Five objections come up every time an engineer proposes replacing a hardware PLC with an industrial PC running CODESYS, TwinCAT or a real-time Linux stack. Ranked by how often they actually cause failures in the field:
- Uncontrolled OS updates rebooting the machine — real, and the most common cause of unplanned downtime. Entirely preventable.
- Behaviour after a power cut — real, and a hardware specification issue, not a software one.
- Storage wear from logging — real on consumer SSDs, manageable with design.
- Timing jitter — real, but usually misattributed. Most jitter comes from the network stack, not the CPU.
- “Nobody on site can troubleshoot a PC” — the most underrated objection, and the hardest to engineer around.
None of these makes a soft PLC unsuitable. All of them make an unconsidered soft PLC unsuitable. Below is what each one actually involves.
1. Operating system updates
A control PC that reboots itself mid-cycle because of a background update is the single most cited reason engineers distrust soft PLCs. The complaint is legitimate. The cause is a general-purpose OS installed with general-purpose defaults.
What addresses it:
- Windows IoT Enterprise LTSC rather than a desktop edition — a long servicing branch with no feature updates
- A write filter so the system volume returns to a known state on reboot
- Update policy under your control, applied during planned maintenance
- Or Linux with a pinned kernel, which removes the question entirely
What to ask a supplier: which OS editions they supply and license, and whether they can ship a pre-configured image. A supplier who only offers consumer Windows has not thought about control applications.
2. Power loss and restart behaviour
A hardware PLC resumes when power returns. That behaviour is expected and unremarkable. A PC that stays off after a power cut, or boots into a repair screen, is a site visit.
This is not a software problem. It is a BIOS and hardware specification question, and it is easy to verify before purchase.
What to specify:
| Feature | Why it matters |
|---|---|
| Power-on-after-power-fail setting | Determines whether the unit restarts automatically. Look for three options: always on, always off, and last state |
| UPS status input | A signal line telling the system mains power is gone, so it can close files and shut down cleanly instead of losing power mid-write |
| Hardware watchdog | Resets the system if the application stops responding — the behaviour a PLC gives you for free |
| Remote power switch input | Lets the unit be restarted from outside the cabinet without opening it |
| RTC wake | Scheduled restart or wake for maintenance windows |
The UPS status input is the one most often missing. Without it, a PC connected to a UPS keeps running until the battery dies and then loses power abruptly — which is the failure mode the UPS was supposed to prevent. With it, the system knows power failed and shuts down in a controlled sequence.
Example configuration. The BI03-IC exposes a UPS status signal (UPS_OK±) and a remote power switch input (SW±) on its terminal block, alongside BIOS settings for power-on-after-power-fail (on / off / last state), a hardware watchdog, and S5 RTC wake. These are ordinary industrial requirements, but they are not present on every box PC — check the terminal pinout, not the marketing page.
3. Storage wear
Control applications log. Logging writes. Consumer SSDs have limited write endurance and no power-loss protection, which is a poor combination in a cabinet.
What to do:
- Specify industrial-grade storage with a stated TBW figure and power-loss protection
- Keep high-frequency logs in RAM and flush periodically, rather than writing every cycle
- Send historical data to a server or historian, not to the local disk
- Use a write filter so the OS volume is not being written at all
A specification detail worth checking: some Alder Lake-N designs route their PCIe lanes to USB, SATA and UFS, which means the M.2 slot supports SATA only, not NVMe. That is fine for control workloads — sequential logging does not need NVMe throughput — but you should know before you buy the drive. The BI03-IC is one of these: its M.2 slot is SATA, with a second standard SATA port alongside.
4. Jitter and determinism
This is where the discussion usually gets imprecise. “Real-time” means several different things:
| What you need | How it is achieved |
|---|---|
| Deterministic fieldbus cycle | EtherCAT master with a NIC that supports hardware timestamping |
| Time synchronisation across nodes | IEEE 1588 PTP, again NIC-dependent |
| Bounded task scheduling | PREEMPT_RT kernel, or the vendor’s real-time extension |
| Guaranteed network scheduling | Hardware TSN (802.1Qbv/Qbu) — a platform feature, not a software one |
The network interface controller matters more than the CPU here. An EtherCAT master needs a NIC with a mature direct-access driver. Intel i210, i225 and i226 are the controllers commonly used for this. Consumer Realtek controllers generally are not, because the driver support for bypassing the normal network stack is weak or absent.
Be careful with the word “real-time” in supplier datasheets. Hardware TSN and Intel Time Coordinated Computing are supported on Intel’s Atom x7000E industrial line, not on the Intel Processor N95/N97/N100 entry line. A box built on an N-series processor can run a soft PLC, act as an EtherCAT master and do PTP synchronisation — but it is not a TSN endpoint. If a datasheet claims TSN on an N-series CPU, ask which document states it.
The BI03-IC uses four Intel i210 controllers, which is why EtherCAT master operation is supported on it. It does not claim TSN.
5. Who maintains it in five years
This is the objection that is hardest to solve with hardware, and the one most likely to decide whether the project goes ahead.
A maintenance technician can swap a PLC and restore a program from a memory card. Restoring a PC is a different skill set, and if the person who built the image has left, the knowledge often left with them.
What reduces the risk:
- A documented, reproducible system image — not a machine someone configured by hand
- Recovery media stored with the machine, and a written restore procedure
- Identical replacement hardware available years later — which means asking the supplier directly about product lifecycle
- Configuration in version control, not only on the device
On lifecycle, be direct with your supplier. Intel’s embedded and IoT product lines carry extended availability commitments; consumer-oriented SKUs generally do not. Ask what the supplier holds in stock, what notice they give before a platform change, and whether a replacement would be mechanically and electrically compatible. An honest answer to that question tells you more than any datasheet.
Where a soft PLC genuinely fits
Good fit:
- Machine control already built around CODESYS, TwinCAT or a similar runtime
- Systems needing control and data processing on one device — HMI, database, OPC UA server, protocol conversion
- EtherCAT motion control with a reasonable cycle time
- Applications where the control logic changes often and PLC programming turnaround is a bottleneck
- Cabinets where a separate PLC, gateway and edge PC could be consolidated into one unit
Poor fit:
- Functional safety — a soft PLC does not replace a safety PLC or safety relay
- Sub-millisecond deterministic control without a TSN-capable platform
- Sites with no IT capability and no appetite to develop any
- Very high vibration or temperature beyond the unit’s rating
Hardware checklist
Before selecting a box for a soft PLC application:
OS edition offered: LTSC / IoT / Linux — and who licenses it?
Pre-configured image: Yes / No
Power-on after power fail: On / Off / Last state — configurable?
UPS status input: Yes / No
Hardware watchdog: Yes / No
Remote power switch input: Yes / No
NIC controllers: Intel i210 / i225 / i226 / other ______
EtherCAT master supported: Yes / No — on which NIC?
Storage interface: M.2 NVMe / M.2 SATA / SATA
Industrial SSD with PLP: Yes / No, TBW rating ______
Serial ports: ___ × RS232, ___ × RS485, switchable how?
DC input range: ____ V to ____ V
Operating temperature: ____ °C to ____ °C
Fanless: Yes / No — at which CPU option?
EMC standard tested: IEC 61131-2 / EN 61000-6-2 / other ______
Third-party report available: Yes / No
Platform availability: ____ years, EOL notice period ______Frequently asked questions
Can a soft PLC be used for safety functions? No. Safety functions require a certified safety PLC or safety relay. A soft PLC can coordinate with a safety system but cannot replace it.
What cycle time is realistic? It depends on the runtime, the OS configuration and the fieldbus. Single-digit millisecond EtherCAT cycles are common in practice. Sub-millisecond deterministic control needs a platform designed for it. Ask your runtime vendor for figures on comparable hardware rather than relying on a hardware datasheet.
Does an industrial PC need to be fanless? For a sealed or dusty cabinet, yes — a fan is a moving part and an ingress path. Note that fanless operation is usually tied to the CPU option: a 6W or 12W processor can be passively cooled in a compact housing where a 15W part may need active cooling. Check which CPU options are fanless rather than assuming the whole series is.
What EMC standard should a cabinet controller meet? For equipment in a control cabinet, IEC 61131-2 is the standard written for programmable controllers and is a stronger reference than a generic industrial EMC standard. Ask for the test report and check which model numbers it covers.
About the BI03-IC
The BI03-IC is a fanless DIN-rail or back-mounted edge controller with four Intel i210 Gigabit ports supporting EtherCAT master operation, two serial ports (COM2 switchable to RS485 in BIOS), 12–24V DC input, and −20 °C to 60 °C operation. It includes a UPS status input, remote power switch input, hardware watchdog, configurable power-on-after-power-fail, TPM 2.0 and Secure Boot. EMC has been tested to IEC 61131-2:2007.