SiteIt's 3am. The palletiser is stood on step 14 and nobody knows what step 14 is waiting for. The inputs all look right. The HMI just says "sequence fault". Somebody has a laptop with the software on it. Somebody else says "just go online and have a look".
Going online is the fastest way to see why a machine won't move. You can watch the logic run, see which condition is false, and go straight to the cause. Done well, it turns an hour of guessing into ten minutes.
Done badly, it's how a breakdown turns into a lost program, a machine that moves when it shouldn't, or a fault that comes back next week because someone "tweaked a timer".
Watch, don't change
Here's our view, and we'd put it on the wall by every panel. When you go online to fault-find, you are there to watch. You're not there to edit, download or force anything.
Monitoring the logic changes nothing in the PLC. An online edit, a download or a force does. And once you've changed the program, you own the change, whether you meant to or not.
That matters more than it sounds. HSE looked at 34 control system incidents. It found the primary cause of a fifth of them was in changes made after commissioning* (HSG238). HSE notes the sample was small, so read it as a warning, not a rule. One case in the same guide involved a change to PLC interlock software that hadn't gone through design staff or testing. It "contained a simple coding error".
Before you connect
Five checks before the cable goes in.
- The right software. Same platform, and a version that can open the project. Opening a project in a newer version can offer to upgrade it. Say no.
- The right project file. It should match what's in the PLC. If you're not sure, do a compare before you trust anything on screen.
- A backup first. Upload what's in the PLC now and save it with the date and time, before you do anything else. If there isn't a known good backup on site, read when the PLC loses its program tomorrow.
- A clean laptop. HSE lists maintenance activities and software upgrades among the routes for threats into control systems (HSE, cyber security). That page is written for major hazard sites, but the point carries. A laptop that's been on every network in the county is a risk.
- The right person. Someone who knows the platform, and knows the difference between monitoring and editing on it.
Finding what it's waiting for
Start from the output or the step that won't happen. Find the rung or block that drives it. Go online and look at which condition is false. Cross-reference that bit and follow it back. Keep going until you hit a real input, a timer or a fault bit.
Most of the time you end up at one of these:
- an input that's flickering too fast to see on the card LED
- a timer that's timing out because something upstream is slow
- a fault latch that needs a reset from a place nobody knows about
- a step that waits for a confirm signal from another machine
- a counter or a data value that's wrong after a power dip
If it's a real input, go back out to the field. Our guide to finding the missing signal covers the card and the wiring.
Write down what you found, with the rung or block number. The next time this fault happens, that note saves the next person the whole search.
Forces and online edits
A force overrides the real world. The PLC acts as if an input is on, or drives an output, whatever the field device says. That means it can bypass the checks the program would normally make.
Under PUWER, a control system isn't safe unless, so far as is reasonably practicable, a fault in it "cannot result in additional or increased risk to health or safety" (L22, regulation 18). A force left on is a fault you put there yourself. HSE's control systems guide also asks designers to "minimise the need to bypass or override safety systems during maintenance" (HSG238). Doing it from a laptop at 3am undoes that work.
If you truly have to force something to prove a fault, treat it like any other bypass. Clear the machine. Isolate what you can. Tell everyone. Have someone watch. Take the force off before you go, and check the force list is empty.
Never touch safety logic. On a safety PLC or a safety relay circuit, the machine stays down until someone competent fixes the fault itself.
If you changed something
Sometimes the fix really is in the program. A timer that's too tight for a slower product. A sensor that's moved and needs a new setting.
Make the change through your site's change process, not on the fly. Write down what you changed, why, and who agreed it. Save a new backup with a clear name. A change to how a machine works can also bring new duties with it, which we cover in machinery modifications. Keep the backup safe too. There's a simple way to do that in PLC program backups and version control.
The person with the laptop
Going online is a skill. It takes knowing the platform, the software and the machine. On many sites it lives in one controls engineer, or with the machine builder, or nowhere at all.
So picture the 3am version of this. The palletiser is down. Nobody on shift can go online. The builder's helpline opens at eight. You've got five hours of trailers stacking up and nobody who can tell you what step 14 is waiting for.
That's the real cost of the skill gap, and it's worth putting a number on before it happens. Work out who on each shift can go online safely. If the answer is "nobody on nights", have a controls engineer you can call before you need one. If you're training your own, our guide to PLC courses is a good place to start.
What to do on Monday
- Check you have a dated, known good backup of every PLC on site, stored off the machine.
- Keep one laptop for PLC work, with the right software versions, and nothing else on it.
- Write a one-page rule: monitor only, no forces or edits without a second person and a record.
- List who on each shift can go online. Fill the gaps before the next night breakdown.
* The HSG238 incident analysis figure correct as of 30 September 2026. Check HSG238 on HSE's website for the latest. Regulation 18 text is quoted from L22, fourth edition (2014, amended 2018).