Overview
Collision control is the most safety-critical function in any CNC machining workflow. On Siemens SINUMERIK controls, the on-controller kinematic monitor handles real-time protection, but the off-line simulation environment determines whether collisions are caught before the part is ever loaded. Two simulation philosophies dominate the SINUMERIK CAM ecosystem: kinematic reconstruction in Proficam / Profikinematik (the "Profi"-line of CAM-simulation utilities) and NC-block-true replay in Real-PRO, often combined with ChekT-it B4 and the controller's own Virtual-NC HMI. This reference documents the architectural differences, the PFR-file workflow, retraction-plane-change monitoring, and the decision matrix an integrator should apply before commissioning a 5-axis cell.
Why Off-Line Simulation Is Not Optional
A 5-axis SINUMERIK program (e.g. SINUMERIK 840D sl with TCPC / ROT / MEAS transformations) carries more degrees of freedom than a 3-axis program. The CAM system emits a sequence of linear blocks (G1) that the controller interprets against a kinematic chain. Small rounding errors in the post-processor propagate through the inverse kinematics and can drive the tool shank into the fixture or the table trunnion. Field experience shows that the visual line of motion on the CAM screen is not a guarantee of safe motion on the real machine. Collision simulation must therefore replay the actual NC program, with the actual tool table, the actual machine zero offsets and the actual kinematic model.
Architectural Models: Profikinematik vs Real-PRO
| Attribute | Profikinematik / Proficam | Real-PRO |
|---|---|---|
| Simulation driver | Internal CAM kinematic engine; rebuilds motion from CAM-internal toolpath | NC-block-true replay; decodes the exact .MPF / .SPF files that the controller will run |
| Source of truth | CAM-side geometry + simplified kinematic | NV-archive and cycle data exported from the SINUMERIK |
| Cycle support | Generic cycle emulation; customised cycles require manual mapping | Native cycles from the controller's NV archive |
| Post-processor coupling | Coupled to the post-processor that emitted the file | Post-processor independent - any conforming post-processor is replayed |
| Plane-change retraction | Travel-limit monitoring is weak on inclined-plane retractions | Strong monitoring of travel limits during retraction from oblique planes |
| Optimised block replay | Cannot always replay the compressor / spline output the controller produced | Replays the controller's optimised output (look-ahead, COMPCAD, COMPSURF) |
The decisive engineering distinction is the data origin. Real-PRO sources its NV archive, cycle definitions, and machine data directly from the SINUMERIK, so the simulation is bit-equivalent to the runtime behaviour up to the limits of the kinematic solver. Profikinematik re-derives the motion from CAM-internal data; a passing Profikinematik simulation is therefore a necessary but not sufficient condition for collision-free runtime.
ChekT-it B4 and Virtual-NC Integration
The ChekT-it B4 workflow uses Virtual-NC, a Windows-hosted HMI that mirrors the SINUMERIK operator interface. Operators launch the simulation against the same program, tool table, zero offsets, and kinematic model that the real control will see. The procedure:
- Export the active
NVarchive (NV-NCK, NV-PLC, NV-DRIVE, NV-COM, NV-MM) from the controller through the SINUMERIK HMI's commissioning menu or viasc;in SINUMERIK Operate. - Import the archive into the Virtual-NC HMI; the same user-interface dialogs are exercised, so cycle parameters and protection zones are loaded identically.
- Run the NC program under the HMI in single-block or rapid-feed mode; observe the 3D window and the axis-actual-value trace.
- Any collision, protection-zone violation or kinematic-limit alarm reported by Virtual-NC will be raised identically on the real control.
Because Virtual-NC consumes the same NC variables ($AA_IM[X], $VA_POS[X], $AC_BLOCKNO) as the real kernel, simulation results are deterministic across the simulation host and the controller.
PFR File Management for Profisimulation
The .PFR file is the Postprocessor-File-Runtime artefact emitted by the CAM system. It binds the CAM-internal toolpath to the post-processor configuration and the kinematic descriptor used by Profisimulation. Failure to refresh the PFR file after any of the following changes produces a simulation that is silently out of date:
- Post-processor revision (new build, new vendor release).
- Machine kinematic descriptor (e.g. head change from
SWIVELtoTCARR). - Tool-table update on the controller side.
- Controller firmware upgrade that changes the NC language dialect.
- Changes to protection-zone or software-limit settings in machine data.
Retraction Programming on Inclined Planes
Multi-axis work frequently requires the tool to retract along the tool vector from an inclined surface, change the working plane (PLANE SPATIAL, PLANE PROJECTED, PLANE EULER), and re-engage. The critical sub-second window between lift and re-engage is where most 5-axis collisions originate. Profikinematik/Proficam reaches its limits here because:
- The CAM-internal toolpath uses a simplified representation of the rotary-axis interpolation; the actual control may add transition arcs via
SOFT,BRISK,COMPCADor spline interpolation. - The travel limits (
$MA_SW_LIMIT_MINUS,$MA_SW_LIMIT_PLUS) and the working-area monitoring (MD20150family) are evaluated by the SINUMERIK on the post-interpolated positions; a CAM-side check that does not honour these exactly is unsafe. - For swivelled heads, the vector to a stationary fixture is rotated by the kinematic transformation; the rotation is correct on the controller and may be approximated in Profisimulation.
Real-PRO, sourcing its data from the controller, inherits the same kinematic and software-limit evaluation. Travel-limit monitoring during retraction from oblique planes is therefore authoritative in Real-PRO and indicative in Profikinematik.
Data Integrity Across the CAM → Post → Control Chain
A frequent commissioning finding is that a CAM system emits a program that simulates green on Profikinematik but alarms on the first block on the machine. The root causes observed in the field are:
| Symptom | Likely root cause | Remediation |
|---|---|---|
| Alarm 10753 "Channel %1 block %2 transformation selection not possible" | Post-processor emits ROT/AROT but the active kinematic chain in machine data is not selected |
Re-export NV archive from the controller and re-run simulation in Real-PRO / Virtual-NC |
| Alarm 10770 "Channel %1 block %2 transformation error" | Singularity in the swivel kinematics (poles of the rotary axes) not handled by the post | Switch the post's singularity handling to match controller's MD20150 $MC_GCODE_RESET_VALUES
|
| Alarm 10651 "Channel %1 block %2 distance too small" | Feed compression (COMPCAD) inserts blocks the CAM-side did not emit |
Disable compression for the simulation run, or feed the optimised output from the controller into the simulation |
| Alarm 10750 "Channel %1 block %2 active tool offset not allowed" | Cutter-compensation (CUT2D, CUT3DFS) entry strategy differs between post and controller |
Align the cutter-comp table and entry/retract macros between post and controller |
For the full SINUMERIK alarm catalogue and reset strategies, consult the SINUMERIK 840D sl Diagnostics Manual and the List Manual (NCK alarms/messages).
NC-Block Simulation Methodology (Recommended)
The procedure that the integrator should follow on a 5-axis cell:
-
Author in CAM. Generate the toolpath, apply the standard
.KIT(cutter) andTOA(tool-offset) data, and emit the post-processed program. - Regenerate PFR. The CAM-side PFR is used only for visual review; it is not the input to safety-relevant simulation.
-
Export NV archive. Pull
NV_NCK,NV_PLC,NV_COM,NV_DRIVEfrom the SINUMERIK in service mode. - Re-simulate in Real-PRO / Virtual-NC. Drive the post-processed program through the controller-mirror solver, with the live tool table and protection zones.
-
Run the controller dry-cycle. With
DRY_RUNenabled,FEEDRATE OVERRIDEat 0 %, and the doors open, dry-run the program on the actual control. Compare the axis-trace against the Real-PRO trace. - Sign-off. Only after the three traces (Profikinematik, Real-PRO, dry-run) agree, and the operator has visually confirmed clearance at the slowest feed, is the program released for production.
Selection Criteria: When Each Tool Is Appropriate
| Scenario | Recommended | Justification |
|---|---|---|
| 3-axis prismatic milling, no swivel | Profikinematik acceptable as primary | Kinematic chain is trivial; toolpath re-derivation matches control output |
| 3+2 indexed 5-axis (pos-controlled swivel) | Real-PRO recommended | Cycle-start blocks PLANE + ROT; controller-side transform must be evaluated |
| Simultaneous 5-axis with spline blocks | Real-PRO mandatory |
CSPLINE / ASPLINE / BSPLINE cannot be re-derived in CAM |
| Tool with long shank (L/D > 5) in soft jaws | Real-PRO mandatory | Shank-deflection compensation (CUT3DFS, TOFF) is controller-resident |
| Verification of NC programs from external CAM customers | Real-PRO mandatory | Unknown post-processor quality, unknown kinematic model |
| High-mix shop, 5+ post-processors in active use | Real-PRO + Virtual-NC | Single authoritative source, low PFR-maintenance overhead |
Pricing vs Functional Justification
The price delta between Profikinematik and Real-PRO is substantial, often 3-5x per seat. The decision is therefore a risk-based one, not a feature-based one. A useful ratio is:
Risk-adjusted cost of a single collision > (Real-PRO licence cost + PFR/Real-PRO integration hours)
For a 5-axis cell machining turbine blades, blisks, or medical implants, a single collision typically costs more than the entire Real-PRO licence fleet. For a 3-axis jobbing shop, the Profikinematik investment is amortised quickly and the residual risk is low. The two halves of the market are therefore rational.
Commissioning Checklist
- Verify the controller is on a firmware version supported by the Real-PRO build. Reference the SINUMERIK support portal for the current compatibility list.
- Configure the post-processor to emit the same language dialect (ISO dialect mode 0/1/2/3) as the controller's
MD20150. - Align the cutter table between CAM and controller, including
STC(sign-of-cutter),TCO(tool-correction-orientation), and the shank geometry. - Establish a signed-off PFR-regeneration procedure and store the PFR in revision control.
- Perform the three-trace sign-off (Profikinematik, Real-PRO, dry-run) on the first article of every new part-number family.
- Document the residual collision risk for the operator; do not let marketing claim "zero collision" without qualification.
Verification Procedures After a Simulation Change
After any of post-processor update, PFR regeneration, controller firmware change, or new tool family:
- Re-run the part in Real-PRO with the new artefacts.
- Re-run a dry-cycle on the controller with
DRY_RUNand feed-override at 0 %. - Compare the
$AA_IMtraces block-by-block; deltas larger than 1 µm require a four-eyes review before release. - Spot-check with a single slow-feed cut (
G1 F100) on a soft material witness before releasing to the operator.
Edge Cases and Field-Proven Caveats
-
Multiple kinematic chains in the same program. When the part uses both a head swivel and a table swivel in different sections, only a controller-mirror solver evaluates both chains correctly against the active
$TC_CARR35[n]data. -
Non-linear kinematic descriptors. If the OEM has implemented a non-standard kinematic (
MD62571 $MCS_KIN_SUPPRESS_ALARM_MASK), confirm the Real-PRO kinematic descriptor matches; otherwise the simulation will be optimistic on the linear range and pessimistic on the non-linear range. -
Compressed G-code. Programs compressed by the controller's
COMPBLOCKare not the same as the original program. Replay the compressed file, not the original, when verifying production output. -
Tool-measurement cycles.
CYCLE971,CYCLE972, andCYCLE982emit motion that is not always visible in the CAM toolpath. Real-PRO replays them; Profikinematik often does not. -
Synchronised actions. Programs that use
IDS=...synchronised actions orASUProutines must be simulated with the PLC-side timers loaded into Virtual-NC; otherwise motion-coupled-with-PLC events are missed.
Comparison with Native SINUMERIK 3D Simulation
Modern SINUMERIK Operate releases ship with a built-in 3D simulation that decodes the same NC language the controller executes. For shops that do not invest in a third-party CAM-side simulation, this in-HMI simulation is the right baseline. It is, however, limited by the HMI's rendering performance and by the fact that it does not import the CAM-internal geometry (the workpiece blank, fixtures, clamps). For final sign-off on a 5-axis cell, supplement the HMI simulation with a CAM-side solver that can show the full part, fixture, and machine envelope.
FAQ
What is the difference between Profikinematik and Real-PRO on a SINUMERIK cell?
Profikinematik re-derives the toolpath from CAM-internal data using a simplified kinematic; Real-PRO replays the actual NC blocks through the controller's exported NV-archive, cycles, and machine data. Real-PRO is authoritative for 5-axis programs; Profikinematik is acceptable for 3-axis work where the kinematic chain is trivial.
Why must the PFR file be regenerated before every Profisimulation run?
The PFR file binds the CAM toolpath to a specific post-processor build, kinematic descriptor, and tool table. If any of these change on the controller side (firmware update, post update, tool-table update, kinematic change), a stale PFR produces a simulation that does not reflect current runtime behaviour and may pass green on something that will alarm on the controller.
Why does a CAM program simulate green in Profikinematik but alarm on the SINUMERIK?
Typical causes are: COMPCAD / compressor inserting blocks the CAM did not emit, cutter-compensation entry strategy mismatch, transformation selection (ROT / AROT / TCARR) not matching the active kinematic, or singularity handling differing between post and controller. Replay the program in Real-PRO with a current NV archive to identify the block.
Is ChekT-it B4 with Virtual-NC equivalent to Real-PRO?
Yes for the purposes of NC-block-true replay: both source their data from the controller's NV archive, so cycle behaviour, kinematic transformation, software limits, and protection-zone monitoring are evaluated identically. Real-PRO additionally offers a 3D machine envelope and a CAM-side collision model, which the HMI-based Virtual-NC does not.
Is the price difference between Profikinematik and Real-PRO justified?
For 5-axis simultaneous work the price difference is justified by the avoided risk of a single collision; for 3-axis prismatic work Profikinematik is the rational choice. The decision is risk-based: a single collision on a high-value part typically costs more than the entire Real-PRO seat licence, so the investment is amortised on the first avoided event.