Reader Setup Recipes
Part of: Devices — starting-point settings for common installations
A reader configuration and the scan event processing behind it have to agree on what a "pass" is. This guide gives starting points for the installations we see most, with the reader settings and the Tagnet processing settings side by side. Every value here is a place to start, not a rule: tag type, packaging, speed, and the RF environment all move the right answer, so expect to adjust after the first few real passes.
The recipes cover Zebra FX readers (FX7500, FX9600 and FXR90) today. The Impinj R700 exposes the same ideas under different names (an RF mode instead of an operating environment, a preset instead of an operating mode), and recipes for it will follow.
Where the settings live
| Setting group | Where to set it | Guide |
|---|---|---|
| Operating mode, environment, session control, antenna stop, delay between cycles, report filter, RSSI filter | Devices, open the reader, Edit, Reader configuration | Zebra FX Configuration |
| Processing mode, quiet period, max event duration, cooldown and scope, motion sensitivity | Device Groups, open the group, Edit, Scan event processing (or the per-device overrides on the reader's edit page) | Device Groups, Scan Events |
A grouped reader takes its processing settings from the group. Put every reader that covers one physical read point in one group so the group's window is the pass, and set the processing values there.
Tune one reader first. Once it reads the way you want, copy its settings to the others in the portal from the local wizard's Starting point step, the edit page's Copy settings from... button, or Create similar reader on its detail page; see Copying Settings from Another Reader.
Drive-through portal
Vehicles carry many tagged items past a fixed portal without stopping: a truck of lumber packs, a trailer of pallets, a forklift through a yard gate. The reader has one chance per vehicle, the tag population per pass is large, and tags in the middle of a load are shadowed by the ones around them.
The goal is to read every tag once. Loud tags should go quiet after their first read so the shadowed ones get air time, and the reader should never idle while the load is in view.
Reader configuration
| Setting | Value | Why |
|---|---|---|
| Operating mode | Custom | Environment, delay between cycles, and report filter are only editable here |
| Operating environment | High interference (the default) or Auto detect to start; Low interference if the load is many strong tags and you are chasing read rate | The environment picks the link profile: the slower profiles hear weaker tags, the faster ones finish more rounds. At portal speeds sensitivity usually matters more than rate; see what the environment changes |
| Session and target | S2 with target A for a dense, slow-moving load; S1 with target AB for a fast pass with tens of tags | Two different bets; see Session choice for a moving load below |
| Tag population | The number of tags on a typical full load | Sizes the first inventory round for the load instead of for one tag |
| Delay between antenna cycles | Disabled | The default idles the radio for 75 ms after any cycle with nothing new, which is most cycles once a load is in view |
| Antenna stop condition | Leave unset | The default gives each antenna one inventory round per cycle, which suits a moving target |
| Report filter | 1 second, per antenna (the default) | One summary per tag per antenna per second is enough to count and to rank antennas; set 0 while diagnosing to get every read |
| RSSI filter (FX9600, FXR90) | Off | Shadowed tags are weak; filter only once you know the read zone catches stray reads from a neighboring lane |
Scan event processing (on the device group)
| Setting | Value | Why |
|---|---|---|
| Processing mode | Multi-tag passage | One scan event per vehicle, all tags together |
| Quiet period | 1 minute | Long enough to bridge a gap between the front and back of a load, or a vehicle that slows, without splitting the pass |
| Max event duration | 3 minutes | Caps a window that a stopped vehicle would otherwise hold open |
| Tag cooldown | 5 minutes, group scope | A tag that is read again during the cooldown, for example a vehicle backing up, does not start a second event |
| Motion sensitivity | Default | Motion detection needs per-read phase, which the 1 second report filter does not supply; it is not driving anything in this recipe |
These processing values suit a portal where vehicles clear the read zone without stopping. Two things change them:
- Vehicles that stop or queue near the portal. A vehicle that pauses in the read zone keeps the window open until max event duration closes it, and the rest of the pass lands in a second event. Raise max event duration so the whole dwell fits, and keep the quiet period above the longest pause you see mid-pass.
- Vehicles that come back through soon after. A load that re-enters within the cooldown is suppressed, by design. If vehicles legitimately return within five minutes, shorten the cooldown; it should be longer than one pass and shorter than the gap between two passes of the same load.
- Stragglers. With S2 and target A, a tag answers once and then stays silent for as long as it is powered. It is read again only after it drops out of the field for a moment, which for the weakest tags on a load can be well after the rest were read. The quiet period has to be longer than that gap, or the stragglers close as a separate event with one or two tags in it. One minute covers it for a pass; a longer trigger run needs a longer quiet period.
Session choice for a moving load
The session decides what happens to a tag after the reader has read it, and on a moving load that is the whole game. There are two honest bets, and which one wins depends on how fast the load passes and how many tags it carries.
Read once: S2 with target A. A tag answers once, flips its S2 flag, and stays silent for as long as it is powered, which on a pass means until the vehicle has left. Loud tags on the outside of the load drop out after one read, and the shadowed tags in the middle get the air. This is the right bet for a dense load that moves slowly enough for the reader to work through it: a trailer of pallets creeping past, a forklift pausing at a gate.
It has a failure mode worth knowing. A tag flips its flag when the reader acknowledges it, not when the reader has successfully decoded its EPC. If the tag's reply was too weak or collided and the reader never captured the EPC, the tag has still gone quiet, and with S2 it stays quiet for the rest of the pass. A tag read once at the edge of the zone, from a bad angle, in the one spot where it was audible, is lost even though it would have been loud a moment later in the sweet spot. The faster the pass, the fewer second chances, and the more that one lost reply costs.
Keep asking: S1 with target AB, and the tag population set. Target AB alternates between tags whose flag is A and tags whose flag is B, so every tag keeps answering round after round for as long as it is in the field. Nothing is lost to a single bad reply; a tag that was inaudible at the edge is read when it reaches the sweet spot. The cost is that the loud tags keep taking air time, so a large, dense load reads more slowly, and the report filter has to absorb the duplicates (the default one summary per tag per antenna per second does). S1 rather than S0 here because S1's persistence smooths over the brief power gaps as antennas switch; S0 with target A is a middle ground, re-reading a tag whenever it loses power for an instant, which on a four-antenna portal is often. Set the tag population to the load size either way, so the first rounds are not a burst of collisions.
Start from S1 with target AB for a fast pass carrying tens of tags, from S2 with target A for a slow pass carrying hundreds, and let the first real passes decide. The signature of each bet failing is visible in the scan event: with S2, tags that appear in one pass and not the next at similar RSSI; with S1 and AB, the weakest tags never appearing at all because the loud ones held the air.
Watch the Closed by field on the first few scan events. Quiet period is what you want to see; Max duration means the pass was cut short, and the second half of the load is in the next event. Keep max duration comfortably above the quiet period plus the longest read you expect: when the two timers land on the same second, the event reports Max duration even though nothing was cut.
Dock door with a trigger
A door where a photo eye, motion sensor, or door contact is wired to the reader's GPI. The reader should read only while something is passing, and each pass should be one event.
Reader configuration
| Setting | Value | Why |
|---|---|---|
| Operating mode | Portal | Reads start on the GPI trigger and stop after no new unique tags for the stop interval |
| Portal start trigger | The GPI port and signal level the sensor drives | See Portal Settings |
| Stop interval | 3 to 5 seconds | How long after the last new tag the radio stops |
| Session | S1 | Short persistence suits a pallet that may be read on more than one antenna as it turns |
| Target | A | |
| Tag population | Tags on a typical pallet |
S1 keeps re-reading a pallet that sits in the doorway, which is what keeps the scan event open until it leaves. If you switch to S2 for a crowded door, each tag reports once and the radio goes quiet while the pallet is still there, so raise the quiet period to cover the whole run; otherwise the event closes early and any tag read again later lands in a second event.
Outside Custom mode the report filter stays at one summary per tag per antenna per second, so phase and motion are not available in this recipe. If you need motion detection at a door, use Custom mode with a GPI radio start condition instead, and set the report filter to 0; see Radio Control Settings.
Scan event processing
| Setting | Value |
|---|---|
| Processing mode | Multi-tag passage |
| Quiet period | 10 seconds |
| Max event duration | 60 seconds |
| Tag cooldown | 30 seconds, group scope |
Conveyor or tunnel
Items pass one at a time, or in small groups, at a steady speed through a tunnel or past a belt-side antenna. Each item should produce its own event, and the same item should not produce two.
Reader configuration
| Setting | Value | Why |
|---|---|---|
| Operating mode | Custom | Conveyor mode also works, but the delay setting below is only editable in Custom |
| Operating environment | High interference if other readers or tunnels are nearby, otherwise Low interference | Tunnels are often installed in rows |
| Session | S0 | The flag resets the instant an item leaves the field, so the same item is read fresh if it comes around again |
| Target | A | |
| Tag population | 1 to 8 | Small, so each round is short and the reader keeps up with the belt |
| Delay between antenna cycles | Disabled | Items are brief; do not idle |
Scan event processing
| Setting | Value |
|---|---|
| Processing mode | Single-tag passage |
| Quiet period | 2 seconds |
| Max event duration | 10 seconds |
| Tag cooldown | 10 seconds, device scope, or none if items legitimately circulate |
Workstation or verification point
A bench where someone places an item, a tote, or a small batch in front of the reader to confirm it. Reads are deliberate, the population is small, and the same item is often presented again a moment later.
Reader configuration
| Setting | Value | Why |
|---|---|---|
| Operating mode | Simple | |
| Operating environment | High interference (the reader default) | Workstations are usually indoors near other readers |
| Session | S1, or leave at reader default | Tags re-read continuously while they sit on the bench |
| Target | A | |
| Alternative: S2 with target A | Each item is read once when it is placed and then goes silent until it is lifted off the bench and presented again. Use it when one report per presentation is what the workflow wants; use S1 when the reader should keep confirming what is on the bench. | |
| Tag population | 10 to 30 | A tote's worth |
| RSSI filter (FX9600, FXR90) | On, raised until tags on the next bench stop appearing | The read zone should end at the edge of the bench |
| Transmit power | Reduced, often 15 to 20 dBm | The cheapest way to shrink the zone; raise the RSSI filter second |
Scan event processing
| Setting | Value |
|---|---|
| Processing mode | Immediate for a scan-and-go bench, Single-tag passage with a 2 second quiet period when items dwell |
| Max event duration | 30 seconds |
| Tag cooldown | 5 to 30 seconds, device scope |
Cabinet or shelf
A fixed population in a confined space: a cabinet with a door, a rack, a set of shelves with an antenna per bay. Tagnet does not have a presence model for this today, so a cabinet or shelf is handled the way a portal is, as scan events, and the settings below are starting points for making those events useful rather than a finished design.
Settle three questions before choosing settings:
- What starts a read? A door contact or button on a GPI gives a clean read with a beginning and an end, and one scan event per audit. A reader that runs continuously has no natural event boundary, so Tagnet's window timers define one, and the result is a rolling series of events rather than audits.
- Read once, or keep answering? For an audit, each tag should be read once and then yield the air to the others: session S2 with target A. For a reader that runs continuously, tags must keep answering or nothing distinguishes an item that left from one that went quiet: session S1 with target A, or target AB on a dense population. Keep in mind that Tagnet does not yet turn a tag going quiet into a departure; that needs a last-seen model with a grace period for tags in hard-to-read spots, which is not built.
- How dense and how hostile? Many tags close together, on metal shelving or around liquids, shadow one another. Set the tag population to the full count, keep the delay between antenna cycles disabled during a timed read so the reader does not idle, and expect the last few tags to take longer than the first hundred.
Starting point for a triggered audit
| Setting | Value | Consideration |
|---|---|---|
| Operating mode | Custom | Needed for a GPI start condition and a fixed stop duration |
| Radio start condition | GPI with restart, the door contact's port and level | See Radio Control Settings |
| Radio stop condition | Duration, long enough to find the last tag | Start with a few seconds per antenna and lengthen it if the weakest tags are missing from audits |
| Session, target | S2, A | Read once; the rest of the cabinet gets the air |
| Tag population | The count when the cabinet is full | |
| Delay between antenna cycles | Disabled | The read is short; do not idle |
| Report filter | 1 second, per antenna (the default) | One report per tag per antenna, which bounds the event size by the antenna count |
| Processing mode | Multi-tag passage | One event per audit |
| Quiet period | Longer than the read duration | A tag read again late in the run must land in the same event |
| Max event duration | Comfortably above quiet period plus read duration | |
| Tag cooldown | None | Every audit should report the full contents |
A workflow that acts on the audit, such as comparing the event's tags to what the cabinet should hold or moving the items to the cabinet's location, runs once per trigger.
Starting point for a continuous read
| Setting | Value | Consideration |
|---|---|---|
| Operating mode | Inventory | One report per tag per interval, with peak RSSI and read count, which is what bounds the data volume |
| Reporting interval | 30 seconds to a few minutes | How often the picture refreshes; the session setting does not change how much reaches Tagnet, the interval does |
| Session, target | S1, A (the reader defaults) | Tags re-answer on their own every few seconds; use AB only if every tag must be confirmed every round |
| Tag population | The count on the shelves | |
| Delay between antenna cycles | No new tags, 100 to 200 ms (No tags on the FX7500) | Nothing changes between rounds most of the time; the idle leaves the spectrum to neighbors |
| Processing mode | Multi-tag passage | |
| Quiet period | Longer than the reporting interval | Otherwise every interval becomes its own event |
| Max event duration | Minutes to tens of minutes | Sets how often a new event, and any workflow on it, occurs |
| Tag cooldown | None |
Inventory mode never supplies phase, so motion detection reports Unknown for every tag. That is expected.
Two readers covering one area
Any of the recipes above, with a second reader whose field overlaps the first. Left alone, the two interfere, and a tag read by one may already be quiet for the other.
| Setting | Reader 1 | Reader 2 | Why |
|---|---|---|---|
| Operating environment | High interference | High interference | Selects the dense-reader link profile on both |
| Session | S2 | S3 | Separate flags, so one reader's reads do not silence tags for the other |
| Target | A | A | |
| Delay between antenna cycles | No new tags, 75 to 200 ms (No tags on the FX7500) | Same | Gives the other reader air time when nothing new is in view |
Put both readers in one device group so their reads merge into one scan event and the cooldown is shared.
Diagnosing a recipe
When a recipe under-reads, change one thing at a time and compare the same pass before and after. The useful views:
- Scan event detail, Tags tab: read count, dwell, and peak RSSI per tag. A tag with a handful of reads at low RSSI is shadowed, not missing.
- Scan event detail, Analytics tab: RSSI over time per tag for one pass.
- Device detail, Data Explorer: reads per antenna per hour. An antenna that reads a fraction of its neighbors is mis-aimed, cabled wrong, or at the wrong power.
- Report filter 0 for a day gives every read, with phase and channel, at the cost of volume. Set it back afterward.
Raw tag reads are kept for seven days by default; raise that retention in Policies before a diagnostic week if you will want to look back over it.