For the past seven years, I have handled replacement and new-order parts for industrial controls. I'm the person who documents the mistakes so the rest of my team doesn't repeat them. I've personally made—and written down—five significant ordering mistakes, totaling roughly $11,000 in wasted budget. That's not counting the delayed shipments.
Here's the thing: there isn't one universal list for choosing a power supply module, processor modules, digital input and digital output modules, an output module of PLC, flex IO analog input module, or an HMI touch screen. The right choice depends on where you are in the lifecycle—new build, failed replacement, or spare stock. I'll walk you through each scenario and then give you a simple way to figure out which one you're in.
When I first started doing this, I assumed the processor was the hardest component to choose. That was wrong. The processor modules are actually pretty forgiving if you get the specs right. The boring parts are the ones that trip you up: the power supply module, the output type, and the exact analog input range.
The event that changed how I think about this happened in February 2021. A 30 ms power dip hit our line. The processor didn't fail—the rack power supply module couldn't hold up the backplane voltage long enough. The PLC restarted, the recipe was lost, and we lost a shift. That's when I learned that the supply is not an afterthought.
You're in One of Three Scenarios
The mistake I see is giving the same advice for every situation. If you're building a new system, you have freedom. If you're replacing a dead module, you have constraints. If you're ordering spares, you have neither of those pressures but a different risk. These are different jobs.
Scenario 1: Building a New Cabinet From Scratch
Start with the power budget, then the processor, then the I/O, then the HMI touch screen. Not in the reverse order.
Power supply module: size it for the worst moment, not the average
The rack power supply module is not the same as the 24 VDC field supply. The rack supply powers the processor modules and I/O on the backplane; the field supply powers sensors, relays, solenoid valves, and the HMI touch screen. Both deserve their own calculation.
For the rack supply, add the backplane current draw from the processor modules and each I/O module. For the field supply, add the external load of every output module of PLC, plus analog transmitters, plus the HMI touch screen's backlight. Then add at least 30% headroom. A relay output module can pull more when every relay coil energizes at the same time. The current spike can dip the voltage below the 20.4 V minimum allowed by IEC 61131-2 for 24 V digital I/O. The result is random processor resets.
If you're in a facility with brownouts, I'd add a UPS or a supply with ride-through time. If your mains are stable, a good 24 V supply with headroom is enough. Honest answer: this is one of the few areas where the cheap supply is actually more expensive.
Processor modules: don't oversize by default
The processor modules need to handle three things: the number of I/O points, the program size, and the communication protocols. The counterintuitive part is that bigger isn't always better. A processor with lots of memory can also need a bigger backplane supply, a different software license, and more commissioning time. I've seen buyers choose a high-end CPU “to leave room to grow” and then grow onto a different platform two years later.
If you're using remote I/O, check that the processor can handle the remote bus and the number of nodes. If you're using a flex IO analog input module on a remote rack, you also need to know whether the processor supports that type of adapter. Most do, but check before you order.
Digital input and digital output modules: source/sink and current per channel
The digital input and digital output modules for a 24 V machine are not all the same. Verify wiring type: PNP (source) or NPN (sink). Check current per channel and total module current. Output modules also differ in isolation and switching frequency. A relay output module of PLC can switch AC or DC but is slower and wears out. A transistor output switches fast but needs external fusing.
I once matched a “24V DC input” module from the same family without looking at the input filter delay. It filtered out 5 ms pulses from a sensor, and the machine counter undercounted. Same signal, different filter, wrong module.
Flex IO analog input: “analog” is only the beginning
The flex IO analog input category includes 0-10 V, 0-20 mA, 4-20 mA, RTD, thermocouple, and sometimes high-speed counter variants. Ordering “an analog module” is not enough. You have to specify the signal range and sensor type. If you use 4-20 mA field devices, check whether the module supplies loop power. If not, add a 24 V power supply for those loops. I made that assumption mistake on a small project a few years ago: I assumed the module would power the sensors. It didn't.
HMI touch screen: protocol beats brand
The HMI touch screen is usually the easiest to size—screen diagonal, resolution, available cutout, IP rating—but the protocol support is where careers end. I once ordered screens that fit the panel perfectly. They didn't support the PLC's Ethernet protocol. 20 screens went back. Freight both ways.
Before you send the PO, write down the exact protocol the HMI must speak: Modbus TCP, EtherNet/IP, PROFINET, or a proprietary driver. Then check the HMI manufacturer's driver list. If you need serial, confirm RS-232/422/485 and the connector layout. The screen is only a screen until it has to talk to your processor.
Scenario 2: Replacing a Failed Module on an Existing Machine
Now the rules change. You aren't designing a new system; you're matching an old one. Here the part number is king.
Photograph the label before you remove anything
For a replacement power supply module, processor module, or output module of PLC, write down the exact catalog number, series, and firmware revision. I have seen modules that look identical and have completely different backplane behavior. The value of a 30-second photo is enormous.
Also check whether the replacement module's firmware version is compatible with your CPU and software. I learned that on a Thursday, mid-afternoon, with the machine down. Not recommended.
Sometimes the fastest fix is an upgrade, not a direct replacement
This is the counterintuitive advice. When a processor module is no longer available from the original manufacturer, my first instinct used to be to hunt for new old stock. Now I quote both: the exact obsolete part and the current equivalent of the same platform. If the current CPU can run the old program with a modest conversion and is in stock, that can be faster than waiting weeks for a surplus module. The old module isn't always the safe option.
I'd say this applies to about 80% of the obsolete processor cases I see. Here's how to know if you're in the other 20%: if the program conversion would require a complete rewrite, or if the machine has no HMI/network dependencies, a direct replacement might still be cheaper. But the point is to check both paths before ordering.
Don't skip the backup
Before swapping anything, ensure you have a full program backup and the current configuration files. Even a “simple” power supply module replacement can force a processor restart. If the program isn't backed up, that restart becomes an emergency. That's a mistake I made once. Just once.
One important limit: if your machine has a safety-rated function, the replacement module must follow the manufacturer's safety instructions and the system's safety plan. Don't use general advice from an online article for safety functions. This isn't that.
Scenario 3: Stocking Spares for an Installed Base
This scenario is the least exciting but the easiest place to waste money. The goal is not to buy one of everything. The goal is to survive the failure you're most likely to have.
In most cabinets, the modules that fail first are power supply modules, relay output modules, and HMI touch screens. Processor modules fail less often, and digital input modules are generally more robust. For a plant with multiple identical machines, one spare of each failure-prone module per set of machines makes sense. For a single production line with very high downtime cost, you may need a spare in the panel, not just in the stockroom.
The honest limitation: spare part lists are unique to your plant. If your facility is remote and shipping takes ten days, your desired inventory is different from a plant with a distributor across the street. I can't tell you the exact number. I can tell you that a drawer full of the wrong modules is worth nothing.
How to Tell Which Scenario You're In
Use these three questions:
- Are you designing, replacing, or stocking? Designing follows Scenario 1. Replacing follows Scenario 2. Stocking follows Scenario 3.
- What is the installed control platform? If you're replacing, match the platform. If you're designing a new machine, the installed base matters because your electricians and programmers already know how to support it. Introducing a second PLC brand without a good reason adds cost in training, spares, and software.
- What is the cost of one hour of downtime? If it's high, you may need to pay for expedite or keep redundant modules on site. If downtime is annoying but not fatal, a faster purchase process may be enough.
If you answer “replacing” and “the old part is discontinued,” then the decision tree from Scenario 2 applies. If you answer “designing” and “this machine is isolated from your other controls,” then you can choose the best fit without legacy constraints.
So, What Should You Order?
That depends on your scenario. Not because I'm trying to be vague—it genuinely depends. A new cabinet needs a power supply module with headroom, processor modules matched to the I/O count, digital input and digital output modules with the right wiring type, a flex IO analog input module with the exact signal range, and an HMI touch screen that actually supports your protocol. A replacement needs matched part numbers. A spare stock needs the failure-prone pieces.
The important part isn't a magical part list. It's a process: define the scenario, verify the part numbers, check the protocol, and leave headroom on every power supply. That's it.
You can thank me later—or, more realistically, thank yourself when the machine isn't down for a week.
