1. Problem Description
A SINUMERIK 840D sl controller equipped with a PCU 50.3-C begins a normal boot sequence and the operating software runs for approximately 10 minutes. The two-digit hexadecimal 7-segment diagnostic display on the PCU then freezes on the value 38 (or, depending on the BIOS build and PCU sub-variant, 88) and the front-panel status LEDs H1 and H2 remain lit simultaneously. A power-cycle returns the unit to normal operation and the fault recurs after a similar warm-up interval. The fault is repeatable, time-delayed, and thermally correlated, which is a diagnostic signature very different from a hard boot failure.
Symptom summary:
- 7-segment POST display latches on a single value (38 or 88) and does not advance.
- H1 and H2 LEDs illuminated and not toggling normally.
- Fault appears after a 5–15 minute warm-up window.
- Fault clears on cold start; recurrence interval shortens as the PCU heats up.
- HMI application is in steady-state at the moment of fault (not during a specific operator action).
2. PCU 50.3 Hardware Architecture
The PCU 50.3 is the industrial PC that hosts the SINUMERIK Operate (or earlier HMI-Advanced) software stack and provides the operator-panel interface for the 840D sl NCK. The PCU 50.3-C variant typically ships with an Intel Core 2 Duo or equivalent mobile-class processor, 2–4 GB of DDR2/DDR3 RAM, a CompactFlash (CF) card for the operating system image, and an optional 2.5" SATA hard disk for shop-program storage and trace data. The front-panel diagnostic elements consist of:
- A two-digit 7-segment display used to surface Phoenix BIOS Power-On Self-Test (POST) codes during the early boot phase. Industrial 7-segment modules of this type follow the standard seven-segment encoding with decimal point; Siemens uses these as a bare-bones diagnostic bus because there is no VGA output before the HMI initializes. Reference: Broadcom 7-segment display portfolio for the device physics of the LED module used.
- Two front-panel LEDs labeled H1 and H2. On PCU 50.3-C, H1 normally indicates CompactFlash / mass-storage activity, and H2 indicates system status or secondary storage activity. The exact mapping is firmware-defined; consult the SINUMERIK 840D sl PCU 50.3 Base Software documentation for the build you have installed.
- A power button, a reset button, and (on -C variants) a service-mode rotary switch.
Refer to the SINUMERIK 840D sl PCU 50.3 manual in the Siemens Industry Online Support portal (support.industry.siemens.com) for the exact hardware revision of your unit and the BIOS version reported by the POST display.
3. Phoenix BIOS POST Code System
The PCU 50.3 family ships with Phoenix BIOS 4.0 Release 6.x, a legacy modular BIOS that surfaces each initialization step as a single-byte hex value on the POST port. The two-digit 7-segment display is driven by the same POST port (typically I/O port 80h), so any code visible on the display is a Phoenix 4.0 R6 POST code.
POST codes are issued in a defined sequence during cold boot. If the system halts at a particular code, the BIOS has stalled on the routine associated with that code. The two relevant values in this case are:
| POST Code | Phoenix 4.0 R6 Phase (typical) | Subsystem Under Test |
|---|---|---|
38 |
Early POST (pre-memory or chipset bring-up on some OEM builds) | On PCU 50.3-C, code 38 in field reports typically maps to chipset register initialization or interrupt-controller probe. Exact meaning is OEM-customized; verify with the BIOS build string on the splash screen. |
88 |
Late POST (BIOS data area) | Initialization of the BIOS data area (BDA) at segment 40h, including keyboard buffer, equipment flags, and INT 13h/INT 19h vectors. A hang at 88 usually indicates a corrupted BDA, an extension ROM that has overwritten BDA, or a RAM timing fault that the late POST stresses. |
38 and sometimes 88. The difference is normally explained by (a) a different BIOS build, (b) the unit displaying the last POST code on halt versus the code at which it actually stalled, or (c) a transient code that latched the display driver. Treat both values as equivalent thermal-failure symptoms and proceed with the same diagnostic tree.
4. H1 and H2 LED Status Matrix
The PCU 50.3-C front-panel LEDs are multiplexed indicators; the meaning of "both on" depends on the BIOS / HMI state machine. Use the following matrix:
| System State | H1 (CF/Storage A) | H2 (Storage B / Status) | 7-Segment Behavior |
|---|---|---|---|
| Normal POST, advancing | Brief blink or off | Brief blink or off | Rapidly cycling codes (02, 04, 06…) |
| POST stalled at code 38/88 | Solid on | Solid on | Latched, no change |
| HMI loading (SINUMERIK Operate) | Blinking with disk activity | Solid on or slow blink | Blank or shows "donE" / "OS-" |
| Normal operation | Activity blink | Solid on | Blank |
| Thermal fault (this case) | Solid on | Solid on | Latched on 38/88, no advance |
If the unit is in the latched H1+H2-on / code-38-88 state and does not respond to the HMI keyboard or panel, the BIOS has halted and the HMI service is not running. The condition is recoverable only by a hard reset.
5. Root Cause Analysis: The 10-Minute Thermal Pattern
A fault that requires 5–15 minutes of operation to appear, clears with a cold start, and shortens its recurrence interval as the chassis warms is, in field service experience, almost always one of the following four mechanisms. They are listed in order of probability based on documented PCU 50.3 failure data and general industrial-PC repair experience.
5.1 Solder-Joint Fracture Under the BGA Chipset
Repeated thermal cycling (machine on at start of shift, off at end of shift) causes the lead-free BGA solder balls under the northbridge, southbridge, or graphics chip to fracture (the well-known chip-cracking / RoHS BGA failure mode). At room temperature the joint is partially conductive enough to function; once the board reaches 60–70 °C, the open circuit becomes a true open and POST halts.
Diagnostic sign: identical warm-up time on a bench test as in the cabinet; fault occurs faster if you pre-warm the chassis with a heat gun held 30 cm above the board for 60 s.
5.2 DRAM Module Thermal Marginality
The PCU 50.3 has one or two SO-DIMM slots populated with DDR2/DDR3. A single bit-line marginality in one of the modules passes the cold POST memory test but fails the in-band refresh tests that the HMI performs after boot. POST halts when the BIOS re-enters SMM or when the memory controller retries a transaction and times out. The reported code being 88 (BDA init depends on low memory) is consistent with this.
Diagnostic sign: removing one SO-DIMM (or seating the module in the other slot) shifts the symptom or clears it.
5.3 Mass-Storage Read Failure (CF or HDD)
The CompactFlash card hosts the Windows Embedded / Windows 7 image. After 10 minutes the HMI executes scheduled disk operations (e.g., the connection-collector writing servplus.log, alarm log rotation, or trace-buffer flush) that hit a marginal block on the CF card. The HMI issues a call to INT 13h / Atapi that re-enters the BIOS data area; the BIOS detects the corrupted BDA or an unhandled interrupt and halts. H1 (CF) being solid-on at the moment of fault, with H2 also on, supports a storage bus-lockup interpretation.
Diagnostic sign: servplus.log stops writing at the same minute the fault appears; chkdsk reports bad sectors on the CF card.
5.4 Electrolytic Capacitor ESR / Bulk-Cap Degradation
The bulk capacitors on the PCU 50.3 power-plane section age with temperature. As the chassis warms, the ESR of one or more caps rises and the local rail (often the 1.05 V or 1.8 V core rail) sags under load. The chipset re-asserts reset, the BIOS re-runs, and the unit hangs at the same code every warm-up.
Diagnostic sign: an oscilloscope on the suspect rail shows a 30–80 mV ripple increase after 8 minutes of operation; this is the most diagnostic test you can perform.
6. Diagnostic Procedure
Run the steps below in order. Each step is a single-variable test; only one variable changes between steps so the cause can be isolated.
- Capture the symptom on video. Record the 7-segment value, both LEDs, and the time-since-boot when the fault occurs. Ambient cabinet temperature should be noted.
- Pull the PCU and bench-test it. Place the unit on an ESD mat, connect a known-good SINUMERIK panel or a stand-alone monitor, and time the fault at 22 °C. Compare to the in-cabinet warm-up time.
- Pre-warm test. Use a hot-air station on low (350 °C, 30 cm distance, gentle sweep) to bring the board to ~50 °C in under a minute. If the fault appears within 30 s of heating, the failure is a thermal BGA or capacitor joint, not a logic fault.
- Reseat the SO-DIMM(s). Remove both modules, clean contacts with isopropyl alcohol, reinstall. If the fault changes pattern, replace the SO-DIMM with a known-good module of identical PC2/PC3 rating.
-
Swap the CF card. Image the original CF card with
ddto a spare, run the bench test with the spare. If the fault clears, the original CF card has bad blocks; replace it and re-image from the SINUMERIK service stick. - Disconnect the HDD data cable. On PCU 50.3-C with both CF and HDD, run with HDD power disconnected. If the fault clears, the HDD or its power rail is the cause.
- Measure rail voltages at the inductor closest to the chipset. Use a 10x probe and a scope with AC coupling. Acceptable ripple is below 25 mV pk-pk. A ripple above 50 mV pk-pk that grows with warm-up time confirms bulk-cap degradation on that rail.
- Run the SINUMERIK diagnostics suite. From the HMI service menu (password-protected), launch the PCU diagnostics. Look for the Temperature and Memory Test sub-tests. Memory test must complete three full passes at elevated temperature.
7. POST Code 38 / 88: Branch Tree
Use the following decision table to select the next repair action based on the bench-test result.
| Observation | Most Likely Cause | Repair Action |
|---|---|---|
| Fault appears faster when board is pre-heated; clears with reflow of chipset | BGA fracture | Reball or replace the chipset (northbridge / southbridge / graphics). Requires BGA rework station. Siemens part numbers vary by PCU 50.3-C sub-revision; order by serial number from Siemens spare-parts. |
| Fault clears when one of two SO-DIMMs is removed | SO-DIMM module | Replace the SO-DIMM. Use only modules validated for the PCU 50.3 (see Siemens spare-parts catalog). |
| Fault appears only when HDD is connected and powered | HDD or its 5 V / 12 V rail | Replace HDD; inspect SATA power connector for burned pins. |
| Fault cleared by CF swap; original CF has bad blocks | CF card wear | Replace CF card. Use only industrial-grade SLC CF cards; MLC cards in this duty cycle typically fail in 18–36 months. |
| Ripple on 1.05 V or 1.8 V rail grows past 50 mV pk-pk after 8 min | Bulk capacitor ESR | Replace the affected capacitor(s) on that rail. Typical values: 330 µF / 2.5 V polymer, 470 µF / 6.3 V, or 220 µF / 16 V. Match the original footprint and voltage rating. |
| None of the above, fault persists with minimum configuration | Mainboard | Replace the PCU 50.3 mainboard (top-level spare part). Re-image the system from service media; restore the CF card image; re-license SINUMERIK Operate if needed. |
8. Component-Level Repair and Replacement
Where the field reports the failure mode is recoverable without a mainboard swap, the most common field-repair operations and the safety / procedural notes for each are listed below.
8.1 SO-DIMM Replacement
- Power off the PCU, disconnect from the NCU and mains, wait 60 s for rail discharge.
- Open the service cover; release the SO-DIMM latches at 30° angle.
- Insert the replacement at the same angle; press down until both latches engage.
- Run the SINUMERIK service memory test (three full passes).
8.2 CF Card Re-imaging
Use the SINUMERIK service stick (USB) that ships with the option PCU-Basesoftware. Boot the PCU into service mode (set the rotary switch to position 2 on -C variants) and follow the on-screen prompts. Re-image time is typically 8–12 minutes. After re-image, restore the machine-specific archives from the backup made before the failure window.
8.3 BGA Reflow / Reball
A controlled BGA reflow is a valid field repair for confirmed BGA-joint failures. Profile the board to peak 235 °C for 45 s, ramp 1.5 °C/s, cool at <3 °C/s. Use a preheater underneath; top-only heating of a populated PCU 50.3 board will warp the PCB. Validate the repair with the bench-test pre-warm step; the unit must run 30 minutes at 50 °C ambient without halting.
8.4 Bulk Capacitor Replacement
Identify the rail by schematic (not by visual buldge — modern polymer caps rarely bulge even when failed). Replace the lowest-ESR equivalents in the same footprint. After replacement, re-measure ripple; the new rail ripple should be <15 mV pk-pk. A failed rail that returns to <15 mV pk-pk and remains so for 30 minutes of warm-up is repaired.
9. Verification and Burn-In Testing
After any repair, run a structured burn-in to confirm the fault is gone. The burn-in must reproduce the original thermal load.
- Reinstall the PCU in the cabinet and reconnect all cables.
- Boot to SINUMERIK Operate. Confirm normal HMI.
- Close the cabinet door and run the machine in normal production mode (or simulated) for 4 hours.
- Monitor the 7-segment display continuously; it must remain blank (or show the operating-system load code) and must not latch on 38 or 88.
- Run the SINUMERIK service diagnostics again, including the elevated-temperature memory sub-test.
- Log the cabinet interior temperature at 30-minute intervals. The repair is considered verified only if the unit survives the full 4-hour burn-in at the highest temperature the cabinet reaches in production.
10. Preventive Maintenance
The PCU 50.3 is a serviceable industrial PC, and the failure mode documented here is the predictable end-of-life behavior of certain sub-components. To push the next failure as far into the future as possible:
- Replace the CF card on a 36-month cycle, regardless of test results. Use only industrial-grade SLC media.
- Clean the cabinet filter and confirm the PCU cooling fan (if fitted on your sub-variant) is operating. The fan rated life is 40,000 h at 25 °C; replace at 35,000 h.
- Record the 7-segment display state on every scheduled service. A value other than blank or the OS-load code at any time is an early warning.
- Keep the PCU 50.3 BIOS at the latest service-pack level approved by Siemens for the machine builder. BIOS updates frequently include POST code table corrections and chipset microcode patches that address the thermal-marginality class of fault.
- Verify that the cabinet interior temperature stays within the PCU 50.3 specification. The unit is rated for operation at 5–55 °C with the standard convection cooling; if the cabinet runs hotter, derate the expected service life accordingly.
11. Notes on Documentation and Support
The Phoenix BIOS 4.0 Release 6 POST code list is a published industry-standard reference. The Siemens-specific PCU 50.3 documentation set (Base Software manual, Service Manual, Spare Parts catalog) is the authoritative source for the LED map, BIOS build string, and approved replacement part numbers for any given PCU 50.3-C sub-revision. Always quote the BIOS build string (visible on the splash screen during POST) and the PCU serial number when contacting Siemens support. The Siemens Industry Online Support portal at support.industry.siemens.com is the entry point for SINUMERIK 840D sl documentation, firmware downloads, and service notifications. Cross-reference the Phoenix 4.0 R6 POST table against the Siemens-specific overlay in the PCU 50.3 Base Software manual to confirm the exact subsystem under test for the code you observed.
What does POST code 38 mean on a Siemens PCU 50.3?
POST code 38 on a PCU 50.3 indicates that the Phoenix BIOS has halted during an early initialization step. On Phoenix 4.0 R6 the code sits in the chipset / interrupt-controller phase; on Siemens-specific builds the exact subsystem is OEM-defined and may map to chipset register initialization or the early memory probe. Treat code 38 and code 88 as members of the same fault class — they both indicate a BIOS-level halt before the HMI loads.
Why do the H1 and H2 LEDs both stay on when the fault occurs?
On PCU 50.3-C, H1 and H2 are storage-activity and system-status LEDs. When the BIOS halts before reaching the HMI, the HMI service never overrides the LED defaults, and the BIOS leaves both indicators latched in the "on" state. A latched H1+H2-on condition with a frozen 7-segment value is a reliable visual signature of a BIOS-level halt.
Why does the fault appear only after 10 minutes and not at cold start?
A fault that needs a warm-up window and clears on a cold start is almost always a thermal-marginality failure. The most common mechanisms on a PCU 50.3 are BGA joint fracture under the chipset, a marginal SO-DIMM module, a degraded bulk capacitor on a low-voltage rail, or a CF card that develops read errors as the media temperature rises. Each of these is a single-variable test away from identification.
Can I keep running the machine with this intermittent fault?
No. The fault is a BIOS halt, not a recoverable HMI error. When it occurs the HMI drops, the NCK / drive communication is suspended, and any axis in motion will be brought to a controlled stop by the drive firmware. The risk is a stop in mid-cycle that damages the workpiece, the tool, or the spindle. Treat the fault as a hard fault and schedule the PCU for service at the next opportunity.
Is it cheaper to repair the PCU 50.3 mainboard or replace the PCU?
For BGA or bulk-cap failures on an otherwise serviceable board, a component-level repair is typically 20–40% of the cost of a replacement PCU and preserves the machine-specific license and image. For a multi-fault board or a board where the chipset itself has died, replacement is the correct call. Always re-quote both options with the actual Siemens spare-part part number from the SINUMERIK 840D sl PCU 50.3 spare-parts catalog before committing to a path.