Non-contact inspection of internal threads of machined parts

Abstract

During the manufacturing process of mass-produced automotive parts, millions of internal threads are produced annually by each automaker. The manufacturers invest a lot of effort and resources to improve the quality of the threads by inspecting their features, mainly by using manual contact gages. The majority of manufactured threads are used in threaded joints that assemble two mechanical parts. However, there are threads that have additional functionality, for example, an automotive spark plug. Recently, automotive designers have found that the angular direction of the spark plug inside the combustion chamber affects the efficiency of the engine. Hence, the manufacturers are interested in measuring the exact position of the thread starting point, allowing them to position the plug in the right direction during the plug assembly process. Our technique allows precise measurement of the thread starting point and meets this requirement. In order to improve the quality of internal threads, it is essential to measure the threads dimensional features. It is also important to inspect its internal surface for defects such as the existence of burrs, thread discontinuities, or internal porosity defects. The paper introduces two novel non-contact electro-optical techniques for inspecting small-size internal threads. The first one enables in-process internal thread geometrical measurements using a high-precision laser sensor which utilizes a motorized periscope. The second technique is based on image processing and uses innovative machine vision; a line-scan technique which uses a CCD camera with a “sight pipe” panoramic optical device. The two methods are complementary; together they allow the practical inspection of all needed features of the internal thread. The methods were tested successfully.

Appendices

Appendix A—self-calibration technique [19]

The precise linear motion system provides a highly repeatable axial scan, and a mechanical fixture can be used to roughly align the axis of the probe with the center axis of the threaded hole. However, it is difficult to align perfectly only using physical adjustment. A proper self-calibration method exploiting the relationship between the probe axis and the axis of threaded hole is introduced in this section to remove the effect caused by the eccentricity phenomena.

Self-calibration is done by utilizing two important thread parameters: the pitch and the phase. Theoretically, for every two longitudinal scans acquired 90° apart along the circumference of the threaded hole, the relative distance between corresponding peaks or valleys on the scans should be 1/4 of the pitch of the thread. This is called the phase factor. If we set the first scan at 0° on the radial cross-section as a reference, we can correct the value of PS by subtracting the estimated distance deviation ∆ _j . This value can be calculated using the following equations.

$$ {\text{Peak}}\_{\Delta_{{ji}}} = {\text{Pos}}\_{\text{Peak}}(j)(i) - \left( {{\text{Pos}}\_{\text{Peak}}(1)(i) - \frac{{\left( {j - 1} \right)}}{4}P} \right) $$

(1)

$$ {\text{Valley}}\_{\Delta_{{ji}}} = {\text{Pos}}\_{\text{Valley}}(j)(i) - \left( {{\text{Pos}}\_{\text{Valley}}(1)(i) - \frac{{\left( {j - 1} \right)}}{4}P} \right) $$

(2)

$$ {\Delta_j} = \frac{{\sum\limits_{{i = 1}}^{{p(j)}} {{\text{Peak}}\_{\Delta_{{ji}}}} + \sum\limits_{{i = 1}}^{{q(j)}} {{\text{Valley}}\_{\Delta_{{ji}}}} }}{{p(j) + q(j)}} $$

(3)

Where i is the sequence number of peaks or valleys, j = 1 ∼ 4 represents four scans at 0°, 90°,180°, and 270°; p(j) is the total number of selected peaks; and q(j) is that of selected valleys on the jth scan. Pos_Peak(j)(i) is the axial position of ith peak on the jth scan, and Pos_Valley(j)(i) is the axial position of ith valley on the jth scan; Peak_∆ _ji refers to the distance deviation between the ith peak on jth scan and the corresponding peak on the first scan at 0°, while Valley_∆ _ji is the same deviation, but with respect to valley comparison.

The resulting distance deviation ∆ _j plays an important role in finding the accurate locations of corresponding points on different profile scans.

Appendix B—dimensional measurement calculations

2.1 B1—pitch calculation

Having measured two or more thread profile scans (e.g., four scans at 0°, 90°, 180°, and 270°), the average pitch can be obtained by computing the distance between the adjacent peaks or valleys on each scan.

$$ P(j) = \frac{{\sum\limits_{{i = 1}}^{{n(j)}} {{\text{Dis\_Peak}}(j)(i)} + \sum\limits_{{i = 1}}^{{m(j)}} {{\text{Dis\_Valley}}(j)(i)} }}{{n(j) + m(j)}} $$

(4)

$$ P = \frac{{\sum\limits_{{j = 1}}^4 {P(j)} }}{4} $$

(5)

Where P is the final average pitch of the whole measured thread while P(j) is the average pitch on each scan, Dis_Peak(j)(i) is the distance between the adjacent peaks on the jth scan while Dis_V alley(j)(i) is the distance between the adjacent valleys on the same scan. Also, n(j) and m(j) represent how many pairs of adjacent peaks or valleys have been selected for analysis respectively on the jth profile axial scan.

2.2 B2—minor and major diameters calculation

The least-squares polynomial for selected consecutive peak points or valley points was determined, as shown in Fig. 10. In this case, a first degree equation was adopted to fit the data with a straight line. Therefore, any combination of two corresponding points on the fitted lines for scans at 0° and 180° as well as 90° and 270° can be used to compute major and minor diameters. Also, the offset distance, L0, should be subtracted from the sensor output data to get those diameters, which represent the distance from the sensor lens to the center of the 45° mirror attached to the periscope tube. L0 is a constant value during the entire scanning process and is a known parameter after probe assembly. The minor and major diameters are determined using the following equations:

$$ {\text{Dia\_Minor}}(i) = \left( {{L_i}\_{\text{Valley}} - {L_0}} \right) + \left( {{L_{{i + 2}}}\_{\text{Valley}} - {L_0}} \right) $$

(6)

$$ {\text{Dia\_Major}}(i) = \left( {{L_i}\_{\text{Peak}} - {L_0}} \right) + \left( {{L_{{i + 2}}}\_{\text{Peak}} - {L_0}} \right) $$

(7)

Where i = 1, 2, Li_Valley represents the distance from the sensor to the designated valley point on the ith longitudinal scan cross-section with a sequence of 0°, 90°, and Li+2_Valley means the same thing with a sequence of 180° and 270°; Li_Peak and Li+2_Peak have the same meaning in reference to the peak point. Dia_Minor is the minor diameter and Dia_Major is the major diameter.