IMD SolutionComputing LibraryShop main store

Troubleshooting

Host blue screen during serial VM connection to a PLC

Remote access and virtualization Published 10 min read

Short answer

A host blue screen during a VM serial session to a PLC is a Windows 10 host crash in the USB-serial and virtualization stack, not a PLC or guest fault. Capture the stop code and version stack, check whether Workstation 14 is documented for your Windows 10 build (Broadcom lists Windows 10 1903 hosts from Workstation 15.x upward), then upgrade Workstation, refresh VMware Tools, and retest before touching adapter or PLC settings.

On this page
  1. Why does my Windows 10 host blue-screen when a VM talks serial to the PLC?
  2. Is it the host or the virtual machine that crashed?
  3. What does Page Fault in Nonpaged Area actually tell you?
  4. Which versions do you record before changing anything?
  5. Is VMware Workstation 14 supported on your Windows 10 build?
  6. Do you fix the VMware version or the USB serial adapter first?
  7. Do you need to touch the PLC or the serial settings?
  8. When do you replace the workstation instead?
  9. Questions asked

Your Windows 10 host is the machine that blue-screens, and the change that ended this case was a VMware Workstation upgrade, not a PLC, cable, or RSLinx setting. On a Windows 10 host running Workstation Pro 14, with a Windows 7 guest using RSLinx through a USB-to-serial adapter to reach MicroLogix controllers, auto-configuration crashed the host with Page Fault in Nonpaged Area, and a manual configuration that connected still crashed it at the going-online step. Upgrading Workstation Pro from 14 to 17 stopped the crashes. That result puts the fault in the host's USB-serial and virtualization stack, and this article works the order: confirm which side crashed, capture the stop, record the version stack, test the documented Workstation pairing for your Windows build, then the adapter driver, and only then anything at the controller.

Why does my Windows 10 host blue-screen when a VM talks serial to the PLC? #

The crash lives in the host's USB and virtualization path, not in the PLC or the guest: USB passthrough still carries every byte through the host kernel, so the machine that draws the blue screen, the Windows 10 host here, owns the fault.

USB passthrough does not take the host out of the connection. The host kernel owns the physical USB controller, and the virtualization layer carries every transfer between that controller and the COM port the guest exposes to RSLinx. The adapter driver lives in the guest in this arrangement, so the guest handles the serial protocol while the host handles everything underneath it. Auto-configuration is not special code. It simply drives enough communication activity through the passthrough path to expose a fault that quieter operation hides, which is why the crash follows driver activity rather than any one setting.

The reported outcome agrees with that reading: upgrading Workstation Pro 14 to 17 ended the host crashes on this installation, and a second user described Workstation 14 nearly taking down a Windows 10 host while version 17 ran clean. Two anecdotes from two installations are not a compatibility certificate, but they point the same direction as the documented support pairing checked in the section below.

Is it the host or the virtual machine that crashed? #

Check which operating system drew the blue screen before you change any setting: if the whole PC reboots and the stop code appeared across the physical monitor outside the VM window, the host crashed, and in this case it did.

Read it off the screen and the reboot behavior. The physical monitor showing the blue screen, the whole PC rebooting, and your open applications gone afterward mean the host died. If instead only the VM window froze or reset while the host desktop stayed up, the guest is the machine to diagnose and its serial driver is the suspect.

The distinction decides everything downstream. A host crash moves the cable, the PLC settings, and the guest's RSLinx configuration to the end of the line, because none of them executes in the operating system that failed. A different VM-to-PLC failure behaves the opposite way: a guest that cannot reach the controller over Ethernet, for example through a NAT adapter, times out without crashing anything, and that non-crashing path has its own fix in fixing a VM remote PLC connection with a NAT adapter.

What does Page Fault in Nonpaged Area actually tell you? #

Page Fault in Nonpaged Area is the Windows kernel reporting that a driver on the machine displaying the message touched memory it could not find, so it names the host's software stack and says nothing about serial framing, baud rate, or the MicroLogix.

Capture before you change anything. Photograph the complete blue screen and write down any module or driver name it shows, exactly as displayed. Collect the crash dump and the matching Windows event records through your normal IT diagnostic process, and preserve them before another test overwrites the evidence. Record the last successful action too, whether adapter attachment, COM port opening, auto-configuration, first communications, or the going-online step, because the boundary between the last working step and the first crashed one is exactly what a fix has to move.

A module named on the screen is a lead, not a verdict: two drivers can fail together while the initiating fault sits in the older one. And once you hold a reproducible trigger and a captured dump, stop re-crashing the machine to prove the symptom still exists. Change one variable per test. In this case the trigger was recorded as the online transition, not auto-configuration, because manual configuration connected cleanly and the crash came later.

Which versions do you record before changing anything? #

Write down the full software stack first, because the fix decision is a version comparison: host Windows build from winver, Workstation Pro release, VMware Tools build inside the guest, guest operating system, and the USB serial adapter's driver provider, version, and date.

Read each value from the installed system, not from memory:

  • Host operating system: edition and build exactly as winver prints them, for example Version 22H2, OS Build 19045.3803.
  • Virtualization software: the Workstation Pro release number from the product's About dialog.
  • Guest side: the guest operating system and the VMware Tools version, both read from inside the guest.
  • Adapter: chipset family plus the driver's provider, version, and date, from the COM port's entry in the guest's Device Manager.
  • Reproduction trigger: the last successful step and the first crashed step, in order.
  • Comparison set: which stations crash and which do not. Here, the two newest Windows 10 desktops crashed and older stations on the same task did not.

Keep the record through every change, and log the changes separately: the Workstation upgrade and a VMware Tools update are two different changes, and a later failure can only be compared against the state that produced it.

Is VMware Workstation 14 supported on your Windows 10 build? #

Run the documented pairing check before any other fix: Broadcom's host-OS support article for Workstation lists Windows 10 1903 as a compatible host operating system from Workstation 15.x upward, so a 1903-or-newer host with Workstation 14 fails the pairing (14 < 15) and the upgrade is the first move.

Work it with your own winver numbers. Suppose the host reports Version 22H2, OS Build 19045.3803:

  1. Host build number: 19045.
  2. Documented floor: Windows 10 1903 hosts are supported from Workstation 15.x, so any build at 1903 or newer requires Workstation 15.x or newer.
  3. Installed release test: 19045 is at least 1903, so the rule applies, and 14 is below 15, so the installed Workstation 14 fails.
  4. Reported fix test: 17 is at least 15, so Workstation Pro 17 satisfies the pairing, closing a gap of 17 minus 14, or 3 major releases, of USB passthrough and driver work between the crashing version and the working one.

Decision: on any Windows 10 host at build 1903 or newer, stop troubleshooting a passthrough crash on Workstation 14. Upgrade to a release the support table lists for your build and retest before touching the adapter or the PLC. Version 17 is the release reported to fix this exact case, and the support pairing is what makes the upgrade defensible on your workstation; the anecdotes only tell you it worked twice.

Upgrade in the documented order. Shut the guest down cleanly and copy the virtual machine folder, the engineering project, and the communication configuration to safe storage. Confirm your license covers the target release, obtain it through the manufacturer's official channel, install, and restart when the installer requires it. Update VMware Tools inside each guest by the documented process for that release, then retest the exact failing sequence, same adapter, same cable, same controller settings, so the only changed variable against your version record is the Workstation release.

If your host build is not listed for any Workstation release you can license, the pairing stays broken no matter what you fix downstream, and the virtualization platform itself becomes the thing to change. Some shops move the engineering guests onto Proxmox for SCADA and PLC engineering virtual machines on supported hardware for exactly that reason.

Do you fix the VMware version or the USB serial adapter first? #

Fix the virtualization pairing first, because it is the documented failure with the reported cure, and test the adapter second: an incompatible chipset normally fails to connect without crashing the host, so a blue screen on a supported Workstation release points above the adapter at a driver interaction.

If the upgraded stack survives the going-online step, the adapter was never the blue-screen cause and you stop there. If the crash survives the upgrade, install the adapter maker's own driver package for your chipset and Windows build instead of the class driver Windows installs on its own, confirm the chipset from the COM port's entry in the guest's Device Manager, and retest.

Chipset choice is a reported field result, not documentation: adapters on Prolific chipsets reportedly fail to connect to Allen-Bradley serial ports on Windows 10, failing without crashing anything, while FTDI-chipset adapters reportedly connect. Read that as a reason to test, not a compatibility certificate. Once the host survives, a swap to an adapter on a different chipset is a legitimate diagnostic, and note the milder failure it produces: a connecting-but-dropping adapter costs you time, not the whole workstation. One adapter family's own host-crash case is worked through the same capture-verify-retest loop in this USB-serial BSOD fix and troubleshooting walkthrough.

Do you need to touch the PLC or the serial settings? #

Leave the PLC alone until the host stops crashing: manual COM-port selection established communications in this case and the going-online step still blue-screened the host, which proves the serial settings and the controller were never the fault.

Wrong framing, a wrong COM port, or a marginal cable produce timeouts and failed connections at the application level, not kernel memory faults in a different operating system. That asymmetry is why serial settings sit last here. Keep the manual configuration that connected: it is your reproduction trigger, and re-running it after each software change tests the change against a fixed baseline instead of adding auto-configuration back in as a second variable.

The auto-configure button is not the root cause either way. Other users run it routinely without a crash, and here a manual configuration that connected still crashed the host, so the button's presence at the first crash was a coincidence of communication activity, not a cause. Do write down the exact controller model and its port configuration, since MicroLogix 1000 and 1100 controllers were the identified targets here, because the model decides the cable and port settings any later valid test needs.

Going online to a running controller is a change to a live machine, so make each retest against a controller that is out of service or under your normal commissioning procedure, never one the line is depending on.

Once the upgraded host survives the online transition, the PLC and its serial settings return to normal commissioning, and not before.

When do you replace the workstation instead? #

Replace the host when the software path is clean and one machine still crashes where its equal-build twin survives: after a supported Workstation release, maker drivers, and current Tools have all been confirmed, a workstation that blue-screens while a sibling runs the identical guest and adapter is the failing component.

In this case the crash followed hardware, not just software: the two newest Windows 10 desktops crashed while older stations running the same guest, adapter, and task did not, which is the signature of a difference in the machines' driver and USB controller stacks rather than of a bad PLC. Work the software path to its end on the failing machine first, because a machine-specific driver conflict is still fixable there.

If one specific desktop still blue-screens after all of that while its sibling runs the same session cleanly, retire it rather than rebuilding it twice. The economical fallback is to move the serial programming station onto a fanless industrial mini PC dedicated to the engineering load, with the supported host OS and virtualization pairing documented from day one.

The decisions, in the order the checks hand them to you:

  • Windows 10 host at build 1903 or newer, Workstation 14 installed, host BSOD on passthrough: upgrade Workstation to a listed release, 17 reported here, and retest the same sequence.
  • Crash survives the upgrade: maker driver package for the adapter, chipset confirmation, then a different-chipset adapter test.
  • Crash gone after the upgrade: stop there; the adapter, cable, and PLC stay untouched.
  • One machine still crashes where its twin does not: replace that machine.

Questions asked

Will upgrading VMware Workstation delete or break my existing virtual machines?

No. The virtual machine files live in their own folder outside the program, and an upgrade does not touch them. Shut the guest down cleanly, copy the VM folder and your project files to safe storage, install the new release, and accept the hardware compatibility prompt on the first start. Update VMware Tools in the guest afterward, then retest the serial sequence.

Can I skip the virtual machine and run the programming software on the Windows 10 host directly?

Yes, when your software version and license support the host operating system. The host copy talks to the adapter's own COM port, which removes the passthrough layer from the path entirely and is the fastest test of whether the virtualization layer is involved. Keep the guest for packages that require the older operating system, and never run the host copy and the guest copy against the same adapter at the same time.

Where is the crash dump, and what should I do with it?

Windows writes a minidump to the Minidump folder inside the Windows directory after each blue screen. Hand it to your IT diagnostic process: a kernel debugger names the module it holds probable, and you match that module against the installed driver versions in your record. The dump supplements the version stack you read off the systems, it does not replace it.

How do I find the right driver for a Prolific or FTDI USB serial adapter?

Read the chipset from the adapter's COM port entry in the guest's Device Manager: the Details view lists the hardware identifier, which names the chipset family. Download the driver package for that chipset and your Windows build from the adapter maker's own support site instead of letting Windows install a class driver. Field reports favor FTDI-chipset adapters over Prolific ones for Allen-Bradley serial work, so treat that as a test result, not a guarantee.

Keep reading

  • Troubleshooting

    Fixing a VM remote PLC connection with a NAT adapter

    Set the guest's virtual adapter to NAT with DHCP, disable every other guest adapter, and keep the remote-access soft key on the host. The failure that repeats is a second bridged adapter pulling the PLC route away from the NAT path. Test the host path first, then the guest, and check that no local subnet matches the plant network.

    Remote access and virtualization

  • Troubleshooting

    Fixing slow automation VMs on a Windows 11 host

    The usual cause is the Microsoft hypervisor layer under VMware, confirmed by the turtle icon in the status bar. Turn off Memory Integrity, disable Virtual Machine Platform and Windows Hypervisor Platform, run bcdedit /set hypervisorlaunchtype off, reboot, and rerun the same workload — with IT approval on a work machine. Guest memory is ruled out: 16 GB did not help, and working setups run smaller allocations.

    Remote access and virtualization

  • Checklist

    Proxmox for SCADA and PLC engineering virtual machines

    Yes — Proxmox runs Siemens SCADA and PLC engineering VMs; field teams use it for TIA Portal engineering over RDP and in-cabinet dedicated hosts. Commit per workload: pass the support-position, Layer 2 bridge, license and restore tests below before production, and size the host from measured demand.

    Remote access and virtualization