What is the correlation of test equipment? 検査装置の相関とは

The test equipment must be reliable in its judgment results. These include repeatability, linearity and correlation. Correlation is measure of the reliability of the results, even when measured by different test equipment.

Correlation of the test equipment is simply an evaluation item that shows whether or not it shows a similar trend for the master equipment.

Examples of Correlation study

Here is one concrete example. This is the value of resistance and inductance of coil. The value of the master test equipment is recorded on the horizontal axis and the value of the slave test equipment on the vertical axis.

Test equipment on the master side is generally used in development centers, such as test benches. This is because they are considered to be the most reliable test equipment. The slave test equipment (the test equipment to be evaluated) is the test equipment for mass production facilities or new test equipment to be introduced.

※For mass production equipment, fine tuning of positioning and control is often difficult due to the simplification and automation of loading/unloading methods, and generally the inspection accuracy tends to be less accurate than that of test bench.

In this case, two items were evaluated simultaneously, but depending on the test equipment, the number of test items could be 10 or more. Correlation of all those items must be achieved.

Key points of Correlation study

Here are a few points for evaluation.
Correlation values can be easily calculated from two groups of data by using Excel's correl function.

How to select a sample, quantity

Depending on the item being tested, I recommend using about 30 samples to increase the reliability of the validation results. Also, samples that have wide distribution of data within specification should be used. A sample group with less data variability will not produce good correlation.

※Look at the above graph and consider the case where the data distribution is small. You can't run an approximation curve, can you?

How to do it when the master equipment is far away

The fact that there is no master test equipment at the same site does not mean that you do not have to perform correlation study. It is common practice to send samples tested at one site to the other site and compare the results of the two tests to confirm the correlation.

To avoid difficulties later, prepare samples for correlation study while the master machine is nearby and available. When the new inspection system is completed, the samples should be shipped to the manufacturer's site for initial evaluation.

Aligns the basic specifications of test system with master system

High correlation cannot be expected unless the equipment and control system used are identical. There are some engineers who decide the specifications of the inspection system without doing preliminary research, but they have a hard time after the inspection system is completed. It is so important to keep the basic specifications of the test equipment consistent with the master machine.

It may have to correlate with the customer's test equipment...

Depending on the product characteristics, it may be necessary to correlate with the customer's inspection system. However, as mentioned above, correlation evaluation is not so easy. It would be nice if the specifications of the customer's test equipment were similar to your own, but in most cases this is not the case.

If a problem occurs on a customer's production line with some characteristic, the reliability of your company's inspection process can be questioned. When these problems occur, the customer may come on board and it can be quite troublesome. If you have the right processes in place and you have the right test equipment in place, there's nothing wrong with that. If you have implemented poor process, you will have to start all over again.