Of the many industrial applications for lasers, fiber laser metal cutting is the largest. Current growth in this relatively mature market is being driven by continued reduction in manufacturing costs and by the ability to address new applications, both of which require increases in cutting productivity and quality.
Real-time monitoring of the cutting process is critical to these advances because it maximizes and maintains tool performance, minimizes downtime, and facilitates automation and closed-loop control. The trend toward higher laser powers is further motivating the need for a robust process monitoring solution because high-power process excursions and unproductive beam-on time can lead to degradation of the cutting bed or head.
A recent breakthrough in process monitoring, in which the collection, detection and analysis of detailed process information is integrated directly into the fiber laser, enables accurate, actionable process monitoring without additional optics or mounts, modification of the cutting head, or other perturbations that increase tool complexity and decrease tool reliability. This new capability allows tool integrators to easily implement sophisticated monitoring, tracking and control of the cutting process without modifying the tool design.
Current state of the art
Process monitoring is infrequently used in fiber laser cutting [1]. It is instructive to compare the situation to laser welding, in which process monitoring is widely used. Most welding process monitoring uses photodiodes (PDs) to detect light emitted by the process. PDs provide high speed (sub-microsecond time resolution), detection of a wide range of wavelengths and various integration options for collection of the process light.
The lack of process monitoring in laser cutting is the result of both technological and practical limitations. Specifically, integration of standard PD-sensing approaches into a cutting tool requires adding free-space optics, mounts, and/or fiber bundles to the cutting head or optical path to sample the process light. This added hardware introduces complexity and new failure modes.
Sampling problems are more pronounced in laser cutting than in welding because the cutting head experiences significant acceleration and shock/vibration, whereas welding often uses a stationary scanner. Furthermore, existing PD-based technologies incorporated into the cutting head suffer from a low signal-to-noise ratio and have struggled to reliably assess cutting performance for thick metals.
A significant unmet need therefore exists for a high-speed, accurate cutting monitor. This monitor must be capable of detecting process variations for a wide range of materials and cutting conditions, and it must not compromise tool performance and reliability.
Integrated cutting monitor
nLight Inc. has addressed this need by developing a new technology, ProcessGuard, that leverages two differentiated capabilities:
- Fiber-integrated sampling of process light: Rather than external optics and hardware to sample the process light, ProcessGuard uses the existing feeding fiber, with no perturbation of the cutting head or optical path. Light emitted from the workpiece over a wide range of wavelengths and angles is collected and transmitted by the feeding fiber back to the laser chassis for analysis, thereby eliminating the complexity, reliability, stability and optical loss problems associated with existing sampling methods.
- Plasmo optical detection and analysis: nLight-Plasmo has developed a PD-based process monitor, ProcessObserver, that has been deployed in more than 1,000 industrial facilities worldwide. The standalone product has been honed over the past 30 years, and ProcessGuard builds on this expertise by incorporating the ProcessObserver PDs and optical filters into the fiber laser to detect the process light collected by the feeding fiber. Plasmo’s sophisticated electrical filtering and optimized algorithms are used to analyze the process light to extract accurate cutting process information in real time
ProcessGuard is compatible with nLight’s portfolio of fiber lasers, including the full power range (400 W to 20 kW), single-mode and multi-mode feeding fibers, and both static and dynamic (Corona) beam shapes. ProcessGuard has already been successful in high-volume E-mobility welding applications [2], and we have now applied it to fiber laser cutting. The following sections present results from the nLight Applications Laboratory and from a leading tool developer (Proteck Machinery, Chennai, India).
Cutting results
ProcessGuard has been evaluated for cutting a wide range of materials (mild steel, stainless steel, aluminum, copper, brass) with thicknesses of 2 mm to 25 mm and with N2 and O2 assist gases. These tests conclusively demonstrated rapid and reliable assessment of cutting performance and process excursions, including:
- Pierce completion
- Loss of cut
- Focus shift
- Speed changes
- Laser power changes
- Gas flow variations
Measurements in the nLight Applications Laboratory employed a Nukon Vento 315 cutting tool equipped with a 15-kW CFX (Corona) fiber laser. Results from Proteck employed a Prolaser Blaze cutting tool (see Figure 1) equipped with a 5-kW CFX fiber laser. Both tools employed standard, fixed-optic cutting heads with 2x magnification. The ProcessGuard sampling rate was 250 kHz (4 microseconds per data point).
Pierce detection
In many cutting applications, piercing represents a substantial fraction of the total production time and thus contributes significantly to the part cost. For example, according to Firoz Koser of Proteck, “when cutting 25-mm mild steel, a typical layout for a 3-m-by-1.5-m plate may have nearly 30 percent of the total beam-on time consumed by piercing.”
Furthermore, “for thick materials, it is common to experience significantly varying piercing times across different areas of the sheet due to the surface conditions and slight variations in material composition.” Recipes thus include extra piercing time, often by more than a factor of two, to ensure pierce completion before beginning to cut. In addition to reduced productivity, this “wasted” piercing time degrades the part quality because of extra heat input into the workpiece, and it can contribute to tool damage (e.g., contamination of the cutting head optics).
Figure 1 shows the ProcessGuard signal when piercing 20-mm mild steel (MS). The pierce recipe has a total duration of 3.4 sec., but the pierce is complete after 1.6 sec., meaning that 1.8 sec. (53 percent) of the piercing time is wasted. Koser notes that “a typical layout may involve more than 600 piercings, and there is no way to improve the reliability of the process without implementing a closed-loop system that efficiently and consistently detects pierce finish.”
Tool integrators can easily implement such closed-loop control using a digital output signal provided by ProcessGuard immediately upon pierce completion. Figure 2 shows superimposed ProcessGuard signals for eight piercings of 6-mm MS. The red lines show detection of each of the eight pierces.
Importantly, the parameters for pierce detection are set automatically by the ProcessGuard software, independent of material and thickness, and do not require expert input. The recipe can stop the pierce sequence and initiate cutting immediately after completion of each pierce. Furthermore, by tracking changes in the piercing time, drifts in tool performance can be diagnosed without the risk of producing bad parts (eliminating scrap), and problems can be corrected when convenient (reducing downtime). ProcessGuard’s rapid pierce-detection capability can also be used to optimize the piercing recipe during process development, further increasing productivity.
Loss of cut
Loss of cut was easily observed by ProcessGuard for all samples tested. For example, Figure 3 shows results for cutting circular parts from 6.4-mm aluminum. The ProcessGuard screenshot shows three good parts (green traces) and one part for which the cut was lost at three positions (yellow trace).
The signal measured during the bad cut is significantly outside the process limits, and the three peaks correspond to the three regions where the cut was lost. The high time resolution of ProcessGuard allows defect locations to be precisely identified. This capability is critical for minimizing yield loss by localizing failed cuts, for efficiently determining the root cause of process excursions and for closed-loop control of the cutting process. Critically, ProcessGuard can even identify the locations of defects that are not visible to the naked eye.
Focus shift
Focus shift is caused by heating of the protective window or other optics in the cutting head, which changes the laser spot size and divergence on the workpiece. This heating is typically caused by contamination and/or by absorption in the optical materials or coatings. The problem gets worse as the power density on the optics increases (i.e., at higher laser powers and smaller beam sizes). Focus shift degrades the cut quality and speed, eventually leading to loss of cut.
ProcessObserver is already used in factory welding tools to detect focus shift. We investigated extending this capability to cutting tools by deliberately adjusting the focus position (f) with a clean cutting head (no contamination) and measuring the impact on cut quality and on the ProcessGuard signal.
Figure 4 shows results for cutting 6.4-mm MS. The process window for producing good parts is f = +2 ± 3 mm. The graph shows the mean value of the ProcessGuard signal for multiple parts cut with f between +5 mm and -5 mm.
The signals are well within the process window for all good parts (green points). The signal decreases near the edge of the process window, at f = 0 mm and -1 mm. Importantly, this decrease is observable while still producing good parts, alerting the operator to a change in the process without generating scrap or low-quality parts. The signal is unambiguously outside the process window for focus positions that produce bad parts (red points).
The ProcessGuard software includes the ability to set a warning level (orange line) to alert the operator or machine software to process drift, allowing the root cause to be determined and corrected with no yield loss and minimum downtime. Furthermore, such quantitative process monitoring enables automation, allowing reliable production of high-quality parts and continuous assessment of the process with minimal or no operator intervention.
Gas flow variations
Figure 5 shows the results of pressure (gas flow) changes for the same cutting conditions as Figure 4. The process window for producing good parts is 0.4 to 0.9 bar. Flow changes are evident in the 0.4 to 0.7 bar ProcessGuard signals (green traces), again demonstrating the ability to detect process changes before the onset of poor quality. The signal clearly falls outside the process limits when producing bad parts (0.3 bar, yellow trace).
Cutting speed
A given cutting recipe typically has a range of cutting speeds for which the cut quality is acceptable. Although cutting at the highest speed within this process window would maximize productivity, in practice the operator usually works below the theoretical maximum to accommodate variations in the material and the process. ProcessGuard offers the possibility to work at the high-speed end of the process window.
Koser of Proteck noted, “Operators gain the confidence to run production at the maximum feed rate without worrying about part scrap due to cut loss, as the system provides early warnings when the process begins to deviate from set limits – allowing for timely corrective action. Based on the initial ProcessGuard results, I analyzed the cycle time for a typical layout when cutting 10-mm MS. Productivity can increase by approximately 25 percent to 28 percent while maintaining high edge quality, which actually improves at maximum feed rates.”
This dramatic improvement illustrates the fact that cutting tools without process monitoring are leaving productivity, and thus money, on the table.
Conclusions
With the advent of ProcessGuard, tool integrators and end users can now rapidly and accurately monitor the cutting process without introducing complex and failure-prone hardware to the cutting head or the optical path. The availability of quantitative, real-time process monitoring creates exciting opportunities to maximize tool productivity and part quality. Specifically, ProcessGuard enables:
- Development of optimized cutting and piercing recipes, including automated optimization, without variability due to qualitative, subjective or non-expert assessments
- Increased productivity by maximizing yield and allowing operation at the high-speed end of the process window
- Detection of process excursions before they unacceptably degrade part quality, thus reducing or eliminating scrap and minimizing downtime
- Root-cause analysis of process changes
- Reduced dependence on skilled operators to achieve high-quality parts
- Statistical process control by noninvasive, real-time tracking of all steps of the cutting process, opening the possibility of automation and closed-loop control
This major advance will fuel growth of the laser cutting market by driving down costs and increasing quality in both existing and new applications.
1. N. Levichev, G.C. Rodrigues, and J.R. Duflou, Procedia CIRP 94 (2020), p. 499.
2. D. Kliner, A. Braglia, A. Califano, P. Huang, and C. Gavrilov, PhotonicsViews, digital special issue D1 (2025), p. 17.
