If two belts do not move at the same velocity while a product changes from one belt to another, the motion of the product during belt change temporarily cannot be predicted. This can lead to considerable measuring errors when detecting the products. These errors can be minimized by means of clever adjustment of the sensor position on the belt and the method with which the product can be detected. When searching for a suitable position of the sensor for a measuring method, you should keep in mind the three phases of the product transition.
oThe product is still located entirely on the previous belt.
In this period, the product will move at the same velocity as the previous belt. Measurements at this point can only refer to the previous belt.
oThe product is partly on the previous belt and partly already on the belt.
In this period, the motion of the product cannot be predicted. Measurements at the product are very prone to errors. The severity of the measuring error depends on the velocity difference between the belts, the surface texture of the belts and of the product (friction) as well as the sensor position.
oThe product is located entirely on the belt.
As of this period it is made that the product will move at the same velocity as the belt. Measurements of good quality can be performed on this product.
The influence of the product sensor and the product detection method on the measuring accuracy and the reaction time of the algorithm shall be presented in the following on the basis of two cases:
1. The sensor is located behind the belt at least by the length of the product.
Offsetting the sensor for at least the product length behind the beginning of the belt is the safest method. If the sensor is located far enough behind the beginning of the belt, it is made sure the measurements of the Touchprobe sensor on the belt represent the positions of the product on the belt. Front edge of the product, rear edge of the product and product length are correct. Thus, the measurement methods from ET_InTakeMode can be used. Correction motions can be executed. However, placing the sensor further onto the belt also has considerable disadvantages.
oThe change of a product onto the belt is only detected when the product has already completed the change. This means for the algorithm that it has only little time for the correction before the product changes to the next belt.
oFor the algorithm it is hard to determine whether a product is changing onto the belt.
oHow far the sensor has to be placed onto the belt is determined by the largest product. However, the correction distance then is not sufficient for the infeed performance, which usually must be higher for smaller products.
2. The sensor is located at the beginning of the belt.
By placing the sensor at the beginning of the belt it is achieved that the change of a product onto the belt is detected very early. This results in a long reaction time for the algorithm in which it can execute e.g. corrections. When detecting the front edge of the product, this, however, still is completely controlled by the previous belt. In the case of different belt velocities, it cannot be determined exactly at which position of the belt the front edge of the product will be located if it is completely controlled by the belt. An inexact measurement of the front edge of the product means that also the length of the product cannot be precisely detected. Thus, the measuring methods from ET_InTakeMode should be prevented at this point that determine the front edge of the product or its length. In this case, e.g. ET_InTakeMode.TpTrialEdgeSetLength should be selected. To this end, the length of the product has to be specified. If the length of the product cannot be defined, it is not possible to align the front edge of the products to the targets.
As in most applications a product flow of the same products are fed in, you can usually set the front edge at a fixed distance to the rear edge.
3. The sensor is located less than the product length behind the beginning of the belt.
This is a compromise between the two cases explained before. The sensor is moved as far to the beginning of the belt as to ensure that the algorithm has enough time to perform the required corrections. However, it is pushed so far onto the belt that the product is already controlled mainly by the belt when the front edge is detected.
Here, again only the rear edge of the product can be detected. However, since the product is already being controlled mainly by the belt when the front edge is detected, the measuring error is quite small. Thus, the product length and the front edge can be used with limitations.
In order for the product to be controlled mainly by the belt, it is necessary that the belt can apply a greater force to the product than the previous belt. In the case of homogenous products and belts, it can be assumed that this is True if slightly more than the half of the product is located on the belt.
