最新案例当前位置:首页>技术 & 产品展示>最新案例

全自动高速3D检测系统--轮廓全尺寸检测

客户要求:产品检查周期:每秒一个
后盖上任何点变形,高度偏差少于0.1mm
全3D扫描检查
Z扫描精确度为0.02mm
技术方案:
1) 全3D扫描,获得全3D数据,与标准值做比较,如果少于0.1mm,则通过,否则为NG
2) 高度Z精确度为0.02mm
3) 由于精度和速度的要求,系统需要采用美国Stockeryale 40mW精密聚能激光线发生器,德国制Ranger 3D高速扫描CCD,工业级圆心镜头,伺服驱动系统。这是目前唯一可以达到要求的配置。
4) 扫描激光线采用40mm宽度,一个产品扫描一次,一秒一个。
5) 最后的机器系统大概如以下图片一致。
请看以下使用现有设备扫描的结果。
1) 由于镜头和激光并非这个项目用途,所以需要扫描2次才能得到一个工件全部3D数据。
2) 工件扫描面的最低位与最高位大概相差13mm,(不算工件厚度),目前梯度级别为640级,即0.02mm,应该满足0.1mm变形的要求。
Technical Proposal:
1) Full 3D scan, get all point 3D data, and compare with standard. If less than 0.1mm, it will pass.
2) Z 3D data accuracy are 0.02mm.
3) Because of the accuracy and speed requirement, the system should be configured with American made Stockeryale 40mW accurate laser line generator, German made Sick Ranger 3D scan CCD, industrial tele lense,servo system. Per to our assessment, it is the only configuration to achieve the requirements.
4) With 40mm width scan laser and telescope lens, it will take about 1 sec to scan 1 pcs.
5) The final machine configuration will be similar to the following picture machine.
Please see the following scan result with the current available machine.
1) Because the lens and laser model are NOT for this scan purpose, it will need 2 scans for each pcs.
2) The depth between top of the cover and bottom (for the scan surface) is about 13mm. The current accuracy is 640 pixels. That is said the accuracy is 0.02mm. It should satisfy Xinhao’s 0.1mm warpage check requirement.

1) 上方与左方有一个突出的矩形,是临时治具突出的部分,挡住了盖子一些部分,不是正常的3D数据显示
2) 显示的Pass和Fail字体是因为提供样品上有的标记
3) 因为测试的精度为1级0.02mm,极限为0.1mm,就是与及格样品的差别为4级0.08mm以下为及格,4级以上为不及格。图上红色区域为不及格部分,即差别为4级以上部分
4) 按照图上的表示,分区域取点做比较,结果如下:
1) There are shadows on top and left. This is the edge of the temporary cover fixture to block part of the cover. These are not the normal 3D data.
2) The “Pass” and “Fail” on the image are the mark on the sample cover.
3) The accuracy for the test is 1 grade 0.02mm and the limit is 0.1mm. That is said if there is more than 4 grades 0.08mm in any point height on the cover (compare to the “Pass” cover), it will be fail and marked in red. The red area are Fail in the image.
4) Take some points for compare from the images. The results are as below:

5) 可以留意到2个样品的前大部分都是及格的,只是在后面部分产生了偏差。取出后部不合格的部分取点做比较
6) We can see most area of the fail sample is PASS with the deviation on the lower part. Highlight the lower part and take points for compare.


下面这个是在后盖上放了一小片薄纸,纸的厚度为0.09-0.1mm (就是要求的高度变化值),扫描可以明显看出轮廓的差别,数据大概是10级(大约0.2mm,因为可能有一些空隙)。






http://www.waltrontech.com/upload/fckeditor/2011-06/22165536.jpg
If we need to show the real height in mm, we need to do a calibration to create a link between mm and 1 pixel. At the moment, we have no software code for this test because it takes time for programming.
Your concerns:
1) How to use the 3D data to check the scan cover warpage?
We have all 3D data for whole surface. Take a point in the cover and gets its Z as Z1. Compare with the same point in the standard cover Z data as ZS. If Z1 and ZS difference is higher than 0.1mm, the cover fails. We can program to compare as many points as requests.
2) How to get the same point in different pcs?
We will use machine vision method to extract the cover profile for the scan surface. Then we set 1 base 0 point, for example the 1st left point on profile of every scan. We need to check the point X1, Y1, Z1 (the point is located on X1, Y1, Z1 according to 0 point). Then we can find every scan surface’s 0 point and its X1, Y1, Z1 point and then compare all the X1, Y1, Z1 with the standard X1, Y1, and Z1.
3) How to set a standard surface to be compared?
Just put a standard cover to be scan and get all its 3D data. All these data can be compared with other covers.
4) How to generate a standard cover?
Every actual cover is not 100% perfect with lower than limit defects. It is not the meaning standard cover. The standard cover is generated by a mathematics model to scan many “pass” covers. The model will analysis all same point data and gets a standard point data.
客户关心的问题:
1) 如何使用获得的3D数据检查工件变形度?
我们已经可以获得全个面得3D数据。取盖子上一点,它的高度为Z1。与标准盖子上相同的点的高度值ZS做比较,如果差别大于0.1mm,就为不及格。可以编程按要求比较各部分尽可能多的点。
2) 如何在不同工件上得到相同的点?
使用机器视觉软件提取后盖的轮廓。可以找一个合适的点设为零点。例如取盖子左上方第一点为零点。我们要检查的点为相对该零点的X1, Y1, Z1。然后我们可以对不同的后盖取相同的零点,然后根据坐标得出不同后盖中相同的点X1, Y1, Z1。
3) 如何设立一个供比较的标准曲面?
放入一个标准的后盖,扫描获得全部3D数据。这些数据给其他待检的后盖做比对。
4) 如何产生一个标准的曲面?
每一个实际后盖都不是100%完美的。提供扫描的所谓标准后盖只不过是误差小于要求而已。标准后盖是通过一个数学模型产生的,软件扫描很多被认为及格的后盖,通过数学分析方法得出一个标准的曲面,这个曲面就是可供比较的标准曲面。