Troubleshooting
Troubleshooting E-Designer and Q03UDECPU Ethernet comms
Short answer
A dead E-Designer panel comes down to three checks before transport settings matter: Controller 1 moved from Unused to Ethernet, the cable in the configured port, and a full power cycle after the Q03UDECPU's parameters were downloaded. Run them in that order.
On this page
- Why does the panel show no values even though the project looks right?
- Is Controller 1 still sitting under Unused in E-Designer?
- Is the cable in the port the configuration actually names?
- Did the CPU get its power cycle after the parameter download?
- Should the panel use TCP or UDP to talk to the Q03UDECPU?
- How do you prove the whole path works before calling it fixed?
- Questions asked
Your E-Designer panel shows no live values from the Q03UDECPU and switching TCP to UDP changes nothing, so start with the fix, not the transport: move Controller 1 from Unused to Ethernet in the controller list, put the cable in the RJ45 port built into the CPU module itself, and power the Q03UDECPU off and on after its last parameter download, because the CPU only loads stored parameters at power-on or reset. On a typical bench layout, a panel at 192.168.3.10 with mask 255.255.255.0 talks to a CPU at 192.168.3.50 over the same 192.168.3.0 subnet with no gateway in the path. That combination points below the transport, to the controller binding, the physical port or the CPU's startup state, and this article walks each check, works the subnet arithmetic that proves the addresses can talk, and ends with the one forced-value test that proves the whole path.
Why does the panel show no values even though the project looks right? #
Decide first whether one object is dead or the whole controller is dead: a single frozen value means the tag or its binding, every PLC-connected object frozen means the common path, and TCP-versus-UDP changes nothing when the controller was never attached to a channel at all.
Open one failed screen object and read the tag it references. Open that tag and read the controller it references. If the object points at Controller 1 and every other object on Controller 1 is dead too, the fault sits below the tag, in the controller assignment, the driver, the network path or the CPU parameters. Work the list top to bottom, because the first two items clear most dead panels:
- Confirm Controller 1 sits under Ethernet in the E-Designer controller list, not under Unused.
- Confirm the cable is in the Ethernet port the E-Designer project and the GX IEC Developer parameters actually describe.
- Confirm the Q03UDECPU was power-cycled after its last parameter download.
- Compare TCP and UDP last, and only to confirm both ends use the same mode.
Two neighboring symptoms belong elsewhere. Values that arrive slowly but keep arriving point to why HMI point updates run slow on an Ethernet PLC link. Values that run for a while and then stop point to diagnosing intermittent HMI to PLC Ethernet comms hangs. A path that never comes up at all is the fault this article fixes.
Is Controller 1 still sitting under Unused in E-Designer? #
In E-Designer a controller can sit under Unused while its tags look fully configured, and an Unused controller gives the runtime no Ethernet channel at all, which is exactly why TCP and UDP fail identically. Move Controller 1 to Ethernet, download once, and leave the transport alone.
Unused is the default a new or re-created controller lands on, so a project you inherited, re-saved or partially rebuilt can carry a complete-looking tag list over a controller that was never attached. Swapping TCP for UDP cannot touch this fault: the transport setting only matters once the driver is riding an active channel.
- Open the controller configuration in the E-Designer project.
- Find Controller 1 and read which channel it is assigned to.
- If it reads Unused, move it to Ethernet.
- Open the controller properties and confirm the driver is the Mitsubishi driver for a Q CPU over built-in Ethernet.
- Download the project to the panel and confirm the driver opens its channel at runtime.
While you are in the project, read the label on the terminal. This request describes both an E110 and an E1100, and they are different Beijer terminals, so make the E-Designer project target the model actually mounted on the door.
Decision: if Controller 1 was under Unused, that was the fault. Move it, download once, and stop changing the communication mode.
Is the cable in the port the configuration actually names? #
A valid cable in the wrong connector leaves the software path just as dead as a faulted driver, so trace the path physically: on a Q03UDECPU the Ethernet port is the one built into the CPU module itself, and every other RJ45 connector in that cabinet belongs to something else.
Both ends of the configuration must then describe the same two connectors and the same addresses:
| Item | Where it is set | Must agree with |
|---|---|---|
| Panel IP address and subnet mask | E-Designer panel Ethernet settings | The CPU's subnet |
| CPU IP address and subnet mask | PLC parameters in GX IEC Developer | The panel's subnet |
| Controller destination address | Controller 1 connection settings | The CPU's IP, octet for octet |
| Link LEDs | Panel port, switch port, CPU port | Lit on every segment of the path |
Note whether the panel and CPU meet through a switch or a single cable. Through a switch, check link LEDs at all three ports and confirm no VLAN or port isolation splits the two apart. Direct, check link LEDs at both ends, and use static addresses on a bench so nothing waits on a DHCP server that is not there.
Worked example, with illustrative addresses rather than documented defaults: the panel is set to 192.168.3.10 with mask 255.255.255.0, and the CPU's PLC parameters carry 192.168.3.50 with the same mask. AND each address with its mask to get the network portion: 192.168.3.10 AND 255.255.255.0 gives 192.168.3.0, and 192.168.3.50 AND 255.255.255.0 also gives 192.168.3.0. Identical network portions mean the panel and CPU can talk directly or through a plain switch, with no router in the path and no gateway setting needed. Had the panel sat at 192.168.4.10 instead, its network portion would have been 192.168.4.0, the portions would differ, and every request would die looking for a gateway this machine-to-machine link does not have.
Reachability is not identity, either. If something answers at the CPU's address, confirm it is the CPU before treating the reply as success.
Decision: matching network portions, link LEDs lit on every segment, and a destination address that names the CPU exactly. Any one of the three missing, fix it before blaming the driver or the transport.
Did the CPU get its power cycle after the parameter download? #
A successful parameter transfer does not make the new settings live: the Q03UDECPU loads its stored parameters when it is powered on or reset, so a CPU that kept running through the download still drives its Ethernet port with the old settings.
A program restart is not that step. Watching the transfer report success changes nothing about which parameter set the port is using; only power-on or reset loads stored parameters. Mitsubishi's QCPU documentation shows the same distinction from the hardware side: after the CPU is powered on or reset, the time it needs to reach RUN depends on the system configuration and the parameter settings, so let the startup take as long as it takes and judge readiness from the CPU's LEDs rather than a stopwatch.
Removing CPU power stops whatever that CPU controls, so schedule the cycle with production and follow the site's lockout practice.
- Save the PLC project and confirm the Ethernet settings are inside the parameter set you transfer.
- Write the parameters to the Q03UDECPU.
- Bring the machine to a state where a controlled stop is allowed.
- Power the CPU off and back on, or reset it, whichever the site procedure permits.
- Wait for a steady RUN, clear any error indication, and read the CPU diagnostics before touching the panel.
The same medicine recurs on serial links: comms that only return after removing and restoring power are a pattern of their own, and serial comms loss that only a power-cycle reset clears walks through that cousin of this fault.
Decision: after every parameter download, the next commissioned step is power off and on, or reset. If the CPU has not been cycled since the download, cycle it before touching anything else.
Should the panel use TCP or UDP to talk to the Q03UDECPU? #
Commission the transport both projects already agree on and stop there: TCP and UDP are not substitutes to flip for luck, because either one only works when the E-Designer driver and the CPU's Ethernet endpoint are set to the same mode with the same endpoints.
The two behave differently when they fail, and that is their only diagnostic value here. TCP opens a session, so a wrong address or a dead path shows up as a failed connection attempt. UDP sends datagrams with no session, so the same faults just time out quietly. Neither behavior repairs a controller left under Unused, a cable in the wrong port or unactivated parameters; swapping the mode while those three are unresolved adds a variable without isolating anything.
If both modes are supported and configured on both ends, keep the one already written into the CPU parameters, because it demands fewer changes and preserves a single diagnostic baseline. Change the panel's mode and you transfer the project again; change the CPU side and you transfer the parameters and re-run the power cycle from the previous section.
Decision: match the modes end to end, prefer the mode already in the CPU parameters, and in a panel that never talked at all, treat the transport as innocent until the first three checks pass.
How do you prove the whole path works before calling it fixed? #
Link LEDs and one lucky screen update are not proof: finish with a forced, reversible change to one known tag, watched in GX IEC Developer and on the panel at the same time, with nothing else altered while the test runs.
Link indication proves the Ethernet layer only. It says nothing about the driver, the controller assignment, the tag address or the data interpretation. One value that travels in both directions proves all of them at once.
- Pick a value the site procedure lets you change freely, on an object that is not safety- or motion-related.
- Read its current value in GX IEC Developer.
- Change it at the CPU and watch the panel object follow within one screen refresh cycle.
- Write to it from the panel if the object is writable, and watch the value change in GX IEC Developer.
- Return the value to what you found and record the passing result.
Decision: a value that moves in both directions closes the job, because binding, driver, port, addresses and parameters are all proved together. Anything less proves one layer and leaves the rest untested.
Questions asked
How long does a Q03UDECPU take to reach RUN after a power cycle?
There is no single figure. Mitsubishi's QCPU documentation states that the time a CPU needs to enter RUN after power-on or reset varies with the system configuration and the parameter settings. Judge readiness from the CPU's RUN and error LEDs rather than a stopwatch, and test the panel only once RUN is steady.
Does a CPU reset count as the activation step, or must it be a full power cycle?
Mitsubishi's QCPU documentation describes the CPU as loading its stored parameters when it is powered on or reset, so a reset brings the new settings active where the site procedure permits it. What never activates them is the download itself: writing parameters to a running CPU leaves the old set in charge until the next power-on or reset.
Can I bench-test the panel straight to the Q03UDECPU without a switch?
Yes, one Ethernet cable between panel and CPU is enough for a bench test. Keep both addresses in the same subnet, check link LEDs at each port, and if a straight-through cable stays dark between two known-good ports, try a crossover cable before calling a port dead. A bench test that passes with a direct cable also passes through a plain unmanaged switch.
How do I prove the CPU's IP address is not duplicated somewhere on the network?
Power the Q03UDECPU down, then ping its address from a laptop on the same subnet. An answer while the CPU is off means another device owns that address, and the panel's requests have been going to it; silence confirms the address is free. Run this before changing any transport setting, because reachability alone cannot tell the CPU from an impostor.