A PCB stacker may appear to be a simple way to hold boards between processes. On an SMT line, however, its value depends on what happens at the two handoffs: how a board arrives, how the stacker decides it can accept it, what information travels with it, and what happens when the next machine stops. A capacity number or product photo cannot answer those questions.
The better RFQ starts with the line interface. Specify the mechanical, control, data, ESD, and service conditions together, then ask suppliers to demonstrate the complete sequence on representative boards. This approach applies whether the requirement is a stand-alone stacker, an accumulator, or a handling module built into a larger cell.
1. Define the actual board envelope
“Handles boards up to X by Y millimeters” is not enough. Give the supplier the minimum and maximum board length, width, thickness, and mass; allowable warpage; and the location and size of components that extend beyond the nominal outline. Include any carrier or pallet. State whether boards can enter with components on one or both sides, the required orientation, and which edges may be contacted.
Then describe the neighboring machines. Record transfer height and its tolerance, rail width and adjustment method, direction of travel, available footprint, access side, and any obstruction at the infeed or outfeed. An equipment layout drawing and a sample of the most difficult board are more useful than a single “standard PCB” dimension. The acceptance test should use the smallest, largest, heaviest, and most fragile representative assemblies, including a carrier if one is part of normal production.
2. Size the buffer for a stoppage, not a brochure number
A stacker’s stated capacity is useful only in relation to line cadence and the downtime it is meant to absorb. Consider a hypothetical upstream process releasing one board every 20 seconds. If a downstream inspection station stops for two minutes and the upstream process must keep running, approximately six usable board positions are needed to hold the output produced during that stop: 120 seconds divided by 20 seconds per board. That is a planning estimate, not a promised capacity for any particular machine. Initial occupancy, transfer time, restart delay, and whether a position is reserved for handoff can all increase the requirement.
There is a second question buyers often miss: when does the buffer empty? If the downstream station also processes one board every 20 seconds, it cannot reduce the accumulated queue while upstream production continues. In a separate hypothetical case, if downstream runs every 15 seconds after recovery, it processes four boards per minute while upstream supplies three. A six-board queue then needs roughly six minutes to clear, assuming uninterrupted operation. The stacker has bought time; it has not created permanent throughput.
Ask for a timing diagram covering normal running, downstream blockage, buffer full, restart, and buffer empty. Define whether the goal is short-stop absorption, planned decoupling, batch collection, or routing. Those are different jobs and may require different handling logic.
3. Specify the machine-to-machine conversation
For an installed SMT line, identify the interface used by each adjacent machine and the data that must cross the boundary. IPC’s standards table lists IPC-SMEMA-9851 as no longer maintained and IPC-HERMES-9852 as the machine-to-machine communication standard for SMT assembly. “No longer maintained” does not mean existing SMEMA equipment has stopped working; the Hermes initiative explicitly notes that installed equipment may continue to use it. IPC revision table; Hermes Q&A.
The distinction matters. A basic board-ready handoff and a board handoff carrying identity and product data are not the same integration task. Hermes describes a vendor-independent protocol that can pass board identifiers, dimensions, and product type between machines. IPC also describes potential workflows such as automatic width or stacker-height adjustment when the necessary equipment and implementation support them. Those are possible applications of the standard, not features to assume in any offered stacker. Hermes Standard; IPC: CFX and Hermes Work Together.
In the RFQ, name the protocol and version on both sides; request the exact supported messages, signals, and optional functions; and define what happens if a board ID is missing, duplicated, or inconsistent with the recipe. If a factory system needs event data, define that separately. IPC positions Hermes for machine-to-machine board flow and IPC-CFX for broader smart-production information exchange; a claim of “network connectivity” alone does not establish either integration. IPC: CFX and Hermes Work Together.
4. Treat ESD control as a system property
An “ESD belt” label does not prove that an assembled board is protected through the entire transfer and storage path. The EOS/ESD Association describes an ESD protected area as a defined environment with appropriate materials, tools, equipment, and connection to a common ground. It also identifies conveyor belts and automated handlers as equipment that may be designed or modified for ESD control. EOS/ESD Association, ESD Fundamentals Parts 2 and 3; Part 3.
Ask how boards contact belts, rails, guides, trays, or carriers; how those items are grounded or controlled; and which verification method will be used after installation. Align the answer with your site’s ESD control program. ANSI/ESD S20.20 provides requirements for establishing and maintaining such a program, but the buyer still has to assess the installed process. EOS/ESD Association, standards overview.
5. Design for jams and maintenance
The recovery procedure belongs in the purchase specification. Where can a board jam? Can an operator reach a pinch point while clearing it? Which energy sources must be isolated before service? OSHA’s general machine-guarding rule addresses hazards including ingoing nip points, while its hazardous-energy rule applies to covered servicing and maintenance where unexpected startup or energy release could injure workers. The final safeguarding and lockout approach should be reviewed for the actual installation. OSHA 29 CFR 1910.212; OSHA 29 CFR 1910.147.
Request a demonstration of jam recovery, guard opening, emergency stop, power loss, and restart. Watch for boards left in ambiguous positions or released twice after a reset. Define the handoff state the machine must restore before production resumes.
Turn the RFQ into an acceptance test
A useful purchase package contains a board-envelope table, line layout, interface map, expected stop duration, ESD requirements, safety concept, and a short acceptance matrix. At minimum, test normal transfer; blocked downstream; full buffer; missing or duplicate ID where traceability is required; jam; power interruption; and restart. State who supplies representative boards, who signs off factory and site acceptance, and what evidence is recorded.
The selection decision then becomes clearer: the best-fit stacker is the one that handles the real board mix and failure states at the required cadence, with interfaces the existing line can actually use. A nominal capacity or isolated demonstration is only the starting point.