Mitsubishi PLC Series: A Buyer's Comparison Across FX, Q, and A Series—What Actually Works for Your Shop Floor

How I Approach PLC Purchasing When Engineers Don't Agree

When I took over industrial parts purchasing in 2020, I figured a PLC was a PLC. You spec it, order it, install it. Simple, right?

Then I watched three engineers argue for forty minutes over which Mitsubishi series to use for a conveyor upgrade. They kept saying "A series is dead" and "Q series is overkill"—and I had no idea who was right. So I started digging.

Here's what I've learned from processing roughly 60-80 PLC-related orders annually—including a painful misorder on an A series module that taught me a costly lesson. Consider this a buyer's comparison, not an engineer's deep dive. I'm comparing the models the way an administrator compares them: availability, cost, support, and whether you can actually get the thing when you need it.

We'll look at four series: FX (mainly FX3U and FX5U), Q series, R series, and the A series (which often gets written off but still shows up in legacy systems).

Performance & Capacity: The Obvious One—But With a Twist

Everyone knows the hierarchy: FX < Q < R. But here's where my gut said something different than the datasheets.

Q series has been the workhorse for years. It handles large I/O counts (up to 4,096 points for the QnU series) and runs multiple CPUs in one rack. For a mid-size system—say, 200-500 I/O—it's typically the sweet spot.

R series is Mitsubishi's flagship. Faster scan times (around 0.009ms per basic instruction), larger memory (20 MB program capacity for R32), and better integration with modern networking. The numbers say it's the best choice for new projects. My gut? Something felt off about recommending it universally.

Turns out my gut was half right. The R series is objectively more powerful—or rather, more capable for high-speed, data-heavy applications like packaging or assembly lines. But for a simple pump control with 50 I/O, you're paying for capacity you'll never touch. That's fine if you plan for future expansion. It's wasted if you don't.

And the FX series? Don't underestimate it. The FX5U can handle 256 I/O, has built-in Ethernet, and is surprisingly robust. We standardized on FX5U for small machines after realizing we'd been overspeccing Q series on some projects (which, honestly, I think the engineers just liked the Q series' programming interface better).

Note: The R series motion control with the MR-J5 servo system is genuinely impressive if you're doing positioning applications. But if you're just turning motors on and off, you won't see the benefit.

Availability & Lead Times: Where I Eat Mistakes

This is the dimension that matters most to my job—and where I've made the most errors.

FX series (especially FX3U and FX5U): Generally in stock at major distributors like Mr. PLC or online vendors. Lead times I've seen: 1-3 weeks for standard units, maybe 4-6 for odd configurations.

Q series: More variable. The popular CPUs (Q03UDVCPU, Q06UDVCPU) are typically stocked. But specialized modules—like a QJ61BT11N CC-Link master—can push to 8-10 weeks. I had one engineer specify a Q172DSCPU motion CPU that took 14 weeks. The project manager was not pleased (a communication failure I'll get to in a moment).

R series: The newest, so distribution coverage is still building. Major models like R04CPU and R08CPU are available, but some I/O modules (especially the new ones) can have longer lead times. As of late 2024, I've seen R series lead times averaging 4-8 weeks versus 2-4 for comparable Q series.

A series: Here's where it gets interesting. Many distributors treat A series as obsolete—and Mitsubishi has officially moved on. But here's the thing: if you're maintaining an existing system, there's still demand. I've found A series modules on eBay, surplus sites, and specialty distributors. The risk? You're buying old stock (sometimes 10+ years old). And if a module fails, finding a replacement that hasn't been stored improperly is a gamble.

The "A series is dead" thinking comes from an era when every upgrade meant replacing cabinets. But some plants have 20-year-old A series systems still running perfectly, and the cost to replace them (downtime, rewiring, new programming) can exceed $50K. So A series survives in the aftermarket.

Pricing: Not Just the Unit Cost

I'm not giving exact prices—they fluctuate and depend on your distributor relationship—but here's the rough tier I see in my ordering:

• FX series (e.g., FX5U-32MT): Budget-friendly. Typically 40-60% less than a comparable Q series CPU.
• Q series (e.g., Q06UDVCPU): Mid-range. Workhorse pricing for workhorse performance.
• R series (e.g., R04CPU): Premium. 20-40% more than comparable Q series, though the performance delta often justifies it.
• A series (aftermarket): Wild card. I've seen A2SHCPU modules priced from $200 (used, no warranty) to $1,500 (new old stock, guaranteed). The inconsistency is the real cost—you never know if the next module will be available.

But here's the real cost that bit me: I said "standard A series module." The vendor heard "any A series replacement." Result: they sent a A1SJ71UC24-R2 communication module instead of a CPU. The difference? $1,200 in restocking fees and a week of schedule slip. That's when I learned to verify exact part numbers—including revision suffixes—before ordering.

The spreadsheets always said saving money on surplus A series parts was smart. But my gut kept warning me about long-term availability. After that restocking incident, I now insist on a documented availability guarantee from surplus vendors—or we spec a migration plan to FX5U/Q series.

Software & Programming: The Hidden Compatibility Snafu

If you think choosing the hardware is the hard part, wait until you deal with software compatibility. This is another place where I've found the obvious choice isn't always right.

GX Works2 covers FX and Q series—it's mature, stable, and most engineers know it. GX Works3 is required for R series and FX5U. It's more modern (the interface is actually better in some ways) but has a learning curve if your team is used to GX Works2.

Here's the real-world impact: We have two senior techs who know GX Works2 inside out. Moving them to GX Works3 means training time—call it a 20-40 hour hit per person up front. If you're just upgrading one small machine, using R series (which requires GX Works3) may cost more in lost productivity than the hardware savings.

For A series: GX Developer (older software, still functional for maintenance). Fewer engineers are comfortable with it, so if you're maintaining A series, factor in that your support options are shrinking.

One tip I picked up the hard way: Check which version of GX Works2 you need for your specific CPU firmware. We ordered a Q06UDVCPU with firmware 6.40 and our GX Works2 (version 1.5something) couldn't connect. A firmware update for the PLC ran $300 plus a tech call. I now ask distributors for the CPU firmware version when ordering.

Support & Documentation: What I Actually Rely On

As an administrator, I don't write PLC code. But I do track who supports what.

Mitsubishi Electric Automation (meau.com) provides official support for FX, Q, and R series. Documentation is solid—manuals, application notes, software downloads. For A series, official support is limited; most documentation is archived.

Distributors like Mr. PLC, PLCHardware, and online industrial suppliers are my go-to for quick questions and availability checks. Some have technical support teams that can confirm compatibility (which saved me after that A series disaster).

For third-party support on legacy systems: There are specialized shops that repair and support A series PLCs. I've used two, and the experience is variable—one was great, the other sent me wrong modules twice. If you're running A series, I'd vet a repair service before you need them (when you need them is usually during an unexpected breakdown, not a good time to be verifying credentials).

When Each Series Makes Sense (And When It Doesn't)

Based on what I've seen work (and fail) across maybe 200 orders—give or take—here's my honest, no-BS recommendation:

Choose FX series (FX3U or FX5U) when:

• Your application needs 256 I/O or less
• You want Ethernet, basic motion, and PID without paying for premium features
• Your team is comfortable with GX Works2 (FX3U) or GX Works3 (FX5U)
• Budget matters—FX is genuinely the price-to-performance sweet spot now

Choose Q series when:

• You need more than 256 I/O (mid-size systems)
• You want multiple CPUs, redundant systems, or specialized modules (CC-Link, safety, motion)
• You need long-term availability and distributor stock is consistent

Choose R series when:

• You need maximum performance (high-speed processing, large data handling)
• You're building a new system from scratch and want the latest ecosystem
• You're willing to pay the premium and accept the learning curve for GX Works3

Stick with A series (for now) when:

• You have a running system that's stable, and replacement cost (downtime + hardware) outweighs migration cost
• You can source reliable aftermarket modules (vet the supplier carefully)

One Final Piece of Honest Advice

Don't automatically go for the newest model. I've seen projects where the engineer spec'd an R series "because it's the latest"—and then the company struggled with software learning curves, longer lead times, and unnecessary hardware cost. Conversely, I've seen plants clinging to A series because "it works"—and then a module failure caused a 2-week shutdown waiting for a replacement from a surplus dealer in Ohio.

The right choice depends on your specific machine, your team's skills, your timeline, and your budget. And if you're the administrator caught in the middle of that engineering debate, here's my survival tip: ask each engineer to write down the top three reasons for their recommendation—and make them include availability and lead time estimates. That simple change cut our spec-conflict time from 40-minute arguments to 10-minute decisions.

Hope this helps you avoid the restocking fee I paid. If I remember correctly, it was $1,200—but don't quote me on that exact figure. It's been a couple years.