What I Thought Was a Siemens PLC Power Supply Failure
September 2022. 2:14 AM. My phone is ringing, and the caller ID tells me it's the plant manager at a packaging line we'd commissioned six months earlier.
"Everything's down. The PLC just went dark."
My first assumption was the obvious one: the Siemens PLC power supply had failed. We'd installed an S7-1500 with a PM 1507 module feeding a 24 VDC rail — a textbook setup. The rail read dead on the meter. So I did what any reasonably experienced engineer would do at 3 AM. I drove in, swapped the module with the spare from the van, and the line was running again by 5:30.
Problem solved, right?
It stayed up for eleven days. Then it went down again. Same symptom. Same dead 24 V rail. Same late-night drive.
That's when I stopped trusting my first assumption.
The Problem Was Never the Power Supply
Here's the part that took me three months and a fair amount of embarrassment to accept: I replaced four power modules on that line before I understood what was actually happening. The modules weren't failing. They were tripping — and the fault was downstream.
When I finally pulled the data logs and cross-checked them against the machine's event history, the pattern was obvious in hindsight. Every failure followed the same sequence: a high-torque actuator would fire, the rail would sag for a fraction of a second, and the power module's overcurrent protection would do exactly what it was designed to do. Shut down. Protect the hardware. Wait for a human to figure it out.
Three things were compounding each other:
- The load calculation was wrong. We'd sized the 24 V rail off steady-state current. Nobody accounted for inrush from the contactor coils and the two VFD pre-charge circuits. On paper we had 35% headroom. In practice, at the moment of switching, we were flirting with 90%.
- The grounding was inconsistent. The machine builder had bonded the panel backplate to earth at one point. Our installation added a second bond at the door. Two paths, slightly different impedance, and a floating potential that showed up as noise on the 24 V rail whenever a nearby welder was running.
- Temperature. The panel sat on the warm side of a south-facing wall. Summer afternoon temperatures inside the enclosure were hitting 48°C — well above what the PM module's derating curve assumed. The ventajas y desventajas of the S7-1500 family are usually discussed in terms of features and software ecosystem. Nobody talks about the derating chart, and that chart is the one that mattered on this job.
I want to be precise here: none of those were defects in the Siemens hardware. The hardware did what the spec sheet said it would do. The failure was in how we specified the system around it.
The hardest part of diagnosing a power problem is accepting that the power supply might be the messenger, not the cause.
What This Actually Cost
Four power modules at roughly $340 each. That's $1,360 in parts, which sounds manageable until you add the rest.
The line was down for a cumulative 31 hours across those three months. At the plant's throughput — around $2,800 an hour of production — that's somewhere near $86,000 in lost output. Two partial batches were scrapped because the shutdown happened mid-process. I don't have the exact scrap figure, but the plant manager told me it was "in the thousands." I believe him.
And then there's the number that doesn't show up on any invoice: credibility. This was a line I'd personally signed off on. Every 2 AM call chipped away at the trust the client had placed in us. When the fourth failure happened in January 2023, the client's maintenance lead asked me — politely, but clearly — whether we were sure we knew what we were doing.
Fair question. At that point, I wasn't sure either.
I'm not a power electronics specialist, so I can't speak to the deeper physics of rail impedance and harmonic interaction. What I can tell you from nine years of doing this is that the vast majority of "power supply failures" I've seen weren't failures at all. They were sizing errors, grounding errors, or thermal errors that happened to present as power supply failures.
The Checklist That Actually Fixed It
After the fourth failure I built a pre-commissioning checklist that our team now runs on every panel. It's boring. It works. That's the whole point.
- Load calculation with inrush, not just steady state. Every coil, every drive pre-charge, every capacitive load gets a number. We add 25% minimum headroom on top of the worst-case simultaneous inrush figure. Not 25% on the steady state — 25% on the transient peak.
- Single-point grounding. One bond to earth. One. If the builder did one and we're adding another, we remove theirs or document why it stays. This alone eliminated the noise issue on the packaging line.
- Thermal derating, checked against the actual enclosure, not the datasheet room. We measure interior temperature under load on a hot afternoon before signing off. If it's above 40°C, we either add ventilation or we re-spec the supply with the derating curve applied.
- Log review at 30 days and 90 days. Pull the diagnostic buffer, look at the fault history, and read it like a story instead of a status report. The pattern was in our logs for two months before I actually looked.
(One thing I'd add if you're running a generator-backed control panel: this problem gets worse. Backup power behaves differently from grid power, and the etiquette of matching a generator to a controls load is its own discipline — not unlike choosing between a Kohler and a Generac for home backup, where the spec sheet numbers rarely reflect what happens at the moment a compressor kicks on. Same lesson, different equipment.)
We've run that checklist on 47 panels since early 2023. Two caught marginal sizing before commission. One caught a grounding conflict the builder had introduced. Forty-four were clean. None of them called me at 2 AM.
If you're specifying a Siemens PLC power supply, the module itself is rarely the thing that catches you. Siemens builds a good piece of hardware. The problem is usually the system you built around it — the load you didn't calculate, the ground you didn't remove, or the heat you didn't measure because the datasheet was written for a lab, not a factory wall in July.
The supply does its job. It's our job to make sure it has one it can actually do.