
CN1	IF.7	E107	IF1:8	7I76EU	IN4	IF_HOLD
CN1	IF.8	E108	IF1:9	7I76EU	IN5	IF_EMERG_TERM
CN1	IF.9	E109	IF2:2	7I76EU	IN6	IF_CONTR_VUOTO
CN1	IF.10	E110	IF2:3	7I76EU	IN7	IF_AGGR_SU
CN1	IF.11	E111	IF2:4	7I76EU	IN8	IF_LAMA_NON_AGGR_SU
CN1	IF.12	E112	IF2:5	7I76EU	IN9	IF_OVERR0
CN1	IF.13	E113	IF2:6	7I76EU	IN10	IF_OVERR1
CN1	IF.14	E114	IF2:7	7I76EU	IN11	IF_OVERR2
CN1	IF.15	E115	IF2:8	7I76EU	IN12	IF_OVERR3
CN1	IF.16	E116	IF2:9	7I76EU	IN13	IF_MOT_PANT_RES
CN1	IF.17	E117	IF3:2	7I76EU	IN14	IF_PREUTE1
CN1	IF.18	E118	IF3:3	7I76EU	IN15	IF_SBLOCCO_UTE1
CN1	IF.20	E120	IF3:5	7I76EU	IN16	IF_PIST_CAMBIO_SU1
CN1	IF.21	E121	IF3:6	7I76EU	IN17	IF_PIST_CAMBIO_GIU1
CN1	IF.22	E122	IF3:7	7I76EU	IN18	IF_FC_ASSEU
CN1	IF.23	E123	IF3:8	7I76EU	IN19	IF_FC_ASSEV
CN1	IF.24	E124	IF3:9	7I76EU	IN20	IF_FC_ASSEW
CN4	IF.25	E125	IF4:2	7I77U	IN0	IF_VEL_RAGG
CN4	IF.26	E126	IF4:3	7I77U	IN1	IF_GRUPPO_ORIZZ1
CN4	IF.27	E127	IF4:4	7I77U	IN2	IF_GRUPPO_ORIZZ2
CN4	IF.28	E128	IF4:5	7I77U	IN3	IF_GRUPPO_ORIZZ3
CN4	IF.29	E129	IF4:6	7I77U	IN4	IF_END_CYCLE
CN4	IF.30	E130	IF4:7	7I77U	IN5	IF_JOGM
CN4	IF.31	E131	IF4:8	7I77U	IN6	IF_JOGR
CN4	IF.32	E132	IF4:9	7I77U	IN7	IF_JOGP
CN4	IF.33	E133	IF5:2	7I77U	IN8	IF_FOTOC_Z1
CN4	IF.34	E134	IF5:3	7I77U	IN9	IF_FOTOC_Z2
CN4	IF.35	E135	IF5:4	7I77U	IN10	IF_POS_RAGG_CU_6
CN4	IF.36	E136	IF5:5	7I77U	IN11	IF_BIT1_GIOSTRA_CU_6
CN4	IF.37	E137	IF5:6	7I77U	IN12	IF_BIT2_GIOSTRA_CU_6
CN4	IF.38	E138	IF5:7	7I77U	IN13	IF_BIT3_GIOSTRA_CU_6
CN4	IF.39	E139	IF5:8	7I77U	IN14	IF_BIT4_GIOSTRA_CU_6
CN4	IF.40	E140	IF5:9	7I77U	IN15	IF_BIT5_GIOSTRA_CU_6
CN4	IF.43	E143	IF6:4	7I77U	IN16	IF_PINZA_SU_CU_6
CN4	IF.44	E144	IF6:5	7I77U	IN17	IF_PINZA_GIU_CU_6
CN4	IF.45	E145	IF6:6	7I77U	IN18	IF_LIVELLO_OLIO
CN4	IF.46	E146	IF6:7	7I77U	IN19	IF_RES_GIOSTRA_CU_6

7I76EU inputs used: 21 of 32. 7I77U inputs used: 20 of 32 (plus 3 home switches).

Outputs

Ribbon	OF.n	Wire	Module:term	Mesa card	Mesa channel	MW310.EQU symbol
CN10	OF.1	A201	UF1:2	7I76EU	OUT0	OF_CNC_OK
CN10	OF.2	A202	UF1:5	7I76EU	OUT1	OF_SBLOCCO_CAMPO_AEI
CN10	OF.3	A203	UF1:8	7I76EU	OUT2	OF_SBLOCCO_CAMPO_DHL
CN10	OF.6	A206	UF1:17	7I76EU	OUT3	OF_RIF_Y_CAMPO_A
CN10	OF.7	A207	UF1:20	7I76EU	OUT4	OF_RIF_Y_CAMPO_D
CN10	OF.8	A208	UF1:23	7I76EU	OUT5	OF_LAMPADA_AEI
CN10	OF.9	A209	UF2:2	7I76EU	OUT6	OF_MOT_PUNTE_VERT
CN10	OF.10	A210	UF2:5	7I76EU	OUT7	OF_PUNTA_VERT1
CN10	OF.11	A211	UF2:8	7I76EU	OUT8	OF_PUNTA_VERT2
CN10	OF.12	A212	UF2:11	7I76EU	OUT9	OF_PUNTA_VERT3
CN10	OF.13	A213	UF2:14	7I76EU	OUT10	OF_PUNTA_VERT4
CN10	OF.14	A214	UF2:17	7I76EU	OUT11	OF_PUNTA_VERT5
CN10	OF.15	A215	UF2:20	7I76EU	OUT12	OF_PUNTA_VERT6
CN10	OF.16	A216	UF2:23	7I76EU	OUT13	OF_PUNTA_OR1
CN10	OF.17	A217	UF3:2	7I76EU	OUT14	OF_PUNTA_OR2
CN10	OF.18	A218	UF3:5	7I76EU	OUT15	OF_PUNTA_OR3
CN10	OF.19	A219	UF3:8	7I77U	OUT0	OF_CONTRO_PRESS
CN10	OF.20	A220	UF3:11	7I77U	OUT1	OF_LAMA_NON_AGGR_GIU
CN10	OF.21	A221	UF3:14	7I77U	OUT2	OF_MARCIA_INVERTER
CN10	OF.22	A222	UF3:17	7I77U	OUT3	OF_INV_INVERTER
CN10	OF.23	A223	UF3:20	7I77U	OUT4	OF_PANT1_SU
CN10	OF.24	A224	UF3:23	7I77U	OUT5	OF_PANT1_GIU
CN13	OF.25	A225	UF4:2	7I77U	OUT6	OF_PUNTA_VERT7
CN13	OF.26	A226	UF4:5	7I77U	OUT7	OF_PUNTA_VERT8
CN13	OF.27	A227	UF4:8	7I77U	OUT8	OF_PUNTA_VERT9
CN13	OF.28	A228	UF4:11	7I77U	OUT9	OF_PUNTA_VERT10
CN13	OF.29	A229	UF4:14	7I77U	OUT10	OF_RIF_X_CAMPO_AEI
CN13	OF.30	A230	UF4:17	7I77U	OUT11	OF_RIF_X_CAMPO_DHL
CN13	OF.35	A235	UF5:8	7I77U	OUT12	OF_PINZA_SU_CU_6
CN13	OF.36	A236	UF5:11	7I77U	OUT13	OF_PINZA_GIU_CU_6
CN13	OF.37	A237	UF5:14	7I77U	OUT14	OF_LAMPADA_DHL
CN13	OF.38	A238	UF5:17	7I77U	OUT15	OFLUBRCHIOC
CN13	OF.40	A240	UF5:23	7I84U	OUT0	OF_SBLOCCO_UT1
CN13	OF.41	A241	UF6:2	7I84U	OUT1	OF_ABIL_GIOSTRA_CU_6
CN13	OF.42	A242	UF6:5	7I84U	OUT2	OF_POSIZ_GIOSTRA_CU_6
CN13	OF.43	A243	UF6:8	7I84U	OUT3	OF_BIT1_GIOSTRA_CU_6
CN13	OF.44	A244	UF6:11	7I84U	OUT4	OF_BIT2_GIOSTRA_CU_6
CN13	OF.45	A245	UF6:14	7I84U	OUT5	OF_BIT3_GIOSTRA_CU_6
CN13	OF.46	A246	UF6:17	7I84U	OUT6	OF_BIT4_GIOSTRA_CU_6
CN13	OF.47	A247	UF6:20	7I84U	OUT7	OF_BIT5_GIOSTRA_CU_6
CN13	OF.48	A248	UF6:23	7I84U	OUT8	OF_MOT_PANT1

Total 41 outputs: 16 on 7I76EU, 16 on 7I77U, 9 on 7I84U (7 spare there).

Terminal blocks (verify against the manuals): on both the 7I76 and the 7I77 the field I/O splits the same way - one 24-pin block carries inputs 0-15 with outputs 0-7, the other carries inputs 16-31 with outputs 8-15. On the 7I76 those are TB6 and TB5; on the 7I77 they are TB8 and TB7. The 7I84's field I/O is on TB2 and TB3.

---

6. The three home switches

Not I/O points, not on any ribbon. They arrive on pin 6 of each axis D-sub, the one pin on that connector with no drive destination.

Axis	Switch	Wire	D-sub pin	→ Mesa
X	-BOX	137	CONN.X 6	7I77U input, first spare after CN4
Y	-BOY	138	CONN.Y 6	7I77U input, next spare
Z	-BOZ	139	CONN.Z 6	7I77U input, next spare

CN4 uses 20 of the 7I77U's 32 inputs, so there is room. But see §7 - these are 24 V signals and the ribbon inputs probably are not, which may force them onto a different card.

The switches return to rail 14 at the machine junction boxes, so they read as 24 V dry contacts or proximity switches. Confirm the sense (make or break on approach) when the ribbon is measured.

LinuxCNC then homes exactly as TEA did: HOME_SEARCH_VEL to the switch, HOME_LATCH_VEL creep, HOME_USE_INDEX on the encoder index already wired to ENC0/1/2 IDX, HOME_OFFSET = Q.TARAT (X is −10.00 mm, Y and Z zero). Full reasoning in homing.md.

---

7. 🔴 Field power - the open question that shapes the build

Mesa field power does two jobs: it powers the outputs, and it sets the input thresholds. All three cards accept 5-28 V.

The problem is that this build has two voltages:

Signal group	Likely voltage	Count
Ribbon I/O (logic side of the module optos)	Probably 5 V TTL - unmeasured	41 in, 41 out
Home switches	24 V	3 in

If the ribbon is 5 V and the home switches are 24 V, they cannot share a field-power domain. Three ways out, in order of preference:

1. Measure first - the problem may not exist. If the ribbon logic side turns out to sit at 24 V, everything is one domain and this section is moot.
2. Use separate field-power domains. The 7I77's TB2 is an eight-pin field power connector, which suggests more than one VFIELD domain on the card. If the input groups can be powered separately, put CN4 on 5 V and the home switches on a 24 V group. Confirm from the manual.
3. Interpose. Three cheap 24 V relays or optocouplers to bring the home switches down to the ribbon's voltage. Costs three relays and a little panel space, and always works.

⚠️ Note also that the 7I84U ends up carrying only 9 ribbon outputs, so its field power follows the ribbon voltage. Its 32 inputs are free and could take the home switches only if it were powered at 24 V, which it cannot be while driving ribbon outputs.

---

8. 🔴 The one survey that must happen before the Monoscheda comes out

Nothing above can be built until each ribbon way is matched to its I/O address. That survey needs the machine live with the Tecnos still in place, so it has to happen before strip-out, not after.

What to establish

Question	Why
Which of the 26 ways carries IF.1, IF.2 … and OF.1, OF.2 …	The missing column in §5
Which two ways are not signals	Almost certainly the module logic supply, +V and 0 V
The logic voltage, in both states	Sets Mesa field power (§7)
Sourcing or sinking on the ribbon	Sets Mesa input and output configuration

The likely answer, and how to test it cheaply

24 points on a 26-way strongly suggests ways 1-24 are the points in order, with 25 and 26 the supply. If that holds, the whole map falls out and no point-by-point survey is needed.

Test it at four points per ribbon - way 1, way 2, way 12 and way 24 - rather than all 24:

• Outputs: with the machine live, drive a known output from the MW310 diagnostic screen and meter the corresponding ribbon way at the Housing header. OF.1 (CNC OK) is on continuously, which makes it a free reference.
• Inputs: actuate a known field device and watch the ribbon way change. The four white pendant buttons are the easiest - IF.30 jog−, IF.31 fast, IF.32 jog+, and IF.29 end cycle.

If those four agree with "ways 1-24 in order", take the map. If any disagrees, the full 96-point survey is needed, and that is a long day.

⚠️ Do not probe with the ribbon unplugged and the module unpowered - a floating opto input tells you nothing. Everything here is a live measurement, so treat it as live work: one hand, meter probes only, nothing actuated without saying so first.

---

9. Sourcing the 26-way ribbon connectors

What the connector family is

The four ribbons are flat ribbon cable with IDC (insulation-displacement) connectors - the same family as every IDC-10 / 16 / 20 / 26 / 34 / 40 you have seen inside a PC. Standard: DIN 41651 / IEC 60603-13. A 26-way is 13 positions × 2 rows.

Three parts make up the system:

Part	Where	Note
Box header (shrouded male header)	The fixed half - on the panel or a board	Has the polarising notch and usually ejector latches
IDC socket (female, "ribbon cable socket")	Crimped onto the flat cable	Has the bumps the ejector latches grab
Strain relief clip	Clips over the back of the socket	Cheap, and it is what stops the cable pulling out of the crimp

Pin 1 is the edge with the coloured stripe on the ribbon; the header marks it with a triangle or a 1.

🔴 You probably need headers only, not sockets

The existing four ribbons stay. Their sockets are already crimped on and their other ends stay plugged into the IF/UF modules. When the Monoscheda comes out, the socket ends are simply free.

So what is needed is four male box headers for those sockets to plug onto - not a whole new connector system.

⭐ The recommended answer: DIN-rail IDC-to-screw-terminal breakouts

A 26-way IDC to screw-terminal breakout module has a box header on top and 26 numbered screw terminals below, and snaps onto DIN rail. Phoenix Contact call theirs the FLKM / FLKM 26 family; Weidmüller and Wago have equivalents, and generic versions are cheap.

Four of those inside the Housing and the whole problem dissolves:

Panel connector needed	None. Pass the ribbons through a slot with edge trim, or leave the front open
Soldering	None
What you get	26 numbered screw terminals per ribbon - which is exactly what you want while the way-to-address mapping is still unknown. Probe and label at leisure, then land Mesa on the terminals
Reversibility	Total. Unplug four sockets and the Tecnos goes back

✅ All three measured - Grant, 2 Sep 2026

#	Question	Answer
1	Pitch	22.4 mm over 10 → 2.54 mm, see below
2	Are the headers panel parts?	❌ No. They are soldered to the Monoscheda board
3	Ribbon slack	✅ Plenty long enough

1. The pitch is 2.54 mm - the measurement has an off-by-one in it

22.4 mm over 10 pins is 9 gaps, not 10. 22.4 ÷ 9 = 2.49 mm, which is 2.54 mm inside a 2% reading error on a 24-year-old board.

Pitch	9 gaps (pin 1→10)	12 gaps (pin 1→13)	Body width, 26 way
2.54 mm	22.86 mm ← matches	30.48 mm	≈ 35.5 mm
2.0 mm	18.0 mm	24.0 mm	≈ 28 mm

2.24 mm is not a standard pitch and does not exist, so the only readings that fit are 2.54 mm over 9 gaps.

⚠️ Confirm it once with an unambiguous span before spending money, because 26-way breakout modules at 2.0 mm pitch are rare and would force a different approach:

• Measure pin 1 to pin 13 along one row - 30.5 mm = 2.54 mm, 24.0 mm = 2.0 mm. Twelve gaps makes a miscount obvious
• Or measure the overall shroud width: ≈ 35.5 mm versus ≈ 28 mm. No arithmetic at all
• Or simplest: a standard 0.1" jumper or DuPont socket should slide onto two adjacent pins. If it fits, it is 2.54 mm

2. Soldered headers - this is the good answer

The headers leave with the Monoscheda. There is nothing on the panel to reuse and nothing to match, so:

• ✅ Free hand on the front panel. Design whatever suits Mesa
• ✅ Confirms headers are what to buy, not sockets - there was never a panel connector
• ⚠️ The existing "panel" is the Monoscheda's own front bracket, so when the card comes out the front of the Housing is an open gap. A new front panel or blank has to be made either way - build the ribbon slot and the RJ45 cutout into it

3. Ribbons are long enough

✅ The breakouts can sit on DIN rail inside the Housing. Ribbons pass through a slot in the new front panel with edge trim. No panel connector needed anywhere in the build.

4. What the photographed socket confirms (2 Sep)

A close photo of one ribbon's connector shows:

Seen	Means
Female socket - square holes on the mating face, no pins	Confirms the cable half is the socket. Buy the male half only
Two rows, standard IDC form	Ordinary DIN 41651 / IEC 60603-13 family. Nothing exotic
Red stripe on one edge of the grey ribbon	That edge is pin 1. Mark it on every ribbon before unplugging anything
A polarising key - the pale rectangular tab set into the middle of one long side	The mating header must be shrouded with a matching notch. Reversal is already prevented mechanically
Moulded strain-relief cover already fitted at the rear	Nothing to add. Do not disturb it

⚠️ The shroud key handles reversal, so the pin-and-plug keying suggested below is for a different job: stopping an output ribbon from entering an input breakout. Both risks are real and they need different fixes.

✅ Answered 2 Sep by a panel photo: the headers ARE latched. White shrouded box headers with integral ejector latches, gold dual-row pins, soldered to the Monoscheda board and showing through cutouts in the metal fascia. So buy latched breakout modules - many cheap IDC-to-terminal modules use plain shrouded headers with no ejectors, which hold by friction only, and this machine vibrates.

The easiest way to settle the pitch, using the cable itself

Forget measuring pin centres. Measure the flat cable width, or the socket's mating face:

	2.54 mm pitch	2.0 mm pitch
Flat ribbon width, 26 conductor	26 × 1.27 = 33 mm	26 × 1.0 = 26 mm
Socket mating-face length	≈ 35.5 mm	≈ 28 mm
Holes per row	13	13

Both are single unambiguous measurements with no gap-counting. The photograph is consistent with 2.54 mm but is not proof - one caliper reading across the grey ribbon settles it.

Panel-mount headers - the Assmann AWH option (Grant, 2 Sep)

Grant found Assmann WSW AWH-26G-E232-IDC on DigiKey AU, a panel-mountable 26-way IDC part, and notes the existing Housing slots have no room for its mounting bolts.

The family is right. AWH 26G is Assmann's 26-way 2.54 mm, DIN 41651 IDC range - which independently corroborates the machine's pitch, since this is the standard family that fits it.

⚠️ Egress to DigiKey is blocked from this environment, so the exact suffix could not be verified. Check three things on the datasheet before ordering:

Check	Why
🔴 Gender - it must be a MALE header (pins)	The cables already carry female sockets. Some -IDC parts in this range are the female cable socket, which would give socket-to-socket and be useless. Others are male headers with IDC rear termination, which is exactly right
🔴 Ejector latches	The Monoscheda headers have them and the sockets rely on them. A panel header without latches holds by friction only
Rear termination and mounting	-IDC should mean a ribbon IDC's onto the back - good, no soldering. Get the panel cutout and bolt-hole spacing from the drawing

🔴 The slot problem probably is not a problem

Grant frames this as reconfiguring the existing slots. It may not need reconfiguring, because the panel may not be staying.

The fascia in the panel photo carries FUSE 1, FUSE 2, AXIS 1-AXIS 4 and the LED legends - all Monoscheda features - and the headers behind it are soldered to the Monoscheda board. If that fascia bolts to the card, it leaves with the card, and a new panel gets made from scratch with whatever slot and bolt spacing suits. There is nothing to reconfigure.

⚠️ Establish which it is: does the fascia come away with the Monoscheda, or is it part of the Housing chassis with the card sliding in behind it? That single question decides whether this is a machining job or a free choice.

Either way a new front panel has to be made, because the Monoscheda's D-subs and fuseholders go with it and leave holes. Design the ribbon cutouts, the RJ45 cutout and any blanks into one panel and have it laser or water-jet cut from a DXF.

⚖️ Counting the joints properly - corrected 2 Sep after Grant's review

An earlier version of this section claimed the slot route added no connections against four for the panel mount. That was wrong and the comparison was misleading. The ribbon terminates in a latching header either way - on a breakout board in one case, on the panel in the other. The breakout board is itself a component in the path.

Counting every joint each point passes through:

Route	Mated pairs	New IDC terminations	Screw joints	Unplug from outside	Numbered terminals
A. Slot → breakout board → wire → Mesa	1	0	2	❌	✅
B. Panel header → rear ribbon fanned straight onto Mesa	1	1	1	✅	❌
C. Panel header → internal ribbon → breakout → wire → Mesa	2	1	2	✅	✅

B has fewer joints than A, not more - it deletes the breakout's screw terminal and the wire from breakout to Mesa. Only C is worse than A, and C is the version that was implicitly being argued against.

⭐ DECIDED - panel header, rainbow ribbon, straight onto Mesa (Grant, 2 Sep)

Two points from Grant settle it:

1. Breakout modules eat Housing space. A 26-way FLKM-class module is roughly 100 mm long and 40 mm tall with the header. Four of them plus three Mesa cards will not sit comfortably in this Housing. That rules out both the slot-and-breakout route and mounting a breakout behind the panel.
2. Rainbow ribbon removes the one advantage a breakout had - printed numbered terminals. The colour sequence carries the numbering instead.

So the build is:

machine ribbon (grey, existing, untouched)
   -> panel-mount latching header, bolted to the new fascia
   -> rainbow 26-way ribbon, IDC onto the header's rear
   -> fanned out at the card and landed directly on Mesa terminals

One mated pair, one new IDC termination, one screw joint per point. Fewer joints than any breakout route, and it uses almost no volume - just the connector depth at the panel plus ribbon routing.

Rainbow ribbon - the colour code, and the trap in it

The usual flat-cable rainbow sequence is the resistor colour code with black at ten, repeating every ten conductors:

Way	Colour	Way	Colour	Way	Colour
1	Brown	11	Brown	21	Brown
2	Red	12	Red	22	Red
3	Orange	13	Orange	23	Orange
4	Yellow	14	Yellow	24	Yellow
5	Green	15	Green	25	Green
6	Blue	16	Blue	26	Blue
7	Violet	17	Violet
8	Grey	18	Grey
9	White	19	White
10	Black	20	Black

🔴 Pin 1 is BROWN, not red. On the grey machine ribbons the pin-1 edge is marked with a red stripe, so the instinct is red = 1. On rainbow ribbon red is conductor 2. That single assumption puts every one of 24 points one position out, and it would look plausible all the way to power-on.

🔴 The colour repeats every ten, so colour alone is ambiguous - brown is 1, 11 or 21. Colour plus position is unambiguous, but the fan-out is where position gets lost.

Mitigations, all cheap:

• Keep the ribbon bonded and flat until the last 30-40 mm, then split only what is needed to reach the terminals. Order stays visually obvious and the bonded section is its own strain relief
• Flag conductors 11 and 21 with numbered heat-shrink or a marker at the fan-out - the two places the sequence restarts
• Verify the sequence against the cable you actually buy. Brown-first is the convention, not a standard. The datasheet will state it
• Leave a service loop so a card can be pulled without unpicking the fan-out

⚠️ All 26 ways may be live, not 24. If the two non-signal ways turn out to be the module logic supply, they land too. If they are genuinely unused, cut them back and insulate rather than leaving bare tails.

🛒 The DigiKey order - reviewed 3 Sep 2026

Grant's cart:

Part	DigiKey	Qty	Verdict
AWH 26G-E232-IDC header, 26 pos IDC, gold	AE11150-ND	4	✅ Male header, correct half. ⚠️ Confirm it is panel-mount AND has ejector latches
AWG28-26/F/300 ribbon, 26 cond, 0.05	123-AWG28-26/F/300-DS-ND	10 ft	✅ 0.05 in = 1.27 mm pitch, correct. ⚠️ 10 ft is tight - see below
400F0-15-1-004S D-sub receptacle, 15 pos IDC	LFM15H-ND	4	✅ RCPT = receptacle = female. Correct after the gender fix

🔴 Six things to add or change

1. Buy spares - 6 of each connector, not 4. At AU$3.81 and AU$5.45 these are trivial money, and IDC connectors are single-use in practice: press one on crooked, or with 15 conductors instead of 16, and it is scrap. Two spares of each costs about AU$19 and saves a re-order plus international shipping.

2. More ribbon. 10 ft is tighter than it looks.

Run	Count	At 300 mm
I/O, full 26-way	4	1.2 m
Axis, split to 16 from 26-way	4	1.2 m
Total	8	2.4 m = 7.9 ft

10 ft leaves about 2 ft of margin for eight first-time IDC terminations plus the splitting. Order 15-20 ft.

3. 🔴 An IDC press, or a bench vise with smooth flat jaws. This is the most likely thing to go wrong. IDC connectors must be pressed on square and with even force. Pliers ruin them - the contacts go in crooked and cut through or miss entirely, and the fault is intermittent rather than obvious. A vise with flat jaws and a sacrificial hardwood block works fine and is what most people use.

4. ~~Strain-relief covers~~ - checked 3 Sep: AE11150-ND has no strain-relief option, and it probably does not matter. Strain relief earns its place on connectors that get handled and pulled. This one is bolted to a panel with a captive 200 mm ribbon inside a closed Housing that never moves - consistent with the decision to skip cable restraint entirely. The connector that does need it is the machine ribbon's socket, which already has one.

If a version with strain relief is wanted anyway, the fastest route is DigiKey's parametric filter: Rectangular Connectors, positions 26, pitch 0.050" (1.27 mm), termination IDC, Features Strain Relief. Compare the AWH 26G variants first (-0202-IDC, -0222-IDC, -E202-IDC, -E232-IDC, -E202-IDC-R) since they share one datasheet and the suffix encodes the options. Other families: TE AMP-LATCH (the strongest lead), 3M 3000 series, Wurth WR-BHD. A separate clip may also exist - search IDC strain relief clip 26 way.

⚠️ Do not trade away panel-mount, ejector latches or 2.54 mm two-row to gain a clip that is not needed.

5. D-sub mounting hardware, matched to the loom. ⚠️ Check what retention the loom's male connector uses - the AXIS 4 connector in the panel photo appears to have a metal hood with a clip, not jackscrews. If it is a slide-latch, the receptacle needs the matching posts, not screw locks. Get this wrong and the connectors will not lock together.

6. Panel-mount RJ45 for the Ethernet, which replaces COM1 and COM2. Already on the build list and not in the cart - worth adding while shipping is being paid anyway.

Also worth throwing in

• A label maker or write-on heat-shrink. Eight ribbons that look identical, and a rainbow sequence that repeats every ten conductors. Labelling is not optional here
• Spare ribbon for practice. Press one connector onto a scrap before doing a real one

The original guidance

Four 26-way (13 × 2), 2.54 mm pitch, IDC-to-screw-terminal DIN-rail breakout modules - Phoenix FLKM 26, Weidmüller equivalent, or a generic. Nothing else on the ribbon side.

⚠️ Gotchas

• 🔴 Two ribbons are inputs and two are outputs, and all four connectors are identical. Crossing an input ribbon with an output ribbon is the worst mistake available here. Key them deliberately: pull pin 8 (say) from the two input headers and plug the matching hole in their sockets, and pin 20 on the two output headers - the standard pin-and-plug method. Then an output ribbon physically cannot enter an input header. Do this before anything is powered
• Buy shrouded box headers, not bare pin headers. The shroud and its notch are what prevent a reversed plug. On 24 V outputs a reversed plug is expensive
• Buy latched headers if the existing sockets have latch bumps, which the rack photos suggest. Unlatched headers still accept the socket but nothing holds it in
• Fit the strain relief clips. They cost cents and prevent the failure that is hardest to diagnose later
• Label every ribbon by the module it lands on - IF1, IF4, UF1, UF4 - not by socket name and not by the marker-pen labels, which do not match the drawing

Where to look, and what to search

Supplier
element14 / Farnell AU, RS Components AU, Mouser AU, Digi-Key	Best range for the breakout modules and for genuine Phoenix / Weidmüller parts
Jaycar, Altronics	Plain box headers and IDC sockets, over the counter
eBay / AliExpress	Generic "IDC to terminal block adapter" modules, a few dollars each

Search terms that work: IDC 26 way box header 2.54mm latched, FLKM 26, IDC to screw terminal DIN rail 26 pin, ribbon cable breakout board 26 pin, IDC bulkhead panel mount 26 way (patchy stock at 26-way).

If a true panel-mount is wanted anyway, the usual homemade version is a box header on a scrap of protoboard behind a rectangular cutout on M3 standoffs, with a short internal ribbon to the breakout. It works, it looks original, and it adds one more joint to go wrong.

---

10. Parts to buy

Item	Qty	Note
DA-15S FEMALE panel-mount, metal shell, IDC ribbon termination	4	🔴 Female - the looms are male. ⚠️ Two-row 15-pin, not the three-row HD-15/VGA part. Metal shell, bonded to the panel
16-way ribbon (rainbow preferred)	~1 m	🔴 16-way, not 15 - an IDC DA-15 takes one conductor more than it has pins
Plain hookup wire 0.5 mm²	small roll	Pin 7 24 V feed, if kept out of the ribbon
Star washers	few	Bond each D-sub shell to the panel. The Housing is unpainted stainless, but stainless passivates - the washers bite through it
26-way panel-mount latching box header, male, 2.54 mm, IDC rear termination	4	⭐ The decided route - see §9. Assmann AWH 26G family; confirm gender, latches and cutout on the datasheet
26-way RAINBOW ribbon cable, 1.27 mm	~2 m	For the four internal runs. 🔴 Pin 1 is brown, not red
⚠️ ~~Ferrules for 28 AWG~~	-	0.08 mm² ferrules are hard to source - smallest common is 0.25 mm². Practical answer: land the bare stranded conductor, folded double if it looks loose in the clamp. Do NOT tin the ends - solder cold-flows and the joint loosens
Strain relief clips	4	Cheap insurance
Panel-mount RJ45, IP-rated	1	Replaces COM1 / COM2 on the Housing front panel
DIN rail + ends for the Housing	-	Three cards
5 V supply for the 7I77 TB1	0 or 1	Only if the manual says TB1 is not fed over the DB25
24 V interposing relays	0 or 3	Only if §7 option 3 is needed

Related

• retrofit-boundary.md - scope, and why no field wiring is touched
• servo-drive-interface.md - the drive end of the axis looms
• homing.md - why the drives do not home themselves
• io-map.md - all 82 points with their field devices, terminals and Harting pins
• connector-schedules.md - the SGM pin-by-pin source for §3 and §4
• mesa-hardware.md - the card chain, firmware and order record