Choosing a fixed RFID reader and antenna for a Thai factory by the catalogue’s maximum read range can produce two failures at once: missed items in the portal and stray reads from the adjacent storage area. The real requirement is a controlled read zone that creates a reliable business event only when the intended object crosses the intended point. This guide follows procurement from process definition and site survey through placement, RFP, factory acceptance testing (FAT) and site acceptance testing (SAT). It focuses on fixed UHF read points at shipping portals and workstations; see our separate guides for tag and total project cost decisions.
Define what “passed the gate” means to the business
A tag detected beside a gate is not proof that a shipment left the plant. A forklift can wait at the entrance, a tagged pallet can sit next to the portal, or a load can turn back. The buyer should write an event contract: which object, crossing which boundary, in which direction, changes which system state? One example is: “When a returnable container assigned to a shipment passes outward through shipping gate A and its contents match the dispatch order, submit a shipment-confirmation candidate to the WMS.” A candidate still needs the defined validation before it becomes final.
GS1 EPCIS treats an observation within the context of what, when, where, why and, where relevant, how. An EPC read is one input to that event, not the event itself. Retain the read point, antenna, direction evidence, timestamp, order, decision version and exception code. Combine repeated observations of the same crossing into one event, while keeping an adjacent-zone read out of the shipment record. A contract that specifies only “read rate” misses this distinction.
Decide whether the unit of identity is an item, case, tote or pallet before selecting the read point. Each implies a different tag population, stacking pattern, dwell time and reconciliation rule. For the choice between RFID and visual codes, see traceability code selection. For overall costs, see RFID implementation cost. The present article starts after a candidate identification method exists.
A portal and a workstation need different antenna designs
A shipping portal must identify multiple tags across a wide path in a limited time. Side, overhead or other antenna positions may be needed, but additional antennas also add cable runs, port scheduling, interference and possible spillover to adjacent gates. Load height, tag position and vehicle speed determine the required illumination pattern.
A workstation often needs the opposite: identify the one fixture or tote deliberately placed in a small area. A proximity antenna, narrower beam, physical positioning and, where appropriate, shielding may control the zone better than a long-range antenna. Impinj lists proximity and broader coverage antennas for distinct purposes. Neither a catalogue range nor a reader’s maximum rate can replace testing with the actual fixture, workpiece, operator position and nearby metal.
In both cases the aim is to give target tags enough opportunities to respond while preventing non-target reads from becoming business events. Increasing transmit power can improve the first measure and worsen the second. Evaluate transmit power, antenna gain and pattern, cable loss, receive sensitivity, reader mode and event logic as a system.
What the site survey must measure
Put the gate, workstation, storage racks and neighboring read points on a scaled layout. Record load dimensions, stacking, speed, waiting places, worker movement and peak-period traffic. Mark steel fences, doors, trolleys, conveyors, liquid containers, motors, electrical cabinets and other radio equipment. An empty-floor demonstration will not reveal cross-reads when a full shift is working.
Take a representative sample set, not the easiest carton. Include minimum and maximum loads, rotated tags, metal and liquid products, empty and full containers, wet or protected labels and people standing between antenna and tag. Metal can reflect radio energy and detune a tag; liquid and human tissue can absorb energy. RAIN Alliance field guidance stresses the need to design for these actual materials and human interactions.
The survey record must also show proposed antenna height and angle, real cable route and length, power and network availability, cleaning conditions, maintenance access and forklift impact risk. Zebra’s technical tutorial shows how cable-loss compensation is configured on an FXR90, illustrating why cable routing belongs in the RF design. Two positions that look equally close on a plan can have different cable loss after routing around a machine. Confirm the route before requesting a firm quotation.
Contain the read zone in four layers
First, position the antennas and the goods. A portal antenna aimed down the travel path may illuminate a waiting pallet or downstream rack. Angle and locate antennas toward the intended crossing, increase separation from background inventory, and map reflections beyond the gate. Shielding can help, but test whether it hides tags behind a load.
Second, configure RF settings per antenna: power, active ports, inventory mode, session and receiver settings as supported by the equipment. Impinj’s portal example addresses many tags passing quickly and interactions with nearby readers. Record the device model, firmware, configuration and results together. An aggressive setting can increase stray reads and event latency.
Third, use a process trigger such as a photoelectric sensor, conveyor PLC or door signal to open a decision window only while a load is passing. For bidirectional movement, use two triggers or a trustworthy conveyor direction signal. Validate any direction inference from RF reads under reflection and overlapping loads. Decide whether a failed trigger causes a stop or a held exception; it should not silently create a shipment event.
Fourth, convert raw reads into events. Deduplicate within a time window, reconcile EPCs against the expected shipment, compare neighboring antenna observations and use direction and trigger evidence. RSSI can assist diagnosis but does not prove physical location. Route uncertain cases to an exception queue. Zebra describes purpose-built portal antennas designed to reduce cross-talk and erroneous reads; onsite event rules still need testing even with an integrated product.
Choose antenna shape before choosing the largest gain
Define the physical zone first: portal width, height and nearby stock; or a fixture’s position and the neighboring workstation. Then test tag orientation. Predictably oriented tags may favor a linear-polarized antenna; uncertain orientations may justify testing circular polarization. Neither option guarantees a read through metal or liquid. Compare the worst real load configuration.
Compare beam width, gain, polarization, supported band, mounting and environmental protection. A higher-gain antenna concentrates energy in its intended direction; it does not guarantee that a nearby unwanted tag cannot respond. Document the trade-off between covering every stacked tag and reducing leakage into a neighboring aisle. Include mounting clearances and metal-surface effects. Catalogue range is not a contractual factory result.
A reader decision covers more than the number of ports: permitted region configuration, fine power control, receive sensitivity, multi-reader behavior, I/O, network, event API, logs, replacements and Thai support. Raising receive sensitivity may recover weak target responses and also accept more distant tags. A headline reads-per-second figure says little about a moving, crowded load when antennas share time. Manufacturer specifications are inputs to a trial, not a promise of site performance.
Treat coaxial cable, connectors and any splitters as part of the RF design. Cable length and loss change both delivered power and the return signal. Label each port physically and in the configuration backup so a replacement is reconnected correctly. Weigh a shorter cable against accessibility for safe maintenance; do not optimize one number in isolation.
Make Thailand radio compliance a purchase condition
GS1 EPC UHF Gen2 describes a widely used air interface. Legal frequency, transmission conditions, device approvals and installation conditions remain country-specific. A reader previously used in Japan or elsewhere should not be assumed valid in Thailand. Ask the supplier for the current NBTC primary rules and evidence for the exact reader, antenna and cable configuration, and have the buyer verify them before purchase. A selectable “Thailand” setting alone does not demonstrate the entire installation is compliant.
The RFP should demand the applicable current primary provision, product approval evidence, allowed settings, a compliance calculation that accounts for antenna gain and cable loss, and ownership of changes. Do not import a power or frequency figure from an old blog into a legal conclusion for this project. Specify how a later antenna or cable replacement is checked, and maintain separate configurations if the system will expand to other countries.
Write an RFP with reproducible acceptance conditions
Start with a one-page process drawing. State whether gate A reads a tote moving outward or bench B reads a seated fixture. Attach the tag identity scheme, allowed placement variation, maximum concurrent population, speed, waiting area, adjacent tags and operator location. Provide a sample register that includes the hardest cases.
Require a read-zone drawing, antenna height/angle/port list, cable schedule, reader settings, trigger wiring, event contract, exception workflow, logs, recovery procedure and compliance evidence. Separate target detection, false business events, direction correctness, event latency, data protection during an outage and manual exception handling. Define denominators and repeated trials. “Read rate 99%” without a load mix or unit of measurement is not an acceptance test.
For example, 100 successful crossings of load type A do not prove success with a full liquid load or a metal trolley. Stratify trials by worst-case material, orientation, speed and neighboring inventory. Negotiate acceptance thresholds against the plant’s actual loss risk rather than treating an example in an article as an industry guarantee. Specify what happens after a miss: rescan, manual reconciliation, hold or line stop.
Compare bids on how each vendor resolves site conditions. Ask who makes the RF map, whether post-installation angle adjustments are included, how FAT and SAT differ, and what a future adjacent portal costs to retune. An inexpensive reader line item can conceal survey, integration and exception work as “extra.” Our RFID budget and PoC guide covers the overall investment decision.
Run a one-zone PoC before choosing the final layout
Use the same tags, loads and movement patterns for each candidate. Make a ground-truth log of expected EPCs, crossing time, direction and non-target tags using an independent reliable method. Hold antenna position constant while varying one setting at a time: power, receiver setting, antenna type or polarization. Otherwise the cause of improvement is unknown.
After a candidate passes in isolation, run neighboring gates at peak traffic. Include simultaneous loads, waiting, reversing, empty pallets, rotated tags and a failed trigger. Repeat after the surrounding metal or liquid layout changes. Save model and firmware, settings file, antenna and cable part numbers, sample photos and test dates for repeatability.
Report correct shipment events, misses, wrong-gate assignments, rack-related false events, wrong direction and decision time with counts and denominators. Classify failures as RF coverage, tag orientation, trigger, event rule or upper-system integration. “Read from five metres once” is a demo result, not an operating criterion.
Separate FAT from SAT
At FAT, use the delivered reader, antennas, cables, simulated trigger and test interface to verify deduplication, direction logic, event schema, retry after a network outage, logs, permissions and backups. A test rig cannot reproduce the reflections from the Thai plant’s actual steel racks and conveyors. Passing FAT does not establish onsite RF acceptance.
At SAT, check mounted angle and strength, cable loss, neighboring equipment, actual loads and the process at operating speed. Include different shift patterns where they change traffic. Place target and non-target tags simultaneously and verify that only correct events reach WMS or MES. Have operators demonstrate a missed-tote procedure and a held false-read candidate. Keep the final onsite configuration and the delta from FAT.
Each acceptance record needs test ID, load, tag orientation, target and neighboring population, speed, trigger state, expected and actual result, log link, pass/fail and retest date. “Adjusted” is not closure without a changed-setting record and a repeated test. When a new metal rack appears beside the portal later, use change control to repeat the relevant SAT cases.
Example: shipping gate A and fixture bench B on one site
Suppose a tagged-tote rack sits beside gate A. A forklift carries two-high loads and sometimes waits before the line. Bench B identifies a tagged metal fixture when an operator seats it. Putting the same high-power four-port reader at both positions ignores their different objectives.
At gate A, map where tags can appear on the load, add antennas only as evidence requires, and correlate entrance and exit trigger signals with the dispatch list. An event can wait until the exit trigger confirms movement; a load that backs out produces an observation but no dispatch candidate. At bench B, try a proximity antenna beneath a fixed fixture position. Test with the complete metal fixture, neighboring benches, a spare fixture in an operator’s hand and stored items under the bench. A seating signal or start button can guard the event and prevent repeats while the same fixture stays in place.
Run both points together. Settings that pass individually can interact when readers operate simultaneously. Record isolated and simultaneous results separately. Reader scheduling and session settings depend on the equipment and tag population, so resolve them with a system trial rather than a generic setting copied from another facility.
Agree on denominators and failure handling
“99% read rate” can mean identified tags divided by tag opportunities, or complete shipment events divided by crossings. A pallet with 20 cases passing 100 times creates 2,000 tag opportunities but 100 shipment events. One crossing missing one case may look small in the tag percentage while still making that shipment incomplete. Report both measures.
Measure raw non-target observations separately from incorrect confirmed business events. A small amount of stray raw reads may remain in a real factory; the decision layers must keep them from corrupting shipment or production history. Track wrong-gate attribution, duplicate confirmation, wrong direction and unexpected EPCs separately. Define who can clear an exception and on what evidence.
Also time the decision: portal exit to WMS candidate, or fixture seating to work permission. Look beyond the mean to slow cases and recovery after a network outage. Specify whether events are buffered locally, whether work stops, and how duplicate replay is prevented. These are acceptance conditions for the complete system, not solely the RF hardware.
Maintain the zone after go-live
The RF boundary changes with racks, tag supply lots, packaging, equipment replacement and cable aging. Monitor passage counts, exceptions, non-target detections, communication gaps, antenna faults and clock synchronization per read point. Set thresholds from the operating baseline and show the denominator so lower throughput is not mistaken for poor recognition.
Separate retained business events from short-term diagnostic raw reads, with defined retention and access. Trace a business event back to read point, antenna, configuration version and source observations. If using EPCIS, distinguish the physical read point from business location and event time from record time. Give retried events a stable unique key so an outage does not create a second shipment.
A support agreement should cover a local spare, configuration restore, RF retest after replacing cable or antenna, and regression tests after firmware updates. Train staff in exception authorization, not just reader screens. Automatic reading cannot compensate for an operator approving the wrong exception.
Handover conditions for future changes
A settings file alone will not reproduce SAT six months after equipment moves. Handover should include photos, mounting height and angle, brackets, port map, coaxial type and length, read-zone drawing, transmit/receive settings, trigger semantics, event rules and firmware version. Photograph the representative load and non-target tag placement used in acceptance. They form a baseline for comparing later changes.
A small antenna-angle adjustment can move both a blind spot and a spillover area. An equal-length replacement cable need not have the same loss or connector quality. Maintenance should record the reason and affected point before the change, then run a short regression set with worst-case load, neighboring zone and reverse-direction movement. Agree with the vendor which changes may be exempt.
Recovery must work even when the RF specialist is absent at night. Restoring a backup should be followed by an event-level check with target and non-target tags together, not only a successful single-tag read. If restoration fails, switch to controlled manual records and reconcile duplicates and omissions later. Include this recovery drill in handover.
Pre-purchase checklist
Before approving the purchase, answer these questions in writing. Are real target and non-target samples available? Have waiting, reversal and simultaneous passage been tested? Were full loads with metal and liquid mapped? Are polarization and tag orientation recorded? Have neighboring portals run together? Are cable loss and post-replacement RF measurement in the quotation? Is the process response to a failed trigger defined? Have current NBTC primary rules and configuration-specific evidence been checked? Does the plan say what FAT and SAT accept separately?
Put unanswered points into an RFP action register with an owner and due date. State assumptions explicitly when a condition cannot yet be measured. Agree who pays for onsite read-zone adjustment before signing, so a missed read found at SAT leads to a clear diagnosis and correction. The success condition is not that one carton was read once, but that ordinary movements become correct events by the same rule every day.
Keep equipment and business-event acceptance separate. If a tag is read but shipment reconciliation fails, the RF test may pass while the event test fails. If the WMS works but a metal trolley shadows a tag, network success cannot excuse the RF miss. Assign an owner to each failure layer and rerun adjacent negative cases after a fix, especially when changing antenna angle or receiver sensitivity.
Frequently asked questions
How many fixed readers and antennas do I need?
Width and port count alone are insufficient. Measure tag orientation, load, speed, nearby stock and triggers at one read point. Compare a minimum arrangement against additional antennas for detection, spillover and maintenance. More antennas can improve coverage and increase false-event exposure or wiring work.
Does metal or liquid make fixed RFID unusable?
Not categorically. Reflection, absorption and detuning change results between an empty demo and a full production load. Test real samples, tag positions, antenna angles, shielding and a proximity zone where appropriate. A repeatable worst-case result matters more than one successful demonstration.
Should we simply lower power when the next gate reads our tags?
Lower power can also miss target tags. Check aim, beam, shielding, spacing, trigger and event logic. SAT should run adjacent gates simultaneously and distinguish stray raw observations from false confirmed events.
Can SAT be skipped after FAT passes?
No. FAT checks logic and connectivity; SAT tests the actual reflections, obstacles, people, traffic and neighboring readers after installation. Hand over the onsite adjusted version as the operating baseline.
Can an overseas reader be reused in Thailand?
Check the exact model, region configuration, antenna, cable and current local requirements first. Ask for current NBTC primary evidence and perform the onsite survey and SAT. Previous operation in another country proves neither local compliance nor local read performance.
Conclusion: procure a trustworthy crossing event
Fixed RFID reader selection starts with a business event definition. Map metal, liquid, people and adjacent tags on site, then contain the zone through placement, RF configuration, triggers and event rules. Put reproducible tests and exceptions in the RFP; validate logic at FAT and real RF behavior at SAT. This turns equipment specifications into a history the factory can trust.
If you are still evaluating a portal or workstation, contact TOMAS TECH. We can help scope a one-zone PoC and turn your layout into specific RFP and FAT/SAT checks.
