OEM Special Machine Retrofit: The Most Expensive Mistake Manufacturers Make

A special machine has a mechanical lifetime of 25 to 40 years. Its automation part becomes obsolete in 12 to 15 years. That's a fundamental asymmetry of the OEM business, and it's what makes retrofit a massive economic opportunity: for 30 to 50 percent of the price of a new machine, you recover an installation that runs another twenty years. Provided you don't blow the initial scoping.

For ten years we've watched French-speaking Swiss special machine manufacturers burn their fingers on retrofits that go off the rails. The failure pattern is almost always the same, and the root cause is almost never what the client thought at the start. This article dissects the three most expensive mistakes and proposes a five-step method to avoid them.

Why retrofit pays off (vs new)

On a watchmaking assembly machine, for example, the mechanical base represents 60 to 70 percent of the total cost: chassis, axes, spindles, heads, specialized grippers, adapted vision systems. This part doesn't really wear out if maintenance has been proper. What dies first is the HMI (touchscreens have a lifespan of about 8 to 12 years), then the PLC (S7-300 has been out of production since 2023, Siemens support ends in October 2026), then drives and network communication (Profibus to Profinet, Ethernet IP, OPC UA).

Going new means redesigning mechanics, redoing capability tests, requalifying the entire chain. For a precision machine, that's 12 to 18 months of full cycle and a CHF 800k to 2M ticket. A well-scoped retrofit is 4 to 8 months and CHF 250k to 600k.

See our dedicated service for [OEM special machines](/secteurs/oem) for intervention modes detail.

But this economic gain disappears if the project derails. And it most often derails on three specific mistakes.

Mistake #1, underestimating mechanics in favor of software

This is the queen of mistakes. The manufacturer looks at their installation and thinks: "We change the PLC, the HMI, the drives, we put modern TIA Portal in, and we're good." Six months later, the retrofit is stuck because we discovered the original servo-axis doesn't support the new command, the analog 0-10V sensors have been broken for three years (the old PLC compensated in software), the original electrical cabinet doesn't have room for the new modules, and everything has to be redone.

The retrofit rule is that software typically represents 30 to 40 percent of total cost. The rest is electrotechnics (wiring, cabinets, distribution), instrumentation (end-of-life sensors, actuators), and sometimes mechanical refurbishment (motors, encoders, belts). If you cost your retrofit as a pure software project, you're wrong by a factor of 2 to 3.

The right reflex: before any costing, do a physical audit on the machine, in situ, with full component inventory, real condition, and identification of elements that won't survive powering up a new system. Count one to two days of audit for a medium-complexity machine.

Mistake #2, neglecting legacy documentation

Machines that are 15 to 20 years old have variable documentary history. Often the original electrical schematic exists as PDF but no longer reflects the installation after fifteen years of modifications. The original PLC program has been modified forty times, comments are in early-2000s German, the previous automation engineer retired in 2019, and no one really knows why block FB42 does what it does.

Ignoring this legacy means condemning yourself to reinventing half the functionalities, losing capitalized production recipes, and introducing regressions that operators will discover in operation. A French-speaking Swiss OEM client told us they had to redeliver a retrofitted machine three times in six months, each time for a functionality the operator considered obvious and that the new code didn't reproduce.

The right reflex: before touching anything, clean reverse engineering of existing software. Read legacy code, document critical functions, interview operators on what works and what doesn't, capture recipes and degraded modes. This phase typically takes 5 to 10 days on a medium machine, and it avoids weeks of after-sales after startup.

See also our [data and AI](/services/dataAi) approach for valuing legacy data during retrofit.

Mistake #3, forgetting operator training

The original machine had an HMI with specific screen logic, shortcuts learned by heart, alarm messages whose implicit meaning the operator knew. You deliver a modern HMI under WinCC Unified or Aveva with touch navigation, responsive design, color codes revised per ISA-101. Technically, better. Operationally, your operators lose 30 minutes per format change for the first three months because they look for functions where they used to be.

The mistake isn't modernizing, it's modernizing without supporting. One day of training per operator team, laminated reflex sheets placed on the machine, on-site presence during the first two weeks of production: that's CHF 5k to 8k on a CHF 400k project, and it makes the difference between a retrofit perceived as success and one perceived as regression.

The right reflex: include training in the scope from the initial quote. And have HMI ergonomics validated by a senior operator before startup, not during.

Five-step method

Here's the method we apply on OEM retrofits in French-speaking Switzerland, refined over twenty-some projects in recent years.

Step 1, audit of existing (1 to 2 weeks)

Complete physical inventory of the machine. List of all automation and electrical components with reference, firmware version, condition (OK, drifting, end of life). Recovery of all existing documentation: electrical schematics, PLC programs (S7-300, S7-400, old Allen-Bradley, ABB AC500), HMI recipes, operator manuals, quality procedures. If some documents are missing (frequent), reconstitute by reading the installation and code.

Deliverable: audit report with exhaustive list of components to keep, replace, study in second phase. Budget estimate within plus or minus 20 percent.

This step typically costs CHF 8k to 15k and is billable separately. It completely secures the final costing.

Step 2, scope definition (1 to 2 weeks)

Based on the audit, precise scope definition. Hardware migration (typically S7-300 to S7-1500, HMI TP/MP to Comfort Panel or Unified, Profibus to Profinet, sometimes addition of OPC UA connection for MES interfacing). Decision on SCADA if applicable. List of functionalities to keep identical, improve, remove. Validation of operational constraints (acceptable machine downtime window, max switchover duration, degraded mode during transition).

Deliverable: retrofit specification with functional and technical specs, detailed planning, firm quote.

Step 3, batch migration (8 to 16 weeks)

The trap here is doing everything at once. Best practice is breaking into functional batches. Batch 1, PLC and network communication. Batch 2, HMI and SCADA. Batch 3, recipes and business logic. Each batch is developed and tested on a platform (bench with simulator or twin) before integration.

For a complex multi-axis machine, prior simulation under TIA Portal with PLCSIM Advanced is hugely beneficial. It validates 70 to 80 percent of the code without touching the machine, drastically reducing the on-site downtime window.

Step 4, tests and FAT (2 to 4 weeks)

Factory acceptance tests of the new system, ideally with client present. Full functional validation on bench, or directly on the original machine if still in service elsewhere. Verification of cycle times, degraded modes, alarms, interfaces with upstream/downstream systems (MES, site supervision).

Our [practical FAT/SAT pharma guide](/blog/fat-sat-pharma-guide-pratique) details the method, transferable to OEM special machines with adaptations.

Deliverable: signed FAT protocol, list of reservations cleared or scheduled, authorization to start up on site.

Step 5, startup and support (2 to 4 weeks on site)

Physical switchover on the machine during the negotiated downtime window (typically 1 to 2 weeks). Progressive startup, full-load validation, operator training, on-site presence to absorb the inevitable anomalies of the first days. Final documentation delivered and archived.

Plan also a month post-startup with on-call presence for residual adjustments. This presence is what makes the difference between a retrofit that inspires confidence and one that generates after-sales for a year.

Field experience: 18-year-old watchmaking assembly machine

Concrete case anonymized. French-speaking Swiss watchmaking assembly machine manufacturer, machine delivered in 2007, still in production at end client, become impossible to maintain following end of life of original Comfort Panel and ET 200S modules. Initial request: "fast HMI modernization."

Audit in two days on site. Discovery: 14 IO cards at real end of life, 2 servo-drives with uncorrected positioning drift, S7-300 program with 380 blocks of which 60 percent without useful comments, 7 critical production recipes never documented, operators having developed unwritten procedures to bypass three known bugs.

Final scope: S7-300 to S7-1500 migration, HMI redesign under WinCC Unified, replacement of IO cards and both drives, recipe logic redesign based on actual operator procedures. Initial budget estimated at CHF 180k by the manufacturer, audit revealed a real need of CHF 320k. Client accepted because the justification was factual (substantiated audit report) and the alternative (new machine) was at CHF 1.4M.

Lead time 7 months, on-site downtime window 9 days, return to production without any major outage in the first six months. The manufacturer has sold a similar retrofit service to three other clients since, using this project as [commercial reference](/references).

Frequently asked questions

From what age should retrofit be considered?

The trigger isn't age but availability of parts and support. For a SIMATIC S7-300 machine, the time is now: Siemens support officially ends in October 2026, and spare parts are becoming rare and expensive. For an S7-400 machine, count 2027-2028 depending on criticality. For old Allen-Bradley SLC or PLC-5, it's been critical for a long time. Audit your machine fleet every three to five years to anticipate rather than suffer.

How much does a typical special machine retrofit cost?

For a medium-complexity machine (1 PLC, 4 to 8 axes, 1 HMI, standard network communication), count CHF 200k to 400k all-inclusive. For a more complex machine (multi-PLC, integrated vision, many recipes, MES communication), count CHF 400k to 800k. A very simple machine (single-axis, basic HMI) can be done for CHF 80k to 150k. Compare to CHF 600k to 2M for the equivalent new machine.

Can we retrofit while staying on a non-Siemens PLC?

Yes, and it's even often recommended to preserve client fleet consistency. Allen-Bradley CompactLogix, Beckhoff TwinCAT, Schneider Modicon M580, B&R on motion application: all these ranges have their relevant use cases. The choice depends on your maintenance team's skills, the existing software ecosystem at the end client, and integration constraints. Our [architecture design](/services/architecture) approach includes this multi-brand analysis without commercial bias.

What's the main risk of a retrofit?

The number one risk, statistically, is under-scoping mechanics and existing instrumentation. End-of-life sensors, out-of-tolerance actuators, full electrical cabinet, wiring modified without updated schematic: all elements that surface mid-project and make budget and planning drift. The defense is the prior physical audit, serious, on-site. A CHF 10k audit that avoids an CHF 80k overrun is an obvious investment. To discuss a concrete retrofit project, [contact us](/contact).