A bridge-type coordinate measuring machine uses a moving gantry and a tactile probe to capture X, Y, Z coordinates of prismatic and contoured workpieces, with typical MPE of 2–6 µm in volumes of 500–2000 mm, per ZEISS product-page category descriptions [S3]. A vision measuring machine replaces the touch probe with a telecentric lens, a stage-mounted camera, and a backlight, and reads edges or surface features in 2D or 2.5D against a pixel-accurate scale.
The two systems are commonly mis-bid against each other on the same RFQ. The decisive variables are feature dimensionality, batch throughput, tolerance stack, and part weight. For dimension-only inspection of small flat stampings or PCB features under 200 mm, a VMM cycle is 3–10 s; for full 3D GD&T on a 600 mm housing, the bridge CMM cycle is several minutes. Cost per machine follows the same gap: entry-level China-made bridge CMMs list around USD 18,000–35,000, while production VMMs sit at USD 8,000–25,000 [S1][S2].
Working principle and measured feature scope
CMM measurement is contact-based: a touch-trigger or scanning probe records the contact point against a calibrated spatial scale, after which the software reconstructs the geometric element, datum, and GD&T call-out [S4]. ZEISS lists the bridge-type CMM as the most widely used CMM architecture for general workpieces, alongside gantry, horizontal-arm, and shop-floor variants [S3].
A vision measuring machine measures non-contact by imaging the backlit or front-lit feature with a fixed or zoomed telecentric optic, then computes pixel-edge positions against an internal linear scale; depth (Z) is limited to a few millimetres unless the VMM is upgraded to a laser or focus probe, in which case the system is often reclassified as a multisensor CMM [S3]. The hard line: a VMM is a 2D/2.5D tool, while a CMM is a true 3D tool, and the part envelope, weight, and feature type dictate the boundary, not the brand. For dedicated 2D optical setups used in ID and code reading, the machine vision system reference is the relevant sibling category.
Accuracy, repeatability, and ISO 10360
Bridge CMM accuracy is governed by ISO 10360-2 for length-measurement error and ISO 10360-5 for probing error; the resulting MPEE value is the single number a quality engineer compares across vendors. Production-grade bridge CMMs typically quote MPEE in the 2–6 µm range across their full measuring volume, with the ZEISS PRISMO family positioned in the precision class and XENOS marketed as the reference device [S3].
Vision systems do not have an equivalent ISO 10360 MPEE; their performance is given as field of view (FOV), optical resolution, lens magnification, and edge-detection repeatability in pixels, typically 1–3 µm at 0.5× magnification on a calibrated glass scale. The shop consequence is direct: if the drawing calls out a 5 µm flatness on a 400 mm housing, a VMM cannot validate it, and a CMM is mandatory. Reverse, if the spec is a 0.05 mm edge-to-edge distance on a 30 mm connector, the VMM is faster and cheaper to operate.
Throughput, automation, and shop-floor footprint
A VMM typically runs 3–10 s per measurement cycle, and most production VMMs in the China-made catalog are built around an automatic CNC stage, ring or back light, and optional laser probe, per Shenzhen Sisheng's CMM/measurement listing [S1]. A bridge CMM cycle for a multi-feature GD&T routine is normally 2–10 minutes depending on probe changes and datum alignment [S3].
Throughput is also constrained by part loading: a VMM stage handles parts up to a few kilogrammes and under 300–500 mm in typical configurations, while a shop-floor CMM such as the ZEISS DuraMax line is rated for harsh production environments at up to 40 °C and high mechanical loads, and gantry CMMs extend measuring volume to multi-metre parts [S3]. Helmel's US-made moving-bridge CMMs target the same general-purpose aerospace, automotive, and medical workloads with 50+ years of in-house engineering [S4].
Decision matrix: CMM, VMM, or multisensor
On 2D features below 200 mm with tolerances above 0.02 mm, the VMM wins on cost, throughput, and operator skill required. On 3D GD&T, prismatic features, or heavy workpieces above 5 kg, the bridge CMM wins because the VMM has no Z resolution and no reach. On small precision parts that need both 2D edge measurement and 3D probing, a multisensor CMM that combines optical and tactile probes in one volume is the engineering answer, and ZEISS explicitly markets the multisensor CMM as a single platform covering optical, contact, and laser scanning [S3].
Selection criteria, four-axis check, cost: VMM wins for budget under USD 25k, CMM wins above that. Tolerance: CMM wins for sub-5 µm, VMM wins for 20–100 µm. Feature dimension: CMM wins for 3D GD&T, VMM wins for 2D edge geometry. Operator skill: VMM is easier, with typical training in days, while a CMM needs a metrology-trained operator familiar with datum alignment and probe calibration.
Use cases by industry
Aerospace and medical device shops default to bridge-type CMMs with scanning probes for tight-tolerance prismatic and freeform surfaces, and they tolerate the 2–10 min cycle time for the accuracy gain [S3][S4]. Electronics, connector, and PCB shops default to VMMs for edge-to-edge, pin pitch, and coplanarity inspection on parts under 100 mm. Stamping and machining shops doing 100% inline inspection sit in the middle, and a shop-floor CMM such as the ZEISS DuraMax or a multisensor CMM is the typical answer when the measuring cell sits next to the machine tool [S3].
For part ID, barcode, or character recognition on the same line, a separate machine vision ID station is normally added next to the CMM or VMM, since metrology and identification have different lighting, focal-length, and software stacks. A VMM is sometimes confused with a full machine vision system; the VMM is a calibrated metrology device with a glass scale, while a general machine vision system is an inspection cell for presence, defect, or code, not a dimensional reference.
Limitations, failure modes, and standards
CMM failure modes: thermal drift, probe crash, scale contamination, and datum mis-alignment by an inexperienced operator; the ZEISS shop-floor CMM line is specified to 40 °C ambient to limit the first [S3]. VMM failure modes: focus drift, edge-detection noise on low-contrast features, and backlight ageing; telecentric optics limit parallax, but a tilt of 1° on the part can move the apparent edge by several pixels.
Standards anchor: ISO 10360-2 and ISO 10360-5 are the CMM accuracy tests a buyer should demand with a quotation, including the test artefact size, the test length, and the resulting MPEE. For VMMs, the comparable calibration is a glass scale traceable to a national metrology institute, and a per-magnification pixel-size certificate. A vendor that refuses to publish an MPEE or a pixel-resolution certificate is a red flag on either side of the comparison.
Vendor landscape and sourcing signals
The CMM market splits into three tiers: high-end European and Japanese (ZEISS, Hexagon, Mitutoyo, Renishaw-probe-based) for the precision class; US mid-size (Helmel Engineering, in production since 1973, with 50+ years of engineering and US-made moving-bridge systems) for aerospace and defence workloads [S4]; and China-made general-purpose bridge CMMs from vendors such as Shenzhen Sisheng and the Yanchent / Ain CMM basic OEM line, which compete on price-volume ratio for entry-level and OEM system integration [S1][S2].
VMM sourcing is dominated by Chinese and Taiwanese OEMs at the entry and mid range, with European and Japanese vendors at the high-precision end. A 2025–2026 sourcing pattern: aerospace and medical shops in North America are still buying CMMs with multi-year lead times on probing systems; general manufacturing is shifting first-article inspection to a VMM and using the CMM for periodic recalibration. For cost-driven flow, the vision measuring machine reference frames the same product class with different terminology; for contour measuring machine selection the sister article clarifies the non-CMM profile-and-roughness alternative.
For deeper spec-driven comparison on a related family, see the vortex flowmeter vs gas mass flow controller selection logic, which uses the same decision-matrix approach applied to flow, not metrology.