When you're standing in front of a dead production line and the fault is somewhere inside a Mitsubishi PLC, the bottleneck is usually the same: getting connected. I've spent 12 years at an industrial automation company handling emergency PLC calls, and by now I've done more than 200 rush repairs, including same-day turnarounds for manufacturers that couldn't afford a second day of downtime.
This checklist is what I run through when the plant manager is watching and every minute costs money. It covers connecting to FX3U, FX5U, QnU, and R series CPUs using a USB cable, plus the mistakes I see over and over. Six steps, and about 90% of your connection problems will be solved by one of them.
My first rule: never assume a Mitsubishi PLC has a USB port, and never assume the port type. I learned this the hard way when I drove to a client site with a Mini-B cable for what turned out to be an R32CPU with a USB Type-C port. Check the front panel before you leave the office.
On cable selection: for models with a standard USB port, a good-quality USB 2.0 data cable from a reputable brand works fine. But cheap no-name cables are a gamble—many of them are charge-only and contain no data lines. I'll come back to this in the mistakes section.
GX Works2 and GX Works3 install the MELSOFT USB driver as part of the package, and the manuals say it just works. In practice, it often doesn't. Windows 10 and 11 have tightened driver signature enforcement, and the driver can be blocked silently.
What I do every time now:
The order matters more than people think. If Windows has already assigned a generic driver to the device, the MELSOFT driver can refuse to take over, and you'll be stuck with an "Unknown Device" in Device Manager.
Plug the cable into the PLC first, then into the PC. Wait about ten seconds. Open Device Manager → Ports (COM & LPT). You should see a Mitsubishi entry with a COM number assigned (something like "Mitsubishi FX3U USB (COM5)").
If nothing shows up, don't panic. Check, in this order:
I learned the hard way that GX Works won't tell you what's wrong. It just fails the connection test. So go through these settings methodically:
If it goes green, you're in. If it goes red, go back through Steps 2 and 3 before touching the PLC settings.
Now the real work starts. First, go to Online → PLC Read in GX Works and upload the current program and parameters from the CPU. This does two things: it gives you a backup, and it gives the diagnostics context.
Then open Diagnostics → PLC Diagnostics. The screen shows the current error code and the step address where the fault occurred. Write the error code down now, before you cycle power on the CPU. Many faults clear on restart, and once they clear, that clue is gone.
Here's where the instruction list view earns its keep. Ladder logic is fine for human overview, but when the diagnostic message says "fault at step 2140," find that step in ladder and you'll spend time scrolling. Switch to the instruction list view, jump to step 2140, and you'll see the exact instruction that was executing when the CPU faulted. If the fault is in a MOV instruction, you can immediately suspect a bad destination device. If it's in an analog read instruction, you know where to aim your multimeter next.
When you're in a hurry, this step feels like a waste of time. It isn't. Read the program, save it with a filename like Line2_FX5U_20250110.gx3, and put it somewhere you won't lose it.
I still kick myself for an early-career rush job where I skipped the backup, made a change that made the fault worse, and had to reconstruct the original rung from phone screenshots. That job added two hours to a repair that should have taken ten minutes, and the client watched every one of those minutes.
If nothing else, remember this: the ten seconds it takes to save a backup is cheaper than the hour it takes to undo a bad change from memory.
Three of these cost real production time in field reports I know about:
If you're through all six steps and Windows still doesn't see the PLC, it's worth suspecting a damaged USB port on the CPU or a port disabled in the PLC parameters. Ethernet is your fallback. But in my experience across 200+ emergency connections, this checklist resolves the vast majority of cases—and a shocking percentage of them come down to the cable.