five_ax_forever
January 11, 2026 · Original Post
Running a 2020 DMU 50 3rd Generation and over the past 6 months our B-axis positioning accuracy has drifted from ±3 arcseconds (acceptance test value) to around ±18 arcseconds measured with our API laser tracker. We're making aerospace brackets in titanium and the compound angle faces are coming out 0.04–0.06mm off on the far end of a 200mm feature. DMG Mori service ran a calibration but it came back 3 weeks later. Machine is in a temperature-controlled cell (±1°C), spindle is warm-cycled, we use RTCP. Something is wearing or drifting structurally. Anyone dealt with B-axis accuracy degradation specifically?

HansReinhardt_DE
January 13, 2026
The B-axis on DMU 50 Gen 3 uses a direct-drive torque motor with an angle encoder (Heidenhain RON 905 or equivalent). After 5+ years of heavy titanium work the preload on the rotary axis bearing pack can reduce slightly due to thermal cycling and mechanical loading — this shows up as backlash that the encoder can't see because it's on the motor side, not the table side. DMG Mori service should be performing a bearing preload check, not just a software calibration. Ask specifically for a "mechanische Grundeinstellung" (mechanical baseline adjustment) on the B-axis — that's the German terminology in their internal service documentation and using it with the technician will get you the right procedure, not just a re-run of the automatic calibration routine.

SandeepK_Aerospace
January 16, 2026
Worth checking your RTCP (Rotation Around Tool Center Point) parameters — specifically the TCP offset vector for the B-axis pivot point. On 2019–2021 DMU 50 builds there was a minor discrepancy in the documented pivot distance that DMG updated in the Celos control parameter set in 2024. If your machine was calibrated pre-2024 and never had the parameter update applied, you may have a systematic geometric error baked into your RTCP that looks like backlash but is actually a wrong kinematic model. DMG's IFP (Intelligent Function Package) calibration cycle in the 2025 Celos software release re-measures and updates this automatically using the standard touch probe — no laser tracker needed for this specific correction.

LorraineT_QC
January 22, 2026
For anyone doing aerospace work on 5-axis machines in 2026, the ISO 10791-6 standard for rotary axis accuracy testing was updated in late 2024 and is now the reference document most AS9100D auditors use when reviewing machine capability records. If you're volumetrically certifying your DMU 50 for aerospace, your calibration provider should be using a laser tracker with the new ISO protocol — not the older ball-bar or R-test methods, which don't capture the full positional error map. A proper VCC (Volumetric Compensation Certification) with the updated standard can also justify tighter part tolerances to your customer rather than building in conservative offsets for machine uncertainty.