Resolving SINUMERIK 840D SL PCU50.3 Startup Alarm Issues

1. Problem Overview

The SINUMERIK 840D SL operator panel often uses a PCU50.3 (Panel Control Unit) running a Windows Embedded Standard image to host the HMI software. On commissioning or after an unplanned power event, technicians frequently observe persistent boot-time alarms that prevent the controller from reaching the run state until the operator manually clears them on every power-up. The affected machine will not initialize the spindle, drives, or PLC until the alarm buffer is acknowledged, while identical machines on the same shop floor boot directly to the run state without intervention.

The alarms are not random. They cluster around three failure surfaces:

• Windows Embedded boot path: corrupted Boot Configuration Data (BCD), missing bootmgr, or a failed Startup Repair pass.
• PCU50.3 hardware: degraded 24 V DC power supply, bulk capacitor wear on the main PCB, or backplane communication faults.
• Operator procedure: hard power removal (breaker pull) instead of a controlled HMI shutdown, which leaves the NTFS journal dirty and forces a chkdsk/autorun on next boot.

Field data from the supporting case shows the root cause was a combination of improper shutdown by the operator and a failing PCU50.3 main PCB / PSU assembly. After repair of the power supply and main PCB and re-training on the shutdown sequence, the machine booted cleanly.

2. PCU50.3 System Architecture

The PCU50.3 sits between the NCU (Numerical Control Unit) of the SINUMERIK 840D SL and the operator panel (OP/TP/TCU). It is responsible for:

• Hosting the Sinumerik Operate HMI runtime (or older HMI Advanced).
• Providing file services for NC/PLC archives, tool data, and part programs.
• Bridging the NCU-PLC I/O field to the network for commissioning laptops.

The two-digit 7-segment LED is the primary visual diagnostic for boot-stage identification. Always note the displayed value at the moment an alarm appears on the HMI; the code often pinpoints whether the failure is in POST, bootloader, Windows kernel, or HMI runtime.

3. Root Cause Analysis

Three root causes must be ruled out in this exact order before swapping hardware.

3.1 Improper Shutdown Sequence

Hard removal of 24 V DC (or upstream breaker opening) while Windows is still writing to %SystemDrive% leaves the NTFS USN journal inconsistent. On next boot, Windows enters autocheck / chkdsk, the HMI starts late, and the HMI-level watchdog in the NCU raises an alarm such as:

• Alarm 410100 "HMI watchdog expired"
• Alarm 802701 "PLC HMI connection failure"
• Alarm 120000 "Wait for HMI connection"

3.2 PCU50.3 Power Supply and Main PCB Failure

Bulk electrolytic capacitors on the PCU50.3 main PCB age. A six-month-old machine that suddenly starts throwing boot alarms after a stable period almost always points to one of:

• DC-DC converter on the main PCB out of regulation (5 V or 3.3 V rail droop).
• Capacitor ESR rise causing CPU reset loops during Windows kernel init.
• 24 V DC input connector oxidation causing intermittent contact.

Symptom signature: alarms appear even after a clean shutdown, multiple machines on the same input bus show different behavior, and the 7-segment LED flickers or rolls digits erratically.

3.3 UPS / AVR Misconfiguration

The supporting case shows an AVR (Automatic Voltage Regulator) was in the supply path but the customer was unaware that a small internal UPS was also wired to the PCU/NCU pair. Mismatched hold-up time between the PCU and the NCU causes one to survive a mains dip while the other resets, breaking the HMI link and producing the alarms. Either both controllers must share the same UPS output, or the PCU must be on raw mains while only the NCU is on UPS — never the inverse.

4. Pre-Diagnostic Inspection Checklist

Run through this list before opening the PCU. Each item takes under two minutes and rules out one entire failure domain.

1. Confirm machine age and history. Six months in service with a recent NC/PLC archive restore is a strong indicator of post-restore HMI binding drift. Restore should be re-flashed using the OEM service image, not the customer archive alone.
2. Inspect the AC upstream. The AVR/transformer output should stay within ±10 % nominal, and the neutral-ground bond should measure under 4 Ω (per IEEE 1100 / NEC 250 recommendations). A 4 Ω reading is acceptable; anything above 8 Ω indicates a loose bonding jumper.
3. Verify 24 V DC at the PCU input terminals with the operator panel backlight ON. Voltage must be ≥ 22.0 V DC under load. If it sags below 21 V, the 24 V supply is undersized for the OP + PCU + NCU combined draw.
4. Confirm independent 24 V DC feeds for the NCU, the PCU50.3, and the OP/TP/TCU. A common feed with a single inrush can collapse the bus during the PCU POST.
5. Verify the OP mounting topology. Is the OP installed directly on the PCU, or is it hung off a TCU on the panel backplane? TCU mode requires an additional power supply and PROFINET termination checks.
6. Physically re-seat the DVI/DisplayPort cable, the USB cable, and the Ethernet cable (X120) on the PCU. Captive screws on the DVI shell are a frequent cold-contact point.
7. Confirm the ground stud on the PCU chassis is torqued and bonded to the cabinet ground bar. The PCU uses chassis ground for ESD drain on the I/O area; a floating chassis can throw spurious alarms on operator touch.

5. Seven-Segment LED Diagnostic Procedure

The PCU50.3 carries two 7-segment displays on its side. Capture the LED value at each of the following instants:

1. Immediately after 24 V DC is applied (within 2 seconds).
2. At the moment the HMI splash appears.
3. When the alarm banner pops up.
4. After the operator presses the Reset / Clear alarm key.

Record the values with a smartphone video at 60 fps — the LED rolls quickly during POST and the offending code is often visible for under a second. Save the clip and correlate to the alarm number on the HMI.

6. Communication and Power Verification

Even when the PCU boots, the alarms are raised because the NCU and PCU cannot complete their handshake in time. The handshake rides over a single Ethernet cable (X120 on the PCU, X120 on the NCU) using a Siemens-proprietary protocol. Verify the link with the following measurements.

6.1 Physical Layer

• Cable: Cat 5e or higher, shielded, ≤ 30 m for the internal PCU-to-NCU link.
• Link LEDs: green link + amber activity must be lit on both ends within 5 seconds of PCU boot. If only one side is lit, replace the cable first.
• Connector seating: the PCU50.3 uses a latched RJ45 — confirm the latch is intact, not just inserted.

6.2 Supply Sizing (Case-by-Case)

The 24 V DC supply must be sized for the combined worst-case inrush of the NCU + PCU + OP. Use the standard DC power formula:

P_total [W] = (I_NCU + I_PCU + I_OP) × V_DC

For example, with a 3.5 A PCU, 2.0 A NCU, and 1.5 A OP at 24 V DC:

P_total = (3.5 + 2.0 + 1.5) A × 24 V = 168 W minimum supply

Apply a 30 % derating for inrush and aging: select a 220 W (≈9.2 A at 24 V DC) supply minimum. If the existing supply is borderline, the PCU may dip during disk spin-up, triggering the same alarms.

6.3 Ground Bond

Measure the chassis ground to cabinet ground bar with a 4-wire milliohm meter. A reading of 0.1–1.0 Ω is typical for a short, clean bond. A reading of 4 Ω — the value confirmed in the supporting case — is acceptable at the far end of the cabinet ground, but anything above 8 Ω should be re-bonded before further diagnosis. A loose ground bond is a hidden cause of intermittent communication alarms because the Ethernet shield drain returns through the chassis.

7. Windows Embedded Boot Recovery

The PCU50.3 ships with a Windows Embedded Standard 7 (or, on older builds, Windows XP Embedded) image. When the HMI fails to start cleanly, treat it with the same boot-recovery discipline as a desktop Windows machine, with the added constraint that the disk partition layout and the bootmgr location are Siemens-customized.

Refer to the official Microsoft Windows boot issues troubleshooting guide for the underlying steps. The five methods listed there apply directly:

7.1 Method 1 — Startup Repair

Boot the PCU from the Siemens Service USB stick (or a generic WES7 recovery media). Choose Repair your computer → Startup Repair. The tool will attempt to repair BCD and missing bootmgr references. Allow 15–30 minutes for the automated repair; do not interrupt power during this phase.

7.2 Method 2 — Repair Boot Codes

From the recovery command prompt, run:

bootrec /fixmbr
bootrec /fixboot
bootrec /scanos
bootrec /rebuildbcd

On the PCU50.3 the active partition is typically partition 2 (a small System Reserved 100 MB partition) and the Windows partition is partition 3. Use diskpart → list partition to confirm before issuing rebuildbcd.

7.3 Method 3 — Fix BCD Errors

If Startup Repair reports a BCD error that it cannot auto-resolve, rebuild the store manually:

bcdedit /export C:\BCD_Backup
del C:\boot\bcd
bootrec /rebuildbcd

Reboot and verify the HMI starts without alarms. If the HMI service (sinumerik_hmi.exe or the HMI Advanced equivalent) does not auto-launch, the BCD entry pointing to winload.exe on the dedicated Sinumerik partition is missing — restore from the OEM service image.

7.4 Method 4 — Replace Bootmgr

If bootmgr is reported as missing or corrupt, copy it from the recovery image to the system partition root. On WES7, the file is found in D:\Sources\boot.wim (or wherever the OEM has staged the WIM). Mount, extract, and place at the root of the active partition.

7.5 Method 5 — System Restore / Image Restore

The final recovery path is a full image restore from the OEM service stick. This will overwrite the HMI runtime and reload the controller's PLC/NC binding. Schedule this during a planned downtime window because the machine will be unavailable for 60–90 minutes.

8. Hardware Repair Procedure

If the boot recovery above is clean but the alarms still appear on cold start, the PCU main PCB or its 24 V DC sub-PSU has failed. The supporting case was resolved only after repair of the PCU power supply and main PCB.

8.1 Visual Inspection

1. Remove 24 V DC and wait 60 seconds for bulk caps to discharge.
2. Open the PCU chassis. Inspect the main PCB for bulging or leaking electrolytic capacitors, especially around the CPU VRM area and the 5 V / 3.3 V buck converters.
3. Inspect the sub-PSU PCB (a separate small board feeding the backplane fans and storage). Look for brown scorch marks around the input fuse and MOV.
4. Re-torque the chassis ground stud to 2.0–2.5 N·m.

8.2 In-Circuit Voltage Check

Power the PCU on the bench with a current-limited 24 V supply set to 5 A limit. Probe the following rails at the backplane connector with a true-RMS multimeter:

Any rail out of window under a 4 A load indicates a failed DC-DC converter on the main PCB. Replace the PCB; do not attempt component-level repair of the CPU VRM — the BGA reflow risk outweighs the cost of the replacement board.

8.3 Capacitor Health Check

For field technicians with an ESR meter, lift one lead of each bulk cap (≥ 470 µF / 25 V) near the input filter and check ESR. A healthy 1000 µF / 25 V cap reads under 0.10 Ω. Anything above 0.5 Ω on a 6-month-old board is a quality control flag; replace the cap set. Note that the original question asked how to check NCU/PCU capacitors — the answer is an in-circuit ESR check, but a board-level replacement is the OEM-preferred remedy.

9. Operator Shutdown Procedure

Once the hardware is repaired, the alarm returns unless the operator's shutdown sequence is corrected. Train the operator on the following steps and post the procedure at the operator station.

1. End the current program: press NC Stop and confirm spindle and axes are at standstill.
2. On the HMI, open the main menu and select Logout. Confirm any open program is saved or canceled.
3. From the user menu, select Shut down operating system. Wait for the on-screen confirmation that the system is safe to power off.
4. Watch the 7-segment LED on the PCU. It must settle to 0 or blank.
5. Wait an additional 30 seconds.
6. Open the cabinet and turn the main breaker OFF.

10. Verification Tests

After repair and procedure training, perform the following validation sequence to confirm the fix.

1. Clean cold boot: Power off via the proper sequence, wait 5 minutes, then apply 24 V DC. The HMI must reach the run state without any alarm banner within 90 seconds.
2. Alarm buffer check: Open Diagnostics → Alarm log. The log must be empty after a clean boot. If the alarms still appear, repeat Section 5 (LED capture) before any further hardware intervention.
3. Communication handshake test: From the service menu, ping the NCU at its default IP (typically 192.168.214.1 for NCU, 192.168.214.241 for the PCU). The round-trip time must be under 5 ms on a crossover link.
4. Spindle and drive test: Issue a small S-word and M3 from MDI. Spindle must reach commanded speed without drive alarm 300000-class events.
5. Power dip test: With the machine in Ready, drop the input mains briefly (under 200 ms) using the AVR bypass. The UPS (if present) must hold both the PCU and the NCU through the dip. If only the NCU survives, rebalance the UPS assignment per Section 3.3.
6. Improper shutdown test: For verification only — perform a controlled hard shutdown, then a clean cold boot. The HMI must come up in autocheck and recover. If it does not, the disk image is still damaged and a full image restore (Method 5) is required.

11. Preventive Measures

• Install a dedicated UPS sized at 1.5× the combined NCU + PCU + OP draw. Both the NCU and the PCU must share the same UPS output. Document the wiring diagram on the inside of the cabinet door.
• Label every shutdown path. Affix a laminated card next to the operator panel that lists the exact shutdown sequence from Section 9.
• Schedule annual 24 V DC health checks. Measure the rail under full load once per year. Capture the PCU 7-segment LED value at cold start and compare to the previous reading.
• Keep an OEM service image on a labeled USB stick in the cabinet so a recovery can be done without a service engineer on site.
• Verify ground bond annually. The 4 Ω reading captured in the supporting case is the upper limit — re-torque the bond if it climbs to 6 Ω or above.
• Avoid mixing AVR and UPS on the same input unless the UPS is downstream of the AVR. AVRs can generate switching noise that the UPS interprets as a line event and switches to battery unnecessarily.

12. Quick-Reference Troubleshooting Matrix