The backbone, shown by its work.
Five shared competences serve all four divisions. Instead of listing them, we show them — moments from real programs, told the way our engineers tell them. Customer names are shared under NDA.
Engineering & NPI.
New products are industrialized as a defined phase: document review, design-for-manufacturing, BOM optimization, fixture and test-rig development, golden samples. Engineering is quoted and scheduled — never hidden in the piece price.
A machine builder handed over a drawing set with three revisions in circulation at once. A structured DFM review consolidated them into a single released baseline before the first panel was built.
Two sister machines carried near-identical BOMs with different part numbers for the same terminal blocks. Normalizing both lists against one component master cut the purchased line items by a third of the catalog.
A new control cabinet entered series production without an agreed reference unit. The team built and jointly signed off a golden sample, so every later deviation had a physical benchmark to be measured against.
A wiring harness family was due for series release with only manual continuity checks planned. A dedicated test fixture was designed in parallel with the NPI phase and was ready before the first article.
An operator panel required the busbar to be removed twice during assembly. Resequencing the build steps and reordering two subassemblies removed the double handling without touching the released design.
A safety relay went to eighteen-month lead times mid-ramp. An alternate was qualified against the original datasheet, tested in the fixture, and released as a documented second source before stock ran out.
A cabinet door fouled its gasket on roughly one unit in ten. A tolerance stack analysis traced the issue to two hinge dimensions, and a single drawing change made the fit repeatable.
A prototype cabinet had grown undocumented field changes during commissioning. The team back-documented every deviation into the CAD and BOM, so the series version started from what actually worked.
One product line used five connector families where two would do. A DFM pass during redesign consolidated the interfaces, simplifying both the crimp tooling on the line and the spares list in the field.
A drawing set arrived with three wire numbering schemes from three subcontractors. The team agreed on one convention with the customer and reissued the labels, so field technicians read every cabinet the same way.
A program moved from ten units a year to ten a month. Before the ramp, work instructions, fixture calibration, and the golden sample were bundled into one readiness file the whole line could build against.
Electrical design.
EPLAN-based electrical engineering for panels, machines and complete lines — from schematic to manufacturing documentation, built to EN 61439 and IEC 60204-1 practice.
Every project schematic was being drawn from scratch, each with its own symbols. Building a shared EPLAN macro library meant new cabinets started from proven circuit blocks instead of blank pages.
A production hall reported that a single drive fault kept tripping the main feeder. A selectivity study across the breaker cascade found the mismatch, and regraded settings confined faults to their own branch.
An aging servo platform was retrofitted with higher-power drives on the original cabling. Recalculating conductor sizes against the new load profile flagged two undersized runs before energization, not after.
A densely packed panel forced technicians to remove the drive to reach the fuses behind it. The layout was reworked around service access, keeping every maintainable component reachable with the cabinet live-locked out.
An analog measurement signal drifted whenever a nearby drive accelerated. Rerouting the shielded pairs into a separated duct and correcting the shield termination brought the readings back to a stable baseline.
A cabinet designed for a temperate hall was destined for a foundry mezzanine. A thermal analysis at the real ambient temperature drove a filter fan upgrade and one component derating before commissioning.
Field wiring on a machine family failed acceptance twice for mismatched terminal assignments. Generating terminal plans directly from the EPLAN project removed the manual transcription step where the errors were born.
Intermittent sensor dropouts on a long conveyor traced back to voltage drop on the 24 V distribution. A proper supply budget with recalculated cross-sections and an added feed point ended the ghost faults.
An aging press line needed a new safety controller wired into legacy circuits nobody had documented. The team surveyed the cabinet, drew the as-found state, and designed the interface against reality rather than folklore.
Four machine variants each had a bespoke cabinet with its own layout. A common platform design with defined option slots let one build standard cover all four, with the variants reduced to plug-in differences.
A camera trigger line shared a route with motor cables and fired sporadically. An EMC-conscious redesign of the routing and grounding concept gave the vision system a clean trigger without moving the camera.
Software & controls.
PLC, HMI and robot programming as part of a system responsibility. Vision, traceability and MES/ERP connectivity where the process demands it.
Six machine families each carried hand-written conveyor logic with subtle differences. A validated PLC function library replaced them one family at a time, so a fix in one place now propagates to all.
A packaging machine hid its sequence logic in scattered flags no one dared touch. Rebuilding the control as an explicit state model made every transition visible and gave commissioning a language to argue in.
A legacy S5 controller ran a line whose spare CPUs had run out. The team translated the program to a current platform, verified it against recorded I/O traces, and swapped over during a planned stop.
Operators had learned to ignore an HMI that raised forty alarms per shift, most of them noise. Reclassifying alarms by priority and suppressing the chatter left a screen where a red message means something.
A fieldbus dropout took a machine down for hours because the fault text said only "bus error". Station-level diagnostics on the HMI now name the failing node, turning a search into a walk.
Format change on a filling line meant twenty manual parameter edits and the occasional typo. Moving the format data into managed recipes reduced changeover to a selection, with the parameters versioned and checked.
A robot cell and its host PLC disagreed about who owned the part during transfer. A formally defined handshake with explicit ownership states removed the race condition that had scrapped parts at shift start.
A vision system judged parts correctly but the reject gate occasionally acted on the previous result. Tying each inspection result to a tracked shift register position made the routing deterministic at line speed.
Every new servo axis was commissioned from scratch with copy-pasted code. A standard axis object with defined homing, jog, and error behavior made the twelfth axis as predictable as the first.
Three departments read production counts from three screens that never agreed. A single SCADA data path from the PLCs, with counts latched at the source, gave the plant one number to plan against.
A control rewrite for a remote plant allowed only a short commissioning window. Testing the full sequence against a simulated I/O model beforehand meant site time was spent on the plant, not the program.
Testing & quality.
Automated functional testing developed per product; 100% testing in series with a digital test record per serial number. Customer audits welcome.
A control unit shipped on sample inspection until a field return cluster raised questions. Moving to 100% functional test at end of line meant every unit now proves itself before it leaves the building.
Manual testing of a cabinet family took a technician most of a shift per unit. An automated rig with switched loads and a scripted sequence runs the same checks identically, every time, with a printed record.
A quality question about one batch used to mean digging through paper folders. Serial-number-linked test records now tie each unit to its measurements, firmware version, and operator, retrievable in minutes.
A supplier flagged a suspect component lot after units had shipped. Because incoming lots were booked against build serials, the affected units were identified precisely, and the recall stayed a list, not a fleet.
A new subassembly went straight to volume at a previous supplier and drifted from the drawing. Here a full FAI against every drawing dimension caught two deviations at unit one instead of unit one thousand.
A customer audit was announced with two weeks notice. Because test records, calibration certificates, and training matrices were maintained continuously, preparation meant tidying a meeting room, not reconstructing history.
Repaired field units used to return with a delivery note and nothing else. Each repair now ships with a test report showing the fault found, the work done, and the same end-of-line pass as a new unit.
A test rig failed good boards on one voltage rail near the tolerance edge. Reviewing the limit against the component specification and measurement uncertainty separated real defects from measurement noise.
An electronics assembly showed early-life failures that end-of-line testing could not catch. A defined burn-in cycle before final test now surfaces infant mortality in-house rather than in the customer’s machine.
Busbar connections on a power cabinet were torqued and trusted. Controlled torque tools that log each joint against the unit serial turned a habit into evidence an auditor can read.
Line defects were fixed on the spot and forgotten, so they recurred. A short nonconformance loop that records each defect and routes it to a cause review made the third occurrence a design change, not a repair.
Supply chain.
Sourcing, second sources and obsolescence management at the BOM line level; kitting and logistics under the same roof as production.
A program depended on a single supplier for its main contactor line. A second source was qualified while the first was still healthy, so a later force majeure at the primary became a switch, not a stop.
A discontinuation notice for an HMI panel arrived through the standing obsolescence watch, not through a failed order. The program team had a full year to qualify the successor at their own pace.
Assembly kept pausing while fitters hunted parts from bulk stock. Kitting each build into serial-numbered trolleys, staged to the line takt, meant a cabinet starts only when everything it needs is on the cart.
A processor at the heart of a controller reached end of life with the redesign still years out. A last-time buy sized against the service forecast, stored under controlled conditions, bridged the gap deliberately.
A lot of crimp contacts arrived with plating out of specification, visually indistinguishable from good stock. Incoming inspection against the drawing caught it at the dock instead of in a harness on a machine.
Order planning treated every component as if it shipped next week. Classifying the BOM by real lead times and ordering the long poles at design release kept the twelve-month items off the critical path.
A shortage tempted the schedule toward broker stock of unknown provenance for a safety component. The team instead requalified an authorized alternate, keeping the certificate chain intact for every shipped unit.
Spot-buying drives left every project exposed to whatever the market did that month. Frame contracts with scheduled call-offs gave the supplier a forecast and the projects a dependable delivery rhythm.
C-parts like fasteners and ferrules consumed disproportionate ordering effort. Moving them to supplier-managed consignment bins freed purchasing to work the parts that actually decide delivery dates.
Approved alternates lived in individual buyers’ memories and inboxes. Consolidating them into a released alternates file on each BOM line means any buyer can act on a shortage without reopening the qualification.
A customer ramp doubled requirements while suppliers still worked to the old forecast. Sharing the updated demand curve upstream, with confirmed capacity checks, aligned the chain before the shortage could form.