PLC Driver Coverage Matrix

What we can read today, what we can read in 12 months, what we cannot read at all (yet). The single source of truth for driver scoping before any pilot. Updated quarterly + after every site survey.

Rule. Before we accept a pilot, the PLCs at the customer site must be green or yellow in this table. Red ones are deal-breakers unless we know exactly how to bridge them.

---

Part 1 — Coverage at a glance

Status	Meaning	Action
🟢 Production	Driver shipped + conformance-tested + deployed in ≥1 pilot	Use as-is
🟡 In flight	OSS lib forked; in development; not yet production-grade	OK if pilot timeline allows ≥6 weeks
🟠 Planned	OSS lib exists; we haven’t started; effort sized	Need 2–4 week sprint to ship
🔴 Gap	No usable OSS; would need custom work or vendor SDK	Pilot blocker until resolved
⚫ Refuse	Vendor-locked, license-incompatible, or strategically bad	Decline pilots requiring these unless we wrap with a gateway

---

Part 2 — Phase 1 (locked, must-have for production)

Per PRD — IA-3 Industrial Protocols & Drivers §Deliverable #1. These 4 cover roughly 85% of the PLCs we’ll see in Moroccan industrials.

Protocol	OSS lib (we fork)	Status	Coverage in Moroccan pipeline	Maturity rating
Siemens S7-300/400/1200/1500	robinson/gos7 (Go, 392★)	🟡 In flight	YNNA Steel EAF, Super Cerame kiln, most rebar mills, ceramic, agro	⭐⭐⭐⭐⭐ very mature
Modbus TCP / RTU	goburrow/modbus (Go)	🟡 In flight	Power meters everywhere, Lesieur Cristal PV inverters, drives, weigh systems	⭐⭐⭐⭐⭐ very mature
OPC UA client	gopcua/opcua (Go)	🟡 In flight	Schneider M580, Beckhoff, newer Siemens, Renault Tanger / Stellantis Kénitra Tier-2 lines	⭐⭐⭐⭐ mature
Omron FINS	maestrohub-labs/fins (Go, v0.1.0-mh.1)	🟠 Planned	Agro-food packing lines, some chemical	⭐⭐⭐ emerging — aligns us with MaestroHub

Combined Phase-1 coverage estimate.

• ✅ ~85% of named pipeline accounts readable.
• ❌ ~15% blocked: Honeywell DCS (notably SNEP), Allen-Bradley legacy, possibly Yokogawa.

---

Part 3 — Phase 2 backlog (12-month, by deal-driving priority)

Priority	Protocol	Best OSS option	Status	Why this priority
P1	Allen-Bradley / Rockwell EtherNet/IP	libplctag (C with Go bindings)	🟠 Planned	Common in Super Cerame legacy presses, some Cosumar lines, automotive Tier-2
P2	Schneider UMAS (M340/M580 advanced)	py3comm — partial	🔴 Gap	Modbus covers basic reads; advanced symbolic access needs UMAS
P3	Beckhoff ADS (TwinCAT)	stansfield/go-ads — incomplete	🟠 Planned	Common in newer Tier-2 automotive + pharma; usually accessible via OPC UA on new versions
P4	Mitsubishi MELSEC (SLMP / MC)	paulgriffiths/melsec — partial	🟠 Planned	Pockets in textile, electronics, some Mitsubishi-spec automotive Tier-2
P5	Honeywell Experion DCS adapter	None viable	🔴 Gap	Critical for SNEP — DCS, not PLC. Needs custom work or Honeywell SDK license
P6	B&R Automation PVI	None mature	🔴 Gap	Common at Pharma 5 — usually OPC UA-accessible on newer X20/X90 controllers
P7	HART over Modbus gateway	pyhart (lab-grade)	🔴 Gap	Process instrumentation in chemicals, refineries; usually via vendor gateway anyway

---

Part 4 — The full driver universe (every PLC family we might encounter)

A. Discrete PLCs (the bulk of the market)

PLC family	Vendor	Native protocol	OSS support	Our path	DigiTPME status
S7-300 / S7-400	Siemens	S7 (ISO-on-TCP)	✅ gos7	Native S7	🟡 P1
S7-1200 / S7-1500	Siemens	S7 + OPC UA built-in	✅ gos7 / gopcua	Either; prefer OPC UA when available	🟡 P1
LOGO!	Siemens	Modbus TCP	✅ goburrow/modbus	Modbus	🟡 P1
M340 / M580	Schneider	Modbus TCP + UMAS + OPC UA	Partial — UMAS = gap	OPC UA preferred; Modbus fallback	🟡 partial via OPC UA
Premium / Quantum (legacy)	Schneider	Modbus + Unitelway	Modbus only	Modbus	🟡 P1
CompactLogix / ControlLogix	Allen-Bradley	EtherNet/IP (CIP)	✅ libplctag	EIP via libplctag	🟠 P1 of Phase 2
MicroLogix / SLC-500 (legacy)	Allen-Bradley	DF1 / EtherNet/IP	Older OSS, weak	EIP via libplctag works for newer firmware	🟠 P1 of Phase 2
CJ / CP / NJ / NX	Omron	FINS + EtherNet/IP	✅ maestrohub-labs/fins	FINS native	🟠 P1
CS1 (legacy)	Omron	FINS	✅ FINS	FINS	🟠 P1
MELSEC Q / iQ-R / FX	Mitsubishi	SLMP / MC	Partial OSS	SLMP planned	🟠 P4
TwinCAT runtime (CX / IPC)	Beckhoff	ADS / OPC UA	OPC UA preferred	OPC UA when available; ADS as fallback	🟠 P3
X20 / X90 / Compact	B&R	OPC UA / PVI	OPC UA on new firmware	OPC UA	🟠 P6
WAGO 750 series	WAGO	Modbus + EtherNet/IP	✅ both via existing libs	Modbus	🟡 P1
Axioline / RFC 4072	Phoenix Contact	Modbus + Profinet + REST API	✅ Modbus + custom REST	Modbus + REST when needed	🟡 P1
DVP / AS / AH series	Delta	Modbus	✅ Modbus	Modbus	🟡 P1
XGB / XGI / XGR	LS Electric (Korea)	XGT FENet + Modbus	Modbus only via OSS	Modbus fallback	🟡 partial
Hitachi EHV / HX	Hitachi	Modbus	✅ Modbus	Modbus	🟡 P1
VIPA SLIO	YASKAWA	S7-compatible	✅ gos7	gos7	🟡 P1

B. DCS (Distributed Control Systems — for chemicals, refineries, large utilities)

DCS	Vendor	Native protocol	OSS support	Our path	DigiTPME status
Experion PKS	Honeywell	Proprietary EWS	❌ Nothing usable	OPC UA gateway (Matrikon, Kepware) — adds cost	🔴 P5 — SNEP blocker
Centum VP / CS3000	Yokogawa	Vnet/IP (proprietary)	❌ Nothing usable	OPC UA via Exaopc (Yokogawa’s own gateway)	🔴 Gap — sugar / refining concern
DeltaV	Emerson	Proprietary + OPC UA option	OPC UA on newer	OPC UA — needs DeltaV OPC server	🔴 Gap
800xA	ABB	Proprietary + OPC UA	OPC UA on newer	OPC UA via ABB OPC server	🔴 Gap — possibly relevant for Managem mining
PCS 7	Siemens	S7 + WinCC OPC UA	✅ gos7 + gopcua	S7 for tags, OPC UA for batch	🟡 partial — uses Phase 1 libs
AVEVA System Platform / Wonderware	AVEVA	OPC UA + MQTT (newer)	✅ gopcua	OPC UA	🟡 partial

C. SCADA / HMI / Historian platforms (often sit above DCS)

Platform	Vendor	Read-out protocol	Our path
Ignition	Inductive Automation	OPC UA + MQTT/Sparkplug B native	✅ Sparkplug B native — peer to us, not a barrier
Wonderware Historian	AVEVA	SQL + OPC HDA	OPC UA (newer) or SQL extracts
OSIsoft PI System	AVEVA	PI Web API + OPC HDA	PI Web API REST
iFix	GE	OPC UA	OPC UA
WinCC OA	Siemens	OPC UA	OPC UA
ZenOn	Copa-Data	OPC UA	OPC UA
TopServer	Software Toolbox	OPC UA / Sparkplug B	We talk Sparkplug B; bypass entirely

D. Drives / motor controllers (often Modbus-readable)

Family	Vendor	Protocol	Our path
Sinamics	Siemens	Profinet + S7	gos7 / gopcua
PowerFlex	Allen-Bradley	EtherNet/IP	libplctag (Phase 2)
Altivar	Schneider	Modbus	goburrow/modbus 🟡
ACS series	ABB	Modbus + EtherNet/IP	Modbus 🟡
MR-J / FR series	Mitsubishi	MELSEC + CC-Link	Phase 4

E. Energy meters & analyzers (mission-critical for CBAM)

Family	Vendor	Protocol	Our path	CBAM relevance
UMG 96	Janitza	Modbus TCP	🟡 P1	⭐⭐⭐⭐⭐ direct
PowerLogic PM5350	Schneider	Modbus	🟡 P1	⭐⭐⭐⭐⭐ direct
EM340 / EM530	Carlo Gavazzi	Modbus	🟡 P1	⭐⭐⭐⭐⭐ direct
Diris A / B	Socomec	Modbus	🟡 P1	⭐⭐⭐⭐ direct
GMP252 (CO₂ probe)	Vaisala	Modbus + 4–20mA	🟡 P1	⭐⭐⭐⭐ Scope 1 process gas
HART instrumentation	Various	HART (over 4–20mA)	🔴 P7	⭐⭐⭐ via gateway

F. Fieldbus / industrial Ethernet (transport, not read-out)

Bus	Where you find it	Direct Edge access?	Our path
Profinet	Siemens-heavy plants	❌ No (real-time bus); read via PLC	gos7 / gopcua on the controller
Profibus DP	Older Siemens	❌ No; read via PLC or Profibus-Ethernet gateway	gos7
EtherNet/IP	Allen-Bradley	✅ libplctag (Phase 2)	Yes
EtherCAT	Beckhoff	❌ Specialized; read via TwinCAT	ADS or OPC UA
Modbus RTU (RS-485)	Older drives, old PLCs	✅ via Modbus-TCP gateway (Moxa, etc.)	goburrow/modbus
CC-Link	Mitsubishi	❌ No; via PLC	MELSEC (Phase 4)
DeviceNet	Older AB	❌ Legacy; via gateway	Replace or live with it
IO-Link	Newer sensors	✅ via IO-Link master (Modbus / OPC UA)	Modbus or OPC UA

G. Building / HVAC / utility (sometimes relevant for energy)

Protocol	Where	Our path
BACnet/IP	HVAC, building energy	Use [vellotis/go-bacnet] when needed; Phase 3
KNX	Smart-building stuff	Out of scope for industrial pilots
M-Bus	Energy meters at building level	Modbus gateway available

---

Part 5 — The strategic gaps (red zones)

These are the deal-blockers as of today. Each one needs a decision: invest now, defer, or refuse the pilot.

Gap #1 — Honeywell Experion (DCS)

• Where it bites. SNEP (the most-mature YNNA subsidiary; Honeywell DCS + SAP integrated). Likely also at OCP downstream + some refineries.
• Why no OSS. Honeywell’s EWS protocol is proprietary; no community has reverse-engineered it cleanly.
• Workarounds today:
  • OPC UA gateway via Matrikon ($)orKepware($) — adds ~$5–10k per site, breaks our open-architecture story.
  • PHD (Process History Database) export via SQL/CSV — slow, after-the-fact, not real-time.
• Recommended path.
  1. Don’t take a Honeywell-only pilot in Phase 1 — refuse politely.
  2. By Phase 2, scope a paid Matrikon gateway integration as a differentiator from pure-OSS options.
  3. By Y2, evaluate writing our own minimal Experion adapter if 3+ named accounts depend on it.

Gap #2 — Yokogawa Centum (DCS)

• Where it bites. Sugar refining (potentially Cosumar), some chemicals.
• Why no OSS. Vnet/IP is proprietary; Yokogawa controls everything via Exaopc.
• Recommended path. Same as Honeywell — bridge via Yokogawa’s own OPC UA gateway, charge accordingly.

Gap #3 — Schneider UMAS (advanced symbolic access on M340/M580)

• Where it bites. Anywhere we want to read symbolic tags rather than raw Modbus addresses on Schneider M-series.
• Workaround. Modbus TCP works for ~80% of needs (raw addresses). UMAS is a quality-of-life upgrade, not a blocker.
• Recommended path. Defer to Y2 unless a customer explicitly demands symbolic access.

Gap #4 — HART instrumentation

• Where it bites. Process plants with smart instruments (flow, pressure, pH).
• Workaround. HART-to-Modbus gateways are cheap and widely available (~$200–500). Use them.
• Recommended path. Skip writing a HART driver. Specify HART gateway in the BOM if site has HART devices.

Gap #5 — B&R PVI (legacy)

• Where it bites. Pharma 5 — possibly. B&R is common in pharma.
• Workaround. Newer B&R (X20/X90 with mapp Technology) speaks OPC UA out of the box → use gopcua.
• Recommended path. Site survey check: B&R version. If new = covered. If old = scope out, possibly refuse.

Gap #6 — ABB 800xA

• Where it bites. Possibly Managem mining, ABB-spec refineries.
• Workaround. ABB AbilityTM Edgenius / 800xA Connectivity Server speaks OPC UA. Use that.
• Recommended path. Treat as OPC UA target; cost adds the ABB connectivity server license to BOM.

---

Part 6 — Coverage mapped to the named pipeline

Crossing the named accounts with the driver matrix. This tells us which deals we can engineer today vs need driver investment first.

Account	Likely PLC stack	Driver coverage	Pilot-readiness
YNNA Steel	Siemens S7-1500 (EAF + drives) + Modbus power meters	🟢 Phase 1 covers it	✅ Ready
Super Cerame	Siemens or Schneider on kiln + likely AB on legacy presses	🟡 Phase 1 (Siemens/Schneider) ✅; AB legacy 🟠 P2	🟡 Mostly ready; AB-press a stretch
Lesieur Cristal	Mixed Siemens + Schneider + Modbus PV inverters	🟢 Phase 1 covers it	✅ Ready
Sonasid (VIP)	Siemens S7 mostly + Modbus meters	🟢 Phase 1 covers it	✅ Ready
Dimatit	Likely Siemens + Schneider	🟢 Phase 1	✅ Ready
SNEP	Honeywell DCS + SAP	🔴 Honeywell gap	❌ Blocker — refuse Phase 1; revisit Y2
Cosumar	Mixed; possibly Yokogawa on refinery	🟡 If Siemens/Schneider ✅; if Yokogawa 🔴	Needs survey
Centrale Danone	Likely Siemens + B&R on packing	🟡 Siemens ✅; B&R 🟠 newer = OPC UA	🟡 Mostly ready
Renault Tanger Tier-2	Mixed — newer plants OPC UA standard	🟡 OPC UA covered; AB pockets 🟠	🟡 Mostly ready
Stellantis Kénitra Tier-2	Mixed — varies per supplier	🟡 OPC UA covered	🟡 Mostly ready
LafargeHolcim Agadir-Souss	Siemens PCS 7 likely	🟢 gos7 + gopcua	✅ Ready
Pharma 5	B&R + Siemens likely	🟡 B&R newer = OPC UA ✅; older 🟠	Site survey first
Tanger Med	Mixed: Siemens + custom + lots of energy meters	🟢 Phase 1	✅ Ready
Managem Tizert	Already runs Sparkplug B (Epiroc); ABB possibly elsewhere	🟢 Sparkplug B native	✅ Ready (Tizert)
OCP Group	Mixed + IROC ecosystem	🟢 Mostly Phase 1 + OPC UA	✅ Reachable when relevant

Pipeline-level read. 9 of 15 named accounts are 🟢 ready today with our Phase-1 driver list. 3 are 🟡 (need site survey to confirm). 1 is a 🔴 blocker (SNEP — Honeywell).

---

Part 7 — Decision matrix (build vs buy vs refuse)

When a new pilot brings a protocol we don’t cover, decide using this:

Condition	Decision
OSS lib exists + maintained + Apache/MIT license	Fork-and-improve. Standard Phase-2 path.
OSS lib exists but unmaintained >12 months	Adopt + maintain ourselves or migrate to alternative.
Closed-source vendor SDK exists, license OK, $1–5k cost	Pay it once if 3+ accounts depend; document migration path.
Closed-source vendor SDK, license bad / $10k+	Refuse the pilot. Not strategic.
No OSS, no SDK, customer needs it	Refuse politely + reference to a SI partner.
Could write our own driver in <2 weeks (e.g. via documented protocol)	Write it — adds to the moat.
Could write our own in >2 weeks	Defer — not worth it for one customer.
Customer refuses to allow OSS on their network	Refuse. Wrong philosophical fit.

Rule of thumb. Driver investment > 4 person-weeks per protocol = needs founder approval. Per ADR process.

---

Part 8 — Roadmap & investment sizing

Quarter	Driver work	Person-weeks	Pilots unblocked
Q1 (now → Jul)	Phase 1: gos7 + Modbus + OPC UA + FINS production-grade	~12 (= IA #3 full-time)	YNNA Steel, Super Cerame, Lesieur, Sonasid, Tanger Med
Q2 (Jul → Oct)	Phase 2 P1: libplctag (Allen-Bradley)	~3	Super Cerame legacy lines, Cosumar, automotive
Q3 (Oct → Jan)	Phase 2 P3 + P4: Beckhoff ADS + Mitsubishi MELSEC	~5	Pharma, electronics, Mitsubishi-spec automotive
Q4 (Jan → Apr)	Phase 2 P2: Schneider UMAS (if customer demands)	~4	Schneider-symbolic-needing customers
Y2	Honeywell Experion (P5) — major investment OR partner	~12 OR partnership	SNEP + future DCS accounts

Investment ratio. Driver work is ~25–30% of IA #3’s time; the rest is SDK + tag config schema + conformance harness.

---

Part 9 — Conformance & testing posture

Per PRD — IA-3 Industrial Protocols & Drivers §Driver SDK, every driver passes the same conformance harness before being marked 🟢.

The harness checks:

1. Connect / disconnect — clean session lifecycle.
2. Read 100 tags concurrently @ 1 Hz for 1 hour — load test.
3. Behaviour on PLC restart — auto-reconnect.
4. Behaviour on bad address — clear error, no crash.
5. Behaviour on network blip — store-and-forward picks up.
6. Tag-list reload-on-config-change — no driver restart needed.

Free simulators we use:

• Siemens TIA Portal PLCSIM (free for read).
• ModRSsim2 (Modbus simulator, free).
• open62541 reference server (OPC UA simulator, free).
• FINS simulator (limited; bench equipment helps).
• libplctag’s built-in simulator for AB EtherNet/IP.

---

Part 10 — How to use this matrix

When founder is scoping a new pilot (in 04 — Pipeline Tracker)

1. Find out the customer’s PLC stack (use Diagnostic Question Bank §Dim 9).
2. Cross-reference here.
3. All 🟢 → green light to discovery call.
4. Any 🟡 → site survey first to confirm.
5. Any 🔴 → either bridge (gateway), defer, or refuse.

When IA #3 plans the next driver

1. Look at the named-pipeline column in Part 6.
2. Build the driver that unblocks the next 2 named accounts.
3. Write the ADR (per 08 — Decision Log).
4. Ship behind the SDK + conformance harness.

When GMSI #1 does a site survey

1. List every PLC found (brand, model, firmware) per PRD — OT-IT Integration & Field Phase 1 §Site Survey runbook.
2. Cross-reference here for each.
3. Flag any 🔴 immediately to founder before scoping commitment.

---

Part 11 — What this matrix is not

• ❌ Not a static list — update quarterly + after every site survey.
• ❌ Not exhaustive — exotic Eastern European or Korean PLCs may show up; treat them as 🔴 until proven.
• ❌ Not a substitute for actually testing on real hardware. A site visit overrides this matrix.

---

Part 12 — Open questions to resolve

Question	Owner	Resolve by
Is SNEP’s Honeywell DCS pilot worth a paid Matrikon gateway integration?	Founder + IA #3	Q3 2026
Does Phase-2 P3 (Beckhoff ADS) actually matter or is OPC UA always available on TwinCAT 3+?	IA #3	After 1 Beckhoff site survey
Should we contribute upstream to MaestroHub FINS lib? Open-source goodwill vs internal velocity.	Founder + IA #3	Q2 2026 OKR session
Can we get a free Janitza UMG 96 dev unit for BOM bench? (Vendor relations)	IA #2 + founder	Q1 2026