How-to
How to set up Pro-face historical trend channels in GP-Pro EX
Short answer
The channel selector reads undetermined because a historical trend pen can only select channels from a saved sampling group. In GP-Pro EX, open Common Settings, create the Sampling group with your RSLogix 5000 tags, then bind the trend pens to those logged channels. No PLC-side buffer list is needed.
On this page
- Why the channel list shows undetermined
- Confirm the tag reads live first
- Create the sampling group
- Choose the sample interval from the history you need
- Sizing the interval for 12 hours of four-channel history
- Bind the trend pens to the logged channels
- Verify before you trust the graph
- What breaks trends later
- Questions asked
The word "undetermined" in the channel selector is not an Allen-Bradley communications fault. A GP-Pro EX historical trend pen never reads a controller tag directly — it reads a channel of a saved sampling group, so with no group in the project there is nothing to select. Create the group under Common Settings, add the RSLogix 5000 tags to it, and the selector fills in. There is no separate "buffer list" to hunt for: the sampling group is the buffer, it lives in the HMI project, and the GP3000 display unit stores the records itself.
Why the channel list shows undetermined #
A historical trend draws timestamped records, and the chain has to be complete before the editor can resolve a channel: RSLogix 5000 tag → EtherNet/IP device in the project → sampling group → stored records → trend pen. GP-Pro EX puts the collection side under the Common Settings menu, in Sampling, and the Historical Trend Graph part displays what that function collected. Break the chain anywhere and the channel fields fall back to "undetermined".
That is why a live numeric display can work while the trend stays empty: live objects read the device directly, trends read stored records. That combination always points at logging configuration, never at communications.
| What you see | What it means | Where to fix it |
|---|---|---|
| Channel selector shows "undetermined" | No sampling group assigned, or the group holds no channels | Common Settings, Sampling |
| Live numeric display updates, trend stays empty | Communications are fine, logging is missing | Create the group, bind the pens |
| Tag cannot be selected anywhere | Device or tag reference problem | EtherNet/IP device setup and tag definition |
| Pens select, graph stays empty | Trigger not running, constant values, or wrong time span | Sampling mode and displayed range |
Confirm the tag reads live first #
Put a simple numeric display on one of the tags before touching logging. If it follows the controller, the EtherNet/IP path and the tag reference are good — stop editing communications, because the missing piece is the logger. If it does not update, fix the device and the tag first: a sampling group built on a broken reference stores nothing usable, and retransferring the project will not change that.
Create the sampling group #
- Open the project in GP-Pro EX and save a working copy before changing logging settings.
- Confirm the Allen-Bradley controller is configured as the project device over EtherNet/IP and that the tags you want resolve.
- Open the Common Settings menu and click Sampling.
- Add a sampling group and open its edit page.
- Add every value the trend must show, one channel per tag, with data types matching the controller values — a DINT channel reading a REAL tag skews the trace.
- Choose the sampling mode: a fixed interval for evenly spaced history, or a condition trigger when only defined events should be recorded.
- Set the storage destination. To write to external memory, open the System Settings for the display unit and, in the Mode tab, set Save Data to CF/SD card or USB storage.
- Apply the settings and transfer the project.
Choose the sample interval from the history you need #
The interval follows from retention, not from screen refresh. In words: records per channel equal the required history duration divided by the sample interval, and total stored values equal that count times the number of channels. A shorter interval catches shorter disturbances and spends storage and controller traffic sooner. Set it from the shortest event an operator must diagnose, sampling at least twice inside that event.
Sizing the interval for 12 hours of four-channel history #
Inputs, a typical case for this kind of station: four channels to trend, a retention target of one 12-hour shift, and a fastest disturbance worth diagnosing that lasts about one minute. The bytes per record depend on the data types plus timestamp and file overhead, which the GP3000 documentation states; the interval decision below does not hang on them.
Rule: divide the required history duration by the sample interval to get records per channel, then multiply by the channel count.
Arithmetic:
- Required history: 12 h × 3,600 s/h = 43,200 s.
- At a 10 s interval: 43,200 s ÷ 10 s = 4,320 records per channel per shift.
- Four channels: 4,320 × 4 = 17,280 stored values per shift.
- Disturbance coverage: 60 s ÷ 10 s = 6 samples inside a one-minute event.
- A week unattended: 7 × 24 h = 168 h → 168 h × 3,600 s/h = 604,800 s ÷ 10 s = 60,480 records per channel.
Decision: set 10 s. It puts six samples inside the one-minute events you are diagnosing and costs 4,320 records per channel per shift — commit to it when the display unit's documented group capacity covers that count. If the target is a full week with no offload, 10 s needs 60,480 records per channel; above the documented capacity, move to 60 s (604,800 s ÷ 60 s = 10,080 records per channel) or schedule card exports.
Bind the trend pens to the logged channels #
- Place a Historical Trend Graph on the screen and open its settings.
- Point its display data at the sampling group created above.
- Assign each pen to a channel of that group; the selector now resolves instead of showing "undetermined".
- Set each pen's vertical range to the engineering span of its signal.
- Set the displayed time span to hold enough samples to read behavior — hours, not seconds, at a 10 s interval.
- Apply and transfer.
Verify before you trust the graph #
- Put a live numeric display of one logged tag beside the trend.
- Move the process so the value changes; the live display following proves the EtherNet/IP path.
- Wait several intervals; new points appearing proves sampling.
- Scroll to an earlier time range; older points still present after the live value changed proves retention.
- If records must survive a power cycle, test it only in a machine state your site approves, then confirm the chosen storage destination kept them.
Screen changes are the same class of problem: the log has to run independently of the displayed screen, which is what it takes to keep a PanelView Plus trend logging after screen changes, and the sampling group gives you that here. Retention deserves the same proof, so run the checks behind verifying HMI historical trend retention before committing before you promise an operator a week of history.
What breaks trends later #
- An interval too fast for the question being asked fills storage and adds controller traffic for detail nobody uses.
- An interval too slow hides the disturbance that prompted the trend between two samples.
- A tag renamed, moved or retyped in RSLogix 5000 after configuration breaks the channel reference; repair the device reference and the sampling channel together.
- Confusing the visible time span with retention: the span is a display setting, capacity decides what still exists.
Questions asked
Do I need a buffer list in the PLC to make the trend channels selectable?
No. The sampling group under Common Settings is the buffer: the GP-Pro EX project defines it, and the GP3000 display unit collects and stores the records itself. Build a controller-side array only when the control program needs to preprocess values for its own logic, never to make trend pens selectable.
Does sampling stop when the trend screen is closed?
No. Sampling is a common setting of the display unit rather than part of a screen, so the group keeps collecting while any screen is displayed. That is why the graph fills in even when nobody has the trend open, and why history keeps growing across screen changes.
Can the same sampling data be exported to CSV?
Yes. The export functions read the same sampling records the trend displays, and alarm data exports the same way. To write to external memory, use the Mode tab setting from the steps above and set Save Data to CF/SD card or USB; the file then contains every channel in the group at the configured interval.
The pens select fine now but the graph draws a flat line. What should I check?
Check the pen's vertical range before blaming the logger. A pen scaled 0 to 100 while the signal runs around 4,000 engineering units compresses a healthy trace into a single pixel line. Set each pen's range to the real operating span of its channel and confirm the displayed time span includes stored samples.