Torna alla pagina italiana | Back to the English page

SMARTMEC

On-line quality control, production process assesment and tracking system for mechanical parts

The Istituto Nazionale di Ottica Applicata (INOA) has been involved in the SMARTMEC European Project for the study of roughness of the mechanical pieces by means of optical technique. In the first part of the project different optical principles and hardware configurations have been studied and compared even by means of representative laboratory tests.

Then two sensors have been designed and realised by INOA for assessing the roughness of mechanical components, both of them are based on non-contact optical techniques :

• a beam scanning optical profilometer based on a CD-player pickup aimed to measure the roughness on the IVECO gear. It utilises the pickup as both the focus sensor and the lateral actuator. Linear scanning on the piece is performed by means of the objective lens tracking motor.

• a sensor based on the angular distribution of the scattered radiation has been applied for evaluating the roughness on a truck crankshaft supplied by RENAULT. By comparing the scattered and reflected portions of the incident light it is possible to assess surface roughness. The device uses a laser-diode and a 16 photodetectors interfaced with a data acquisition board, it performs on-line evaluation of the finishing quality.

Both the devices are completely controlled by a personal computer. The performance of two laboratory demonstrators have been verified on various samples provided by the end users of the European Project.

List of SMARTMEC Reports by INO.

OPTICAL CONTROL OF CRANKSHAFT FINISH

Abstract - Roughness is monitored on crankshaft surfaces using a compact optoelectronic device. The surface quality is evaluated by comparing the light scattered from two different ranges of angles by means of 16 photodiodes. The experimental data show that the ratio between scattered and reflected light intensity increase with the surface roughness.

ANALYTICAL MODEL: THEORETICAL SCATTERING PROFILES

The scattered light is similar for all periodic surfaces, since these profiles can always be considered as a sum of sinusoidal components, and for the corresponding solutions the superposition principle holds. The mean diffused power, defined as , for a mono-dimensional surface normally distributed is given by the following expressions:

            for g<<1 ;

                  for g>>1

where ,  is the roughness, T the correlation distance, m the scattering order and L the illumination length. While the other parameters are:

with  and  incidence and scattering angle, respectively.

Fig. 1 presents some theoretical scattering profiles calculated for increasing roughness values (indicated in microns in the legend), with incidence angle 15° and T value 5 microns. It is important to note how the peak value decreases as the roughness increases from the value 0.1 micron (g value almost 0) towards the value 1 micron (g>>1), whose curve is the lowest one in the figure.

When the roughness increases the intensity of the specular component becomes weaker and the scattered light spreads over a wider angle.

SENSOR FOR ROUGHNESS ASSESSMENT ON FLAT SURFACES

The PD8 sensor, with 8 photodiodes on an arc of circle as shown in Fig. 3, has a geometry especially adapted for collecting the light scattered from flat surfaces. It has been calibrated and tested on a plate of roughness samples with different surface finishing (grinding, lapping and milling).

Fig. 3 presents the PD8 sensor view from the top and from its front face (upper figure).

A simple elaboration of the eight signals has been performed considering the ratio between the scattered and the reflected light, whose results are reported in the last column of Table 1 for nominal roughness values between 1.6 microns and 0.05 microns. All S/R ratios show a non-linear increase versus the roughness value Ra for all type of surfaces. Therefore the scattering sensor allows performing relative roughness measurements on plane surfaces for controls during manufacturing or checks on the final mechanical products.

SENSOR FOR ROUGHNESS CONTROL OF CURVE SURFACES

The successive step of the study was to adapt this roughness assessment technique to measure the curve surface of the RENAULT crankshaft. Hence the detection geometry required a modification to follow the distribution of the light scattered by the curve surface of the crankshaft. As Fig. 4 indicates, the light scattered from the crankshaft surface shows a pattern oriented in the direction of the shaft axis, due to the mechanical working (which produces small furrows perpendicular to the cylinder axis of the crankshaft).

Fig. 4

The scattering sensor for curve surfaces, indicated as PD16 sensor, is presented in Fig. 4, where the laser and the 16 photodiodes array are schematically drown. The scheme in Fig. 4 also shows a crankshaft portion (on the left) and acquisition and elaboration hardware (on the right) of the PD16 sensor. The laser beam incidence is almost normal to the curve crankshaft surface in order to minimise the curvature of the scattering pattern, which can thus be detected by a linear array. The length of the photodiodes array is 16 x 4 mm = 64 mm, and it is placed parallel to the shaft axis in order to detect the scattered light, which is directed along the shaft axis. The PD16 sensor detection is always performed with the diode array outside the "shaft wings", but as near as possible to the surface for maximising the detected portion of the scattered pattern. In Fig. 4 the reflected light is indicated by I_r, while the two scattered light portions have been named I_H and I_L (for High and Low, considering a vertical detection). In a general detection case the two scattered light quantities are symmetric and they can be both detected if the distance from the wings is sufficient. While in the two extreme positions, near the wings, part of the scattered light is shaded and can not be detected by the diode array. Thus double symmetric detection of I_H and I_L should overcome the problem of having part of the scattered light shaded by the shaft wing.

The elaboration principle is the same of the PD8 sensor: roughness data are obtained by comparing the scattered light to the reflected light. The 16 signals are suitably amplified and converted, thus they are acquired through an acquisition board on the PC. Finally the 16 channels data are elaborated for obtaining the Scattering/Reflection ratios, reported in Table 2 and in Fig. 8 for two series of measurements performed on all crankshaft surfaces.

Fig. 5

Fig. 6

The photo in Fig. 5 shows a crankshaft surface examined by the scattering head of the PD16 sensor, whose geometry has been designed to match the pattern of the light scattered by these crankshaft surfaces. The more suitable position of  the scattering head in Fig. 5 is very close to the curve surface and at a very small scattering angle, since the scattered light distribution increases its curvature as the source-detector angle increases.

The demonstrator system performing roughness measurement, which was tested in the final meeting at RENAULT plant, is shown in Fig. 6. This demo system is completely portable and it is composed of:

1. The scattering head with the diode laser and the 16 Photodiode array

2. The support shifting the scattering head vertically

3. The electronics amplifying the 16 diode signals

4. The DAQPad-MIO-16XE-50 acquisition board for notebook

5. The Texas Instrument notebook TI711 displaying the results

The display in Fig. 6 reports the scattering profile (plotted as the 16 acquired signals) and gives three roughness values calculated from I_L, I_H and I_L + I_H (indicated in Fig. 4). If the scattering profile is symmetrical the first two roughness values are very similar and the final mean value is just slightly more precise than the first two. When there is vignetting by a crankshaft wing or other caused of asymmetry of the scattered light it is useful to have the two separate S/R ratios giving only one correct roughness, which can be chosen on the base of the scattering profile displayed.

Fig. 7

The profiles of the light scattered from the crankshaft surfaces are plotted in Fig. 7 for increasing roughness values (Ra from 0.6 microns to 3.6 microns).

The trend of these measured scattering curves is in complete agreement with the theoretical scattering profiles of Fig. 1 and with the description of the scattering properties.

The working principle of the sensor is based on the behaviour of these experimental profiles: when the surface roughness decreases, approaching a mirror surface, the reflected light increases and the scattered pattern spreads over a smaller angle. On the other hand, as the surface becomes rougher the reflection peak disappears and the light is mainly scattered.

The elaboration of the 16 PhotoDiodes data, giving the scattering profiles of Fig. 7, is still based on the comparison of scattered and reflected light portions.

Fig. 8

Fig. 8 presents the results of two series of measurements performed on all crankshaft surfaces evidencing the relationship between the Scattering/Reflection ratio and the roughness Ra.

The Scattering/Reflection ratio reported can thus provide an evaluation of the surface quality.

Moreover, in the case of the crankshafts examined, from the S/R ratio we can determine the type of surface finishing, since for lapped surfaces the roughness is Ra < 2mm, while ground surfaces have Ra > 2mm.

Measurement results are reproducible, but there is a theoretical difficulty to assign an absolute value to the roughness measurement. For this reason a more suitable application of this technique can be the off-tolerance detection of the produced mechanical pieces.

Several series of measurements of the Scattering/Reflection ratio have been performed on all crankshaft surfaces. The first two series, denominated Meas1 and Meas2 have been used to calibrate the sensor. Table 2 presents the results of these two calibration measurements of S/R in the last two columns, while the corresponding values of Ra are reported in column 4 and 5. The R/S data are graphically shown in Fig. 8 versus the nominal values of Ra. This calibration plot evidences the almost linear relationship between S/R ratio and roughness Ra .

Results of the roughness measurements, performed by the calibrated sensor of the demonstrator shown in Fig. 5, are presented in Table 3. Columns 4, 5 and 6 report best  Ra measurement, minimum and maximum Ra values, respectively. For comparison the last two columns of Table 3 again report the Ra values corresponding to the calibration data.

Torna alla pagina italiana | Back to the English page