"We should probably replace this old Allen-Bradley PLC, right?" I get some version of that question at least once a week — usually in a hallway, usually from someone holding a printout of a quote for a new CompactLogix.
And if they're hoping for a clean yes or no, my answer tends to be a letdown. Because there's no universal upgrade path. A SLC 500 running a non-critical packaging line and a MicroLogix 1400 that faults twice a shift on an automotive weld cell are in two completely different worlds. The advice "anything over 10 years should get swapped out" or "if it ain't broke, don't touch it" only tells you that the person giving it hasn't asked the right question.
In 2023, after reviewing specs for roughly 50 control cabinets, we rejected about 15% of first deliveries — and the root cause was almost always the same: someone picked a "standard" upgrade without matching it to the actual production scenario. So here's the decision tree I use to sort old Allen-Bradley PLC hardware into four buckets: non-critical with spares available, critical with long remaining life, fault-prone with compliance exposure, and short-life with constrained budget. See which one matches your situation.
Scenario A: Non-critical machine, low cycles, spares still exist
Typical case: a pump station, a fan control, a conveyor that runs two shifts and shuts down cleanly every night. You've got a MicroLogix 1400 or a SLC 5/03 that predates your newest technician, and the machine it controls produces something that, if it stopped, would be annoying — not catastrophic.
My advice here is contrarian, so hear me out: don't replace it just because it's old. The Allen-Bradley PLC MicroLogix 1400 line was discontinued in 2023, which lit a fire under a lot of maintenance managers — but a discontinued unit with a stable program and a spare CPU in the cabinet can easily run another decade. Replacing it early introduces commissioning risk, redrawn schematics, and a learning curve for operators, all to solve a problem you don't have yet.
What you should do instead is buy yourself options. Stock one spare base unit and one I/O module. Keep a copy of the program on a USB stick stored somewhere that isn't the machine's cabinet — I've seen a washdown destroy every printed drawing taped inside a door (thankfully we had an off-site copy). If you're using an 8000 watt quiet generator for backup power during outages, check its total harmonic distortion spec. Higher-end inverter models keep THD under 5%, while some conventional units push past 8%. PLC power supplies are more tolerant than people think, but repeated brownouts from a dirty generator will eventually cause the kind of intermittent faults that make you blame the PLC when the problem is upstream.
(Side note: "non-critical" only holds if you re-check it every year or two. A line that was non-critical in 2019 might now be your main revenue driver. Just ask anyone who "temporarily" rerouted production during the pandemic.)
Scenario B: Critical process, long remaining life, budget approved
Now we're in different territory — a machine that can't stop, has more than five years of planned service ahead, and where the cost of a bad hour dwarfs the price of hardware. This is the case for a phased migration, not a rip-and-replace.
The counterintuitive move: don't start with the machine that fails most often. Start with the one whose failure costs the most — there's usually a difference. The frequent-failure machine may have a solvable root cause (a bad output card, a worn starter, a messy ground). Replacing it first gives you the emotional satisfaction of fixing the obvious squeaky wheel, but does nothing about the low-frequency, high-impact machine sitting quietly in the corner until it takes the whole line down for a week.
Migration also means being honest about the ecosystem. Going from MicroLogix to CompactLogix isn't just a CPU swap — it's software (Studio 5000 vs. RSLogix 500), training, network rework, and the operator interface library. If you're already invested in the Allen-Bradley platform and your maintenance team knows it, staying inside that family keeps support simple. That's a real advantage, not a sales pitch. But be clear-eyed: this path only makes sense if the phasing is actually funded through the end. A half-migrated panel is worse than leaving it alone.
Scenario C: Intermittent faults, safety or compliance exposure
This is the easiest decision, even though the project is the hardest. If the old PLC is causing nuisance faults, has lost spare-part support, or — the one that ends the debate for me — shows up in a safety or compliance audit finding, replacement isn't optional.
The trap in this scenario is the "diagnose forever" loop. By the time a PLC is faulting unpredictably, you've likely spent weeks swapping cards and blaming noise. Before you commit to replacement, though, there are two thirty-second checks worth doing. First: look at the control panel LED lights. If the specific output LED is lit but the starter doesn't pull in, the problem is downstream — wiring, control-circuit fuse, or the starter itself. If the LED is dark, look upstream at the program, a force bit, or the output module. That split alone eliminates half the "the PLC is bad" reports I see.
Second: if the fault is in a motor starting circuit, test the starter solenoid coil with a multimeter (power off, locked out). A good hold-in coil will read somewhere in the 10–30 ohm range, but check your motor's datasheet for the actual spec — I'm a quality reviewer, not an electrician, and solenoid coil values vary by frame and starter class. If the coil reads open or shorted to ground, the PLC was never the problem.
Scenario D: Short remaining life, constrained budget
The last bucket: the machine has one to three years left, the budget committee is looking at you sideways, and the current PLC still works on most days. Here, the goal isn't modernization — it's controlled extension.
My advice is to stop thinking in terms of "replace the PLC" and start thinking in terms of "replace the failure points." On legacy Allen-Bradley hardware, the parts that die first are usually not the CPU. They're power supplies, output modules (especially relay outputs driving inductive loads), and communication cards. If you can buy used-but-tested replacements for those three categories and store them, you've bought yourself most of the runway at a fraction of the cost. Add a surge protection device at the panel feed and inspect it annually. That's a couple hundred dollars spent to avoid a mystery fault every summer.
The risk with this path is that used parts get scarcer and more expensive as time goes on — the upside is that it can carry a machine past its planned end-of-life without a capital request. I kept asking myself: is another two years of uptime worth the risk of a dead secondhand card in the middle of a production run? I'll be honest — I second-guessed this route every time I approved it. What if the replacement card fails and the vendor won't take it back? The answer, in every case so far, was to buy two and test the spare on a bench before storing it. Not glamorous, but it worked.
How to figure out which scenario you're actually in
If you're still not sure where you land, work through these five questions in order:
- What does an hour of downtime actually cost? Not what the CFO says in a meeting — what it cost the last time the line stopped. That number sets the ceiling on every other option.
- How many years does this machine have left, honestly? Under three, and extension beats replacement. Over seven, and bad money is going into a dead platform.
- Do you have spares for the failure-prone items? Not for the CPU — for the power supply, the output modules, the comm card. If not, you're already in Scenario C whether you've admitted it or not.
- Is there any open safety, compliance, or insurance finding? If yes, the decision is made; everything else is project planning.
- Does your team really know the old platform? A legacy system your technicians can troubleshoot at 2 a.m. is worth more than a modern system they're still learning.
The honest answer is that most old Allen-Bradley PLC decisions aren't technical. They're scheduling and budget decisions wearing a technical costume. Once you know which of the four scenarios you're actually in, the technical path gets a lot simpler — and you can stop arguing about CPUs and start arguing about the two things that matter: how much a failure costs, and how long you still need the machine.
An informed version of that conversation is the one worth having. (And if you can't answer question one, I'd start there.)
Ask about this topic
Share your application conditions and the project stage so the discussion can focus on relevant ratings, interfaces, and documentation.