A custom line is usually quoted as mechanics and cycle time. The electrical and control scope arrives later, often as a rack drawing attached to a purchase order that has already been signed. That sequence is where most change orders come from: the frame fits the building, but the cabinet does not fit the supply, and the software has no agreed boundary. Before you review any layout, write down the power, the network, the data and the people who will maintain the line, because those four items shape the cost of a machine far more than the supplier's headline throughput figure.
Start with the supply, not with the equipment list
Build the electrical brief from your own distribution board outward. Record the available voltage, the number of phases, the fuse or breaker rating on the feeder you intend to use, and where that feeder physically sits relative to the planned line. Then note the earthing arrangement and whether any part of the process needs a separate reference. A machine designed for one supply arrangement and wired into another is a rework job, and rework starts at the cabinet, not at the process head.
Split the load before you talk about individual machines. Heating elements, servo drives, vacuum pumps and compressors behave differently at start-up, and that difference decides how many feeders you need and how the cabinet is divided. The equipment this supplier publishes already shows the split: a mains-fed resistive heater drawing several kilowatts, and a hand-pushed unit whose heater runs from a 48 V battery instead. Inside a production line the same separation applies — each drive family wants its own protection, its own switching device and its own path back to the board.
Give the control boundary a name and a line number
Write one sentence that says what your PLC controls, what the machine builder's controller controls, and where a handover happens. In most integration projects the argument is not technical: it is that nobody specified whether the line controller or the cell controller decides when a part is accepted, who commands a stop, and which controller owns a fault. Decide it in the specification, and list the signals for each handover with a name, a direction and an owner. A list of named signals is also what turns a commissioning argument into a five-minute check.
Do the same for the human interface. Say who is allowed to change a recipe, who may bypass an interlock during a set-up, and which screens must be readable by a shift operator rather than an engineer. Access levels written into the specification are cheap; access levels invented during a night shift are an audit finding.
Decide what leaves the line before choosing a protocol
Ask what your MES, ERP or quality system must receive, and at what moment: at part completion, at shift end, or when an operator flags an event. That answer fixes the amount of data and the required timing; the protocol choice follows from it. Keep the interface to a documented set of records with agreed field names. An undocumented live tag list becomes a dependency that nobody can maintain after the builder leaves.
Then specify what the line must keep locally when the network drops. A line that stores completed records and replays them is a nuisance to commission; a line that loses them stops production the first time a switch is replaced. Also state the physical wiring rule: which cables run in which tray, how the network is separated from power cabling, and where the line's switch lives. Those details get ignored until a fault produces intermittent data corruption that no one can reproduce.
Specify safety stops so the line can be restarted, not just halted
Draw the zones first: where a person can be, where a robot or press can move, and which guarded opening serves set-up as opposed to maintenance. Then define what a stop means inside each zone — remove power, hold position, or vent stored energy — and what has to be true before motion is allowed again. Include the components you will not accept as substitutes, such as a software timer standing in for a monitored guard. A stop that is designed only to halt a machine produces a line that nobody dares to restart.
Plan the maintenance envelope while the cabinet is still empty
Name the people who will service the line and what they are trusted to do on their own. If your team replaces drives and sensors, require terminal markings, a terminal schedule and a spare-parts list with real orderable codes rather than module families. If the builder keeps that work, agree the response terms now and record which remote access is permitted, by whom, and for how long. Remote diagnostics are worth having, but an untracked remote connection is the least protected point on your network.
Ask for the drawings as deliverables with a revision number, and insist that the as-built set matches what is installed. A control supplier that has documented processes is easier to audit: you can check hardware testing records and see whether a controller was exercised at all before it shipped.
Write the acceptance test as one shift, not one cycle
A single good cycle proves nothing about a machine. Specify a run that includes a changeover, an intentional stop with a part under the tool, a power interruption during a record write, and a set-up performed by an operator who has never seen the line. Record what the controller displayed, what the data system received and what the operator had to do by hand. If your MES records agree with the machine counters after that shift, the control scope is finished.
Put the scope in the hands of one supplier who will own it
Custom lines fail commercially when the electrical and control work is split across three vendors with one shared deadline. Ask the automation partner directly whether they carry that scope from panel build through commissioning, and what equipment manufacturing automation services include in writing: cabinet build, wiring to the machine, I/O proof-out, fault response and the handover documents. Where the process itself is new, the same partner should be able to say how the design gets validated — integrated engineering team services make sense only if they cover the control engineer as well as the mechanical designer. If the line is a pilot or demonstration cell rather than volume production, say so in that conversation, because automated production line solutions aimed at demonstration need an interface you can reconfigure in front of visitors and data you can hand to a customer, while a production line needs spares, redundancy and a maintenance shift. Both are useful; they are not the same electrical and control brief. Hand over your feeder data, your interface records, your zone drawing and one planned shift, and ask for the cabinet drawing, the signal list and the acceptance test in return — the project either closes there or it does not.
