Smart laser sensors spy tiny defects in tires before they roll off the assembly line.
Dr. Walt Pastorius
LMI Technologies Inc.
Windsor, Ont., Canada
LMI Technologies (USA) Inc.
Tire makers inspect product in process and when it's finished. For both jobs, lasermeasurement sensors have become the tool of choice over contact/ mechanical followers and capacitive sensors. Recent lasermeasuring systems that employ highspeed digital data communications nullify electrical noise and eliminate separate analog-to-digital converters. This makes 100% inspection feasible, which lowers scrap rates.
A big advantage of laser sensors is that they work without getting close to or touching a tire. Capacitance sensors, in contrast, must get extremely close and perpendicular to a tire surface to be measured. This requires complex, multiaxis positioning mechanisms and frequent sensor recalibration because material properties change during the manufacturing process and capacitive sensors are sensitive to those changes. Also, capacitance sensors don't work reliably on certain materials such as high-silica rubber. Not to mention, tires can hit and damage improperly positioned sensors. Capacitance sensors measure over a relatively large surface area and can't image grooves, lettering, and bar codes. They must instead be positioned over a clear path on tire sidewalls.
Contact sensors such as LVDTs (linear-variable-differential transformers) have similar shortcomings. Contact pressure deforms the rubber, resulting in erroneous data. In-process measurements, such as extrusion profiling, are impractical with contact sensors because tire surfaces tend to be hot and gummy at this stage of production. To measure consistently, sensors used for final inspection need a clean path on the tire as it rotates at 60 rpm. Tire lettering or embossments can destroy touch probes at these rotation speeds. And at lower speeds, these same features as well as grooves, sipes, and pin vents can trigger contact bounce that hurts repeatability.
In contrast, noncontact laser triangulation sensors measure surface features while safely remaining several hundred millimeters or more away, eliminating sensor crashes. The sensors gather reliable data even when not perpendicular to the surface, which simplifies sensor mounting and positioning. Properly designed laser sensors need no recalibration,are insensitive to changes in material properties and surface condition, such as color, finish or the presence of bead lubricant. There are two basic types of lasermeasurement sensors: point triangulation and line.
Point-triangulation sensors, as the name implies, take data at a single point, effectively acting as a noncontact LVDT. Such sensors employ a low-powered laser that projects a spot about 0.1 to 0.3 mm in diameter on the surface to be measured. At an angle to the laser beam is a lens that forms an image of the laser spot. As the surface moves towards or away from the sensor, the position of this imaged spot shifts laterally at the image plane. A position-sensitive detector (PSD) placed at the image plane tracks spot position. A circuit then calculates sensorto-surface distance from the resulting triangle. Point-triangulation sensors support high data rates (16 to 32 kHz) and resolutions to 25 m.
Laser-line sensors operate much the same way as point-triangulation types. A key difference is that the laser beam is optically expanded in one dimension to form a line of laser light on the surface to be measured. A 2D digital array or camera serves as a detector. Measurements are taken at multiple points along the laser line in each camera image or frame. Camera output feeds to a circuit that does the triangulation calculations.
Laser-line sensors take data over multiple points on the surface. Most laser-line sensors output at a slower rate than point sensors typically 10 to 60 frames/sec for each line of data though some are capable of 1-kHz frame rates or higher. Resolution depends on how the laser light is projected and imaged. Also, lighting is an issue because the camera images the full line width for each exposure and needs a reasonably uniform surface for good results.
Directly related to the data or frame rate is data density. Analysis software filters out all points related to lettering, bar codes, and other acceptable surface variations. A high data density ensures adequate resolution for detecting bulges and dents on tires as they rotate at 60 rpm during final inspection. Bulges are typically spaced 0.2 to 0.3 mm apart with heights from 0.3 to 3.0 mm and widths from 5.0 to 7.0 mm. PSD-based laser sensors take 16,000 to 32,000 data points/rev at 60 rpm, far greater than possible with a line sensor. As such, PSD sensors can measure bulges and other deformities to better than ±0.025 mm. The ability to reliably measure narrowly spaced bulges is important because many are not normal cord-related features but defects such as air blisters.
High sensor data rates let manufacturers inspect tires reliably without limiting production rates. But sensors must also communicate data to a PC or other device. Common digital interfaces such as RS-232 and RS-422 have limited transmission speed, so most point-triangulation systems communicate data by analog methods at the full bandwidth of the sensor. However, analog transmissions can pick up unwanted electrical noise, and receiving devices need an analog-to-digital converter, which adds cost.
A special high-speed RS-422 port capable of streaming transmissions to 460 kbit/sec gets around these limitations. For example, "real-time" point-triangulation-sensor data can combine with data from encoders that monitor tire-rotational position during the inspection cycle. This allows radial-runout inspections that identify the circumferential position of maximum tire radius. Also, the ability to sync sensor analysis to tire speed eliminates the need for close rotational speed control.
So-called "Smart Sensors" go a step further. They run customized algorithms embedded in the laser head itself, eliminating PLCs and PC controllers. Highspeed data pipes are no longer needed because sensors transmit only finished analysis. Software functions include linearization, conversion to engineering units (digital output), automatic gain control that compensates for rapid-object reflectivity changes, low-pass filtering, and in some cases analog-to-digital conversion.
A 32-bit Risc processor and large memory (both RAM and nonvolatile) lets the sensors host complex application programs and store operating parameters for specific applications. The technology simplifies software routine programming and sensor control, improves robustness by eliminating external processing devices, and enhances security for integrators installing proprietary processing algorithms. It also simplifies the retrofitting of older contact or capacitance-sensor-based systems.
LMI Technologies Inc.,