A vacuum pressure switch can report that pressure has reached a setpoint, yet a robot may still drop a part. Conversely, an intact part may be rejected because the switch does not turn on before a poorly chosen timeout. Selection therefore starts with the measurement point, the physical condition the signal is meant to represent, the time allowed to reach it, and the response if the condition is lost. This guide is for purchasing, engineering and maintenance teams specifying or upgrading robotic vacuum handling in Thailand. It covers the switch, PLC logic, request for proposal (RFP), factory acceptance test (FAT) and site acceptance test (SAT).
Suction confirmation is different from vacuum-source condition monitoring
A running vacuum generator, negative pressure in a pipe and secure retention of a workpiece are different facts. A valve command does not prove cup contact. A switch output proves only that pressure at its port satisfied its switching condition. It cannot by itself establish friction, workpiece presence, load sharing among cups, resistance to acceleration or the absence of peel-off. SMC’s vacuum-equipment selection guide shows checking a suction verification signal before lifting and setting vacuum pressure with the required lifting condition in mind [1]. Treat “suction OK” as an engineered permission to move, supported by workpiece trials, rather than a device lamp.
Our article on vacuum-pump condition monitoring concerns the health of the source. This guide concerns the individual pick and the pressure observed near the workpiece. A healthy pump cannot compensate for a torn cup lip, a hole in a part or a disconnected branch.
Define what passes at each stage
Divide the requirement into four events: attaining the required pressure within the allowed time after suction starts; maintaining it immediately before lift; detecting loss during acceleration, posture changes and transfer; and confirming release at placement. A single snapshot of the switch at lift cannot test all four. Record traces from good, borderline and deliberate failure samples across accepted product variants. If those traces overlap, redesign cup contact, circuit zoning or a separate presence check before relaxing the setpoint. A new and a worn cup, a clean and an oily part, and slow and maximum-speed moves can yield different pressure histories. Identify the accepted combinations rather than approving one successful demonstration. Where a single pressure criterion cannot separate good and bad picks, document which extra physical or sensing measure closes the gap.
Sensor port and location determine which faults are observable
Possible points include the generator outlet, either side of a filter, a common manifold, a cup branch and the cup itself. A sensor at the source may see adequate vacuum while a long hose, blocked filter, loose fitting or leaking branch prevents adequate vacuum at the cup. Schmalz describes mounting a vacuum switch directly at the cup in decentralized systems and monitoring leakage or part control through its outputs [2]. Festo warns that when a sensor sits between a filter and a Venturi generator, it can switch even though a dirty filter leaves insufficient vacuum at the cups [3]. Draw the sensor port on the pneumatic diagram and state whether it serves source diagnostics, circuit monitoring or individual-pick confirmation.

When one common-manifold measurement works, and what it misses
One common-manifold switch can be reasonable with a single cup or a circuit where all cups contact the same part in comparable conditions. It cannot automatically prove that every cup sealed. Check valves and flow restrictions can isolate an uncovered cup and preserve system vacuum, as Schmalz explains [4]; that useful behavior can also make a local failure less obvious at the manifold. Choose branch switches or an independent part sensor where the failure consequence justifies the extra wiring and upkeep. Test a partially seated cup at the planned maximum motion, not only a perfectly sealed static part. The decision is a trade-off in fault observability, replacement access and consequences of a drop, not a rule that more sensors are always better. On a large part, one failed cup may shift the load and create peel forces even while common pressure remains acceptable.
Specify the port, tubing, filter and environment
Compare candidate devices for measuring and proof pressure ranges, port thread, tubing, allowable fluid, supply voltage, response and filtering, NPN/PNP or communication output, display, enclosure and ambient limits. A blow-off pulse can create positive pressure, so check transient pressure against the selected model’s limits. Long narrow sensing tubes add volume and can change the apparent response. Dust, oil and moisture call for a serviceable filter and a plan to diagnose its blockage. At a moving robot tool, include cable flexing, connector restraint, collision protection and replacement access. In the cabinet, recheck the pressure delay created by the tubing to the cup. Record actual fitting and tube lengths in the test configuration: a convenient bench connection is not evidence for the installed robot. A filter may protect the switch but also changes evacuation time as it loads; the maintenance plan needs a measurable inspection trigger and a replacement method.
Setpoint and hysteresis must come from workpiece measurements
A catalog range says what the switch can be set to, not what should be accepted in your cell. Begin with the required holding condition and verify it with actual workpieces. The simplified relation of pressure difference times effective area does not capture all acceleration, shear friction, lip deformation, peel forces or uneven cup loading. SMC recommends calculating the required vacuum and then optimizing the suction confirmation value [1]. Do not copy that calculation directly into a pass/fail setpoint.
Collect pressure-time data for normal and difficult parts: clean and contaminated, flat and warped, new and worn cups, different lots and robot postures. Determine whether one recipe covers all accepted parts or approved part-specific settings are needed. Lowering the threshold can reduce nuisance stops but may reduce the margin before lift. Raising it can reject safe parts. The choice needs evidence of both good operation and deliberate misses.
Hysteresis is the distance between switch-on and switch-off points [6]. It prevents repeated transitions when pressure fluctuates around a threshold. SMC’s ZK2 documentation says hysteresis should exceed the pressure fluctuation width near the set value to avoid chattering [5]. Yet an excessively wide band may delay the loss signal. Record ON pressure, OFF pressure, filter time, PLC scan and response to a pressure drop together. State whether the numbers are gauge pressure, absolute pressure or a positive “vacuum level”; mixed sign conventions are a common source of commissioning errors.

Build an application-specific acceptance table, not a copied number
For each product recipe, document the part variant, cup condition, measurement port, source condition, ON and OFF values, maximum time to confirmation, holding criterion during travel and release criterion. No universal kPa or millisecond number applies to every cell. If the application needs pressure trends, an analog or IO-Link device may be useful, but include network failure behavior, compatible hardware, spare parts and maintenance capability in the estimate. A richer pressure record does not substitute for a sound acceptance rule. With a digital contact, log its ON/OFF times against valve commands and robot state. Check that the value shown on the device and any PLC conversion use the same units and zero reference. If the PLC analog value differs from the switch display, verify scaling and calibration before moving the threshold to hide that discrepancy.
Cycle timeout, leakage and no-pick are separate diagnoses
Time to switch-on depends on cup seating, hose volume, vacuum-source capacity, leakage and workpiece position. Specify whether the timer begins with the vacuum valve command or the robot’s arrival at the contact point. Set the limit from the slowest accepted good cases and observed equipment variation, not from an average cycle. A no-switch event can mean no workpiece, missed contact, a large leak, a disconnected hose or weak source vacuum. Pressure alone may not distinguish them.
Use independent part-present sensing at the feed point where needed. A signal that rises and then falls during travel suggests a different fault from one that never rises. Gradually increasing evacuation time may point to filter loading, cup wear or a growing leak; this is a diagnostic hypothesis, not a conclusive diagnosis from one trace. Log valve command, robot position, source supply, branch pressure or flow as the risk and cost warrant. If double-picking matters, thickness, weight or optical detection may be necessary because a pressure switch cannot prove that only one part was lifted.
Add signals to distinguish a leak from a missed pick
Limit retries and define their sequence: release, return to a safe position, confirm a part is present, retry within the approved count, then route the failed part or stop. Endless retries conceal production loss and can cause a double pick. After an emergency stop, determine whether the tool still holds a part, is empty or retains vacuum before restart. A blocked hose or closed valve can hold pressure without a secured part; porous material may be held despite a lower pressure. Test the relationship between the measured signal and physical retention with actual samples. A one-bit pressure signal may report the same timeout for no part, a hole in the part, a disconnected hose and insufficient source capacity. If the distinction changes the recovery action, log part-present, robot position, source pressure, branch pressure or flow, and valve command or feedback on a shared timeline. Keep a list of workpiece conditions that the pressure rule misclassifies and decide whether they require a design change or an explicit exclusion.
Integrate the switch into PLC and robot state transitions
A typical sequence is position → vacuum on → wait for suction confirmation → permit lift → monitor transfer → release → verify release. Assign ownership of every transition to the PLC, robot or vacuum controller so two controllers do not advance on inconsistent states. Compare timestamps of suction OK and robot motion permission in FAT. Verify sensor polarity, PNP/NPN compatibility, supply voltage, input behavior on cable break and loss of sensor power against the exact model’s wiring diagram. SMC’s wiring FAQ explicitly directs users to model-specific manuals [7].

Define separate events for timeout before lift, pressure loss during travel, open circuit, absent feed part, failed release and residual workpiece after an emergency stop. For each, specify robot position, valve state, operator message, recovery authority and test evidence. A stop command alone does not necessarily prevent a part from falling; a support tray or mechanical restraint may be needed. An ordinary suction OK contact is not a substitute for a machine safety function. Determine required safety measures through the machine’s risk assessment.
Decide fault-state transitions before commissioning
Control changes to setpoints. For a recipe-based cell, do not allow a pick when part number and active recipe disagree. Back up the PLC, robot program and switch settings under compatible revisions. If operators can change the switch panel, define access and approval. Maintenance personnel should be able to identify which setting was used for a failed pick. A screen that only says “vacuum fault” is inadequate at a stopped robot: the operator needs to know whether the fault occurred at pick, during travel or at release, and whether a part may still be held. Specify safe recovery, who can acknowledge the event, and how to restart without an unintended second pick. Verify sensor-power loss and cable disconnection as separate cases, including the value the PLC actually sees. The route over people or valuable equipment may require a catch tray or mechanical support; the chosen response follows a machine-level risk assessment.
Write an RFP around the required decision, not a component name
“With vacuum pressure switch” leaves the supplier free to choose different locations, outputs, wiring, thresholds and tests. Provide workpiece IDs, mass, center of gravity, available contact zones, holes, oil, dust, warpage, feed variation, number of cups, transfer posture, acceleration, target cycle and consequence of a dropped part. See our guides on vacuum cup selection and vacuum ejector selection for the contact and source decisions that precede this switch specification.
Ask bidders to mark the measurement port on a pneumatic diagram and identify faults visible and invisible from that point. Include pressure and transient limits, response, output and PLC I/O, connector, filter, local spares, parameter access and replacement procedure. Specify ON/OFF points, hysteresis, filtering, timeout, travel monitoring, release confirmation, alarms, retry limit and event log. Where numerical values remain open, state that FAT with real workpieces will set approved values, and identify who signs them off. Request a state chart, wiring drawing, test plan and spare-parts list.
Include the hidden costs in bid comparison
Compare full installed cost: switch, fittings, branch tubes, filters, cable, I/O, network master if required, programming, trial fixtures, FAT/SAT, training and replacement stock. One common switch and several branch switches differ in troubleshooting time as well as purchase price. The economically sensible choice depends on which fault must be found and the consequence of a stop or dropped workpiece. Use the same table to record items excluded by each bidder. The nearest-to-cup location is not automatically best if tool mass, collision exposure and flexing cable increase substantially; ask bidders to compare observable and unobservable faults and service access at each proposed port.
FAT and SAT must use the same acceptance logic
Prepare identified samples spanning accepted dimensions, roughness, warpage, holes, oil, dust and lot variation. Photograph contact surfaces and the intended cup positions. An unplugged hose simulates a large leak but may not mimic a partially lifted lip; describe what each injected fault represents. Test wiring first, then static pick, ordinary transfer, maximum acceleration and posture change, fault injection and recovery. Freeze sensor settings, robot/PLC versions and source conditions for each run. If a value changes, assign a new revision and repeat affected tests rather than mixing results. Include a temporary blocked filter, a loosened branch, sensor power loss and cable disconnection under controlled, guarded conditions. Test the longest route, dwell, planned maximum speed, release and the retained-workpiece state after emergency stop. The display agreeing with a PLC input is only an electrical check, not the end of FAT.
Record pressure curves or switch transition times together with valve commands, PLC inputs, robot position, sample ID, cycle time, retries, slips, drops, double picks and failed releases. Capture failures as well as successes. Logging must be fast enough to resolve the fault being tested; if the production PLC log is too slow, add a synchronized test logger and state what evidence remains available in production. Demonstrate switch replacement: port sealing, connector orientation, settings restore, I/O check, one good sample and one simulated failure. Provide operating messages and maintenance instructions in the language the local team uses.
FAT checks the built cell under planned conditions. SAT repeats the same samples, definitions and pass criteria after installation in Thailand, where routing, supply, electrical connections, feeder alignment and motion settings may have changed. Re-measure evacuation time and retention if a hose, filter, cup, robot acceleration or workpiece changes. “It ran without a problem” is not a measurable acceptance criterion. Specify samples, repetitions, allowable failures, measuring method, instruments, conditions and decision owner for the project. Where a wrong decision affects personnel safety, perform a separate machine risk assessment and dedicated safeguards.
Write acceptance criteria around state and evidence
For each named sample and motion condition, require confirmation by the agreed deadline, the holding condition throughout transfer, a defined transition on an empty pick, leak or disconnection, and evidence that a released part no longer remains on the tool. Separate three questions in the acceptance sheet: was a part at the feeder, did a cup seal, and could the part be carried through the motion? A feed sensor helps with the first but does not prove the cup seated; a pressure switch helps with the second but does not establish shear retention; a physical transfer test proves an outcome but alone does not explain a recurring failure. Use the combination to authorize the next motion.
A practical fault table can say: part-present ON plus pressure timeout prompts inspection of cup contact and part position; part-present OFF points first to the feeder. Pressure achieved before lift but lost at robot rotation prompts investigation of peel, hose pull and inertia; gradual loss while stationary prompts inspection of fittings, lip, part surface and valve behavior. These are triage hypotheses, not definitive diagnoses without photos, logs and repeat tests. Operator messages should identify the stage: “pick pressure not reached,” “pressure lost in transfer,” or “release not confirmed.” Align units, signs and decimal places across the switch, PLC screen and test report.
After start-up, monitor the distribution of time-to-confirmation, not just a success percentage. A gradual increase can motivate inspection of filters, cups and hoses even before a hard timeout, but a change in part mix or motion recipe can shift the distribution too. Keep product ID and setting revision with the record; evaluate false alarms and missed faults across lots and shifts. Maintenance alerts and the per-cycle “suction OK” permission serve different purposes and need not share a threshold. Demonstrate replacement of an identical switch: check model, output polarity, default filtering and parameters; force both PLC input states while stopped; then run one good part and one deliberate empty pick to show permit and inhibit paths. Include this in the handover procedure.
Hand over model numbers, port photos, pneumatic and I/O drawings, setpoints, software versions, raw FAT/SAT data, replacement steps and the conditions that trigger a retest. A replacement cup material or altered workpiece surface can change holding behavior even when the switch value is untouched.
Frequently asked questions
Can one vacuum switch confirm every cup in a multi-cup gripper?
It can monitor common pressure, but it does not prove each cup is sealed. Branch isolation and leakage may hide a local loss. Use branch sensing, independent part detection and retention trials according to the consequence of failure.
What pressure and hysteresis should I set?
There is no universal kPa value. Estimate holding requirements, measure good, borderline and failed parts at the actual port, then approve ON/OFF thresholds, filtering and timeout together. A catalog’s programmable range is not an application recommendation.
Is a PLC timeout enough to detect leakage?
It detects failure to reach pressure by a deadline. Leakage after pick requires monitoring during travel, with different actions for pre-lift and in-motion faults.
Can a Thailand SAT be omitted after FAT passes?
Installation can change tubing, supply, feed and robot motion. Repeat the agreed acceptance conditions on site, document differences and retest after relevant changes.
Conclusion
Specify what state must be observed, where it is measured and when it must be true before choosing a vacuum pressure switch. Link the port location, measured setpoints and hysteresis, timeout, in-motion monitoring, release check and PLC fault response. A shared RFP and the same evidence at FAT and SAT give the “suction OK” signal a precise meaning.
If you are defining a robotic suction cell for a Thailand plant, contact TOMAS TECH while the sensor position and PLC acceptance rules are still being scoped. We can help organize the workpiece, cup and pneumatic conditions that need to be tested.
References
- SMC, Vacuum Equipment Model Selection: https://www.smcworld.com/catalog/BEST-Guide-en/pdf/6-4-p0002-0048-sinku_en.pdf
- Schmalz, Vacuum Switch VSi-HD: https://www.schmalz.com/en-nl/products/automation-743270/switches-and-system-monitoring-308804/electronic-vacuum-and-pressure-switches-308827/vacuum-switches-vsi-hd-385894
- Festo, Vacuum suction cups and vacuum monitoring: https://www.festo.com/ie/en/c/products/vacuum-technologies/vacuum-suction-cups-id_pim113
- Schmalz, Check Valves, Flow Restrictors and Sensing Valves: https://www.schmalz.com/en-us/support/know-how/vacuum-knowledge/the-vacuum-system-and-its-components/valve-technology/check-valves-flow-resistors-and-sensing-valves
- SMC, Series ZK2 Vacuum Unit catalog: https://www.smcworld.com/catalog/en/vacuum/ZK2-E/6-4-p0055-0100-zk2_en/data/6-4-p0055-0100-zk2_en.pdf
- OMRON, Pressure Sensor FAQ on hysteresis: https://www.ia.omron.com/support/faq/answer/35/faq01443/
- SMC, ZK2 series pressure switch wiring FAQ: https://www.smcworld.com/support/faq/en/s.do?ca_id=929&id=2513&lang=en