Siemens SINUMERIK 840D Alarm Codes: How to Read the Number and Find the Faulty Part

A SINUMERIK alarm number is not a random ID. The block it falls in tells you which part of the system raised it, and that is usually enough to know whether you are looking at the NCU, the drive rack, an encoder, or a piece of logic your machine builder wrote fifteen years ago. Learn the bands and you can point at the right cabinet door before anyone opens a manual.

This guide covers how SINUMERIK 840D numbers its alarms, which bands matter, the alarms that actually turn up on machines that have been running since the early 2000s, and where each one usually ends: a setting, a connector, or a module that needs replacing. If you already know what failed and just want the part, our SINUMERIK 840D spare parts guide covers ordering.

The alarm number tells you who is complaining

SINUMERIK splits its alarm range into blocks by source. Siemens owns some of those blocks. Your machine builder owns others. Knowing which is which saves an hour of searching for text that Siemens never wrote.

Read the cancel criterion before you read the text

Every SINUMERIK alarm carries a clear condition, and it tells you how bad the situation is before you have understood a word of the message.

• Cancel key clears it. Informational. Something was rejected, nothing is broken.
• RESET clears it. The channel stopped. Usually a program, a limit or a monitoring function that tripped.
• POWER ON required. The NC or a drive has to be restarted to clear the condition. This is where hardware faults live.
• NC START inhibited with no clear route out. Something upstream has not released, often the PLC or an interlock.

A POWER ON alarm that comes straight back after the restart is a hardware fault until proven otherwise. That is the point where you stop cycling power and start measuring.

The first alarm is the real one

The alarm on screen is often the third or fourth consequence of something that happened a second earlier. Open the alarm list under the Diagnosis area and read from the bottom, oldest first. An axis that loses its encoder will throw a hardware alarm, then a following error alarm, then a channel stop, then a PLC interlock, and the operator will only ever quote you the last one.

On machines with the alarm log enabled, the timestamp resolution is good enough to see the order clearly. If it is not enabled, turn it on. It costs nothing and it pays for itself the first time an intermittent fault shows up on night shift.

The alarms you will actually meet on an aging 840D

4060: standard machine data loaded

This one puts machines out of service for days, and it is almost always a dead NCU battery discovered too late. The NC lost its retentive data and came back with defaults. The machine will move, badly, or not at all, and every axis is misconfigured.

There is exactly one good outcome here, and it is a current series machine start-up archive restored from a USB stick or the card. If nobody has one, you are rebuilding machine data from the builder's documentation, which is a specialist job measured in days. Take the archive from every 840D machine you own this week, before you need it. The NCU battery is a consumable and it does not warn you politely.

25000 and 25001: encoder hardware fault

Alarm 25000 is a hardware fault on the active encoder of an axis, 25001 is the same on the passive one. Machines with a direct scale plus a motor encoder have both, and which number appears tells you which measuring system to go look at.

Before condemning the encoder: check the connector at the motor, check the shield clamp in the cabinet, and check whether the cable runs in a drag chain. Motion cables in drag chains break internally after a few million cycles and the fault is intermittent long before it is permanent. We keep Siemens pre-assembled Motion-Connect cables in stock for exactly this reason, because a cable is a far cheaper first swap than a motor.

25010 and 25020: pollution and zero mark

Pollution of the measuring system means the scale or the encoder disc is dirty or the signal amplitude has dropped. On machine tools this is coolant ingress nine times out of ten. Zero mark monitoring failing on an axis that used to reference fine points the same way. Both are worth a clean and a reseal attempt before you buy anything, and both come back within weeks if the ingress path is still open.

25040, 25050 and 25030: the axis is not going where it was told

Standstill monitoring (25040) trips when a stopped axis drifts out of its window. Contour monitoring (25050) trips when the difference between commanded and actual position exceeds the tolerance during motion. Actual velocity alarm limit (25030) trips when the measured speed goes past what the machine data allows.

These three are mechanical or drive faults far more often than they are electronic faults. A sticking way, a failing ballscrew bearing, a brake that is not fully releasing, or a drive that is current limited will all produce them. Check the load before you check the module. If the axis is hard to move by hand with the brake released and the drive disabled, the answer is in the machine, not in the cabinet.

25201 and the 300000 band: the drive itself

Alarm 25201 is the NC saying the drive reported a fault. The useful detail is in the 300000 range alarm that comes with it, which names the axis and the drive number, and in the modules themselves. Walk to the cabinet and read the LEDs and the display on the SIMODRIVE 611 rack. The infeed module tells you about DC link and mains, the axis control modules tell you about the individual drives.

What fails on a 611 rack in practice: the control modules, then the infeed, then the power modules. We keep tested SIMODRIVE 611 spares including 6SN1118-0DM33-0AA2 control modules, the 6SN1118-1NH01-0AA0 digital control module and 6SN1123-1AA01-0FA1 power modules. For the diagnosis side, our SIMODRIVE 611 troubleshooting guide goes deeper than this page can.

On 840D sl machines the drives are SINAMICS instead, and the fault numbering changes to the F and A codes. Those are covered in our SINAMICS fault codes guide.

20000 and 10203: referencing

Reference cam not reached (20000) and NC start without reference point (10203) are the two you meet after any power loss, any encoder swap, and any battery incident. On an incremental system the machine simply does not know where it is yet.

If referencing worked yesterday and fails today, suspect the cam switch, its wiring, or the approach speed set in machine data. If it fails only on one axis after you replaced something on that axis, you have almost certainly changed a direction, a resolution or a reference mark setting.

21610: encoder frequency exceeded

The encoder signal came in faster than the NC can count, which physically means either a genuinely overspeeding axis or a corrupted signal. On an axis that was moving slowly at the time, this is a signal integrity problem: shielding, connector, or a scale head starting to fail. It is also a classic symptom of a broken shield in a drag chain cable.

When the alarm really is a dead module

Most alarms end in a setting, a cable or a mechanical problem. A minority end in a module that has to be replaced. The ones we see most often on 840D machines, and what we hold for them:

• NCU. The machine is dead, will not boot, or throws NCK system errors that survive a restart. We stock the 6FC5210-0DA21-2AA1 840D sl NCU, tested.
• PCU. The HMI side is gone, black screen or Windows failure, while the NC underneath is often still alive. We hold the PCU 50 6FC5210-0DF21-2AA0 and the MMC103 6FC5210-0DA20-2AA1 for older systems.
• Operator panel. Backlight failure, dead touch layer, or failed keys. Available now: OP012 front panel 6FC5203-0AF01-0AA0 and OP010 6FC5203-0AC00-0AA2.
• Machine control panel. Feed override or key matrix faults that show as PLC alarms rather than NC alarms. We stock the MCP 483 6FC5203-0AF22-1AA0.
• Axis expansion. On 840D sl, the SINUMERIK NX10 6SL3040-0NC00-0AA0, currently three units on the shelf.
• Servo motor. When the encoder inside the motor is the fault and the motor is worth replacing whole. Our 1FK7 stock includes the 1FK7042-5AF71 SIMOTICS S, five available.
• I/O. Peripheral faults that show up as PLC alarms, handled by the 6FC5111-0CB02-0AA0 module.

Every SINUMERIK part we ship is tested, comes from EU stock and carries a 30 day warranty. Full range in the CNC controls collection.

Quick answers

Where do I find the full SINUMERIK alarm text?

The Diagnostics Manual for your specific SINUMERIK version, available from Siemens Industry Online Support. The manual for 840D and the manual for 840D sl are separate documents and the numbers do not fully overlap, so pick the right one. On the control itself, the help key on the alarm line brings up the same text and the clear condition.

Why can I not find any information about my alarm number?

Check whether it starts with a 7. Alarms from 700000 to 799999 are defined by the machine builder, not by Siemens, and the text comes from an alarm file in the PLC project. No Siemens manual and no forum will have it. The machine documentation, the builder's service department, or the alarm text file in the project are the only sources.

The alarm clears on RESET and comes straight back. What does that tell me?

That the condition is still present rather than latched. On a monitoring alarm in the 25000 band this usually means a real mechanical or drive problem is sitting there waiting, not a one-off glitch. Do not keep resetting it. Find what is different about that axis.

Can I swap an NCU without losing my machine data?

Only if you have a current series start-up archive and the correct NC card handling for your system. The machine data lives with the NCU and its card, not in the new module you bolt in. Make the archive first, always, even when you are convinced the old unit is completely dead. Our 840D spare parts guide covers the swap sequence.

Is a used SINUMERIK module a sensible replacement?

For hardware this old it is often the only sensible one. Many 840D modules are long out of production, and an OEM route means either a repair with a lead time or a migration project. A tested used unit of the correct MLFB and version drops in and the machine runs. What matters is that the version letter matches, that the unit was tested, and that the seller takes it back if it does not fit. More on that in our used industrial parts buying guide.

A diagnostic order that works

1. Write down the full alarm number and the exact text, including the axis name in the message.
2. Open the alarm list and find the oldest entry in the group. Work on that one.
3. Use the number band to decide where you are going: NC, drive, PLC, or builder logic.
4. If it is a 700000 alarm, go to the machine documentation and stop searching elsewhere.
5. If it names an axis, go to that axis in the cabinet and read the drive module LEDs before touching anything.
6. Check connector and cable first, especially anything in a drag chain, then the mechanics, then the module.
7. Take a series start-up archive before you replace an NCU or a card. No exceptions.
8. Quote the full MLFB including the version letter when you order the replacement.

Related reading on Samonde

• SINUMERIK 840D spare parts guide: NCU, PCU and OP panels
• SINAMICS fault codes: common causes and fixes
• SIMODRIVE 611 troubleshooting
• When a SIMODRIVE 611 power module dies
• Heidenhain encoder problems: error messages and fixes
• Matching a Heidenhain encoder by Id.Nr.

840D machines are still earning their keep in shops all over Europe, and they will keep doing it as long as somebody can read the alarm line and get the right module on the next van. Learn the bands, keep a start-up archive, check the cable before you blame the module, and keep one spare of whatever stops your production line. That is most of the job.