Ask an equipment buyer when the control part of a machine is designed and the usual answer is "during build". That ordering explains why so many retrofit and new-build projects stall: control work gets left until the machine physically exists and is then bolted onto it. Our own engineering work follows a staged route in which control-system development runs alongside the machine design, and a stage is only closed once a document or a test result backs it. The stages below describe how the electrical control side of that work develops, from the first written requirement to the point where a production team can repeat the work without the engineer who drew it.
Before any wiring: who owns which decisions
Most control disputes are not technical. They are ownership disputes. One supplier draws the machine frame, another supplies the cabinet, and nobody signs the point where a motor cable leaves the machine and enters the cabinet. The first stage of control development is therefore a written boundary, covering who specifies the actuators, who selects the sensors, who draws the power feed, who provides the operator interface, and who signs off the final behaviour. A product R&D accelerator for equipment development that stops at a control drawing leaves that boundary open, and the cost surfaces later as a second cabinet, a duplicated cable tray, or two parties testing the same machine against different acceptance criteria.
The buyer's own input has to be fixed at this point too: what the machine must do, what it must refuse to do, what it must do when a sensor disagrees, and how quickly it has to recover. A written functional requirement is the reference every later stage is checked against.
Stage one: process and machine behaviour, not a control list
The first real design work is a process analysis. Each step the machine performs is listed, with what starts it, what ends it, and what can go wrong inside it. Steps that overlap, steps that wait on an outside signal, and steps where a person stands in the path of motion are all marked, because those are exactly the places control logic must be strong. Interlocks and hold conditions come out of this analysis, not out of a preferred controller brand. A control drawing drawn before this analysis reflects how someone imagined the machine, not how it behaves.
Stage two: control architecture and the choice of control level
Now the architecture question is answered: a single controller or several, where the operator interface sits, how the machine communicates with the equipment upstream and downstream, and what must continue to function when a network or a single input fails. This stage also settles what remains in hardware and what moves into software. Emergency stop paths, guarding circuits, and any function that protects a person or a product stay as verifiable hardware behaviour; sequencing, timing, recipes, and diagnostics are where software takes over. The output here is an architecture description and an input/output list that says what every signal is for, so the next stage can size the panel against reality instead of against a guess.
A non-standard automation equipment build usually carries this stage further than a single machine of a repeatable type, because several stations must agree on shared signals and a shared stop behaviour. The architecture is where that agreement gets written down.
Stage three: electrical design drawings and panel layout
With the architecture fixed, electrical design starts: power distribution, motor circuits, sensor wiring, terminal assignment, cabinet layout, cable routing between the machine and the cabinet, and the naming convention that every wire and terminal will follow. Drawings that carry consistent identifiers are what make the build stage cheap; drawings that do not are what make commissioning expensive. A single engineering team covering structure, electrical, and control handles this work together, because cable paths, connector positions, and panel locations are all physical decisions with a mechanical consequence; splitting them across separate desks after the machine is drawn is what produces panels that do not fit and cable routes that cannot be built.
Stage four: control program and operator interface
Only now does the control program get written, and its structure follows the process analysis from stage one: automatic sequence, manual handling, fault handling, recovery, and safe state each get their own place, so a change in one does not disturb the others. The operator interface is designed in the same pass, not afterwards. An interface that reports only a machine state is nearly useless at the line; an interface that shows which step the machine is in, which condition is holding it, and which action an operator is permitted to take is the difference between a short stop and a long one. Recipe handling, alarm text, and the prompts that appear during changeover are all part of this stage.
Stage five: prototype build, testing, and the list of findings
The prototype is where the control system is tested rather than demonstrated. Every interlock is challenged on purpose. Every fault a plant can realistically create — a lost sensor signal, a jammed mechanism, a power interruption mid-cycle, an operator opening a guard — is simulated, and the machine's response is recorded against the written requirement from the first stage. Whatever fails becomes a findings list, and that list, not a clean demonstration, is the deliverable of this stage. Prototype testing that ends without an honest findings list has not tested much.
Program changes made during this stage have to be reflected back into the drawings. This is where projects go wrong quietly: the prototype works, the code in it is three revisions ahead of the drawing set, and the machine that is later replicated is built from stale documents.
Stage six: documentation freeze and handover to production
A machine that can only be built by the engineer who designed it is not finished. Before control work is considered complete, the drawing set, input/output list, program with a recorded version, alarm list, and the testing results are frozen as one package, so that the same machine can be built again, repaired by someone else, or modified later without any guesswork about what the designer intended. The acceptance test for this stage is practical: can a technician who never met the designer trace a fault from the operator interface to a terminal and identify the cause?
Stage seven: field commissioning and the second round of tuning
Field commissioning proves nothing on its own; it shows whether the control system copes with this particular plant, these inputs, and these operators. Material variation, the behaviour of equipment feeding the machine, and staff habits all generate findings the prototype could not generate. A sensible plan allows for a second round of tuning once the first production weeks are done, with drawings and program updated again so the delivered machine and the delivered documents describe the same thing.
How to tell whether a supplier has actually done this work
Ask what closes each stage. A supplier who has developed control systems can name the artefacts without hesitation: the functional requirement, the interlock list, the input/output list, the drawing set, the findings list from prototype testing, the versioned program, and the acceptance condition. A supplier who has not will describe the work in verbs — they will "handle the controls", "take care of programming" — and the absence of named documents is itself the answer. A second test is the handover question: who maintains the machine after the project team leaves, and what documents do they hold once that team is gone. A Shenzhen equipment and control-system engineering company that has repeated this sequence across many builds can walk through those artefacts in order; a team assembled for one project usually cannot. The same staged route — defined documents and test results at each step — applies whether the equipment being developed is a cleaning system or a production line.
The stage most buyers try to skip, and what skipping it costs
The stage most often compressed is the process analysis in stage one, because it feels like paperwork while the drawing feels like progress. The cost shows up in the last two stages: interlocks added during commissioning, cabinets enlarged on site, cable routes relocated, and control logic patched while the line is already expected to run. It also shifts maintenance onto whichever individual still remembers why the machine behaves as it does. Fixing the requirement and the behaviour on paper first is not a slower route; it moves the expensive corrections to a moment when they are still only corrections on paper.
