B
Hey guys,
I'm sure this must have been asked more than once so apologies in advance if the answer is already available - I just haven't yet been able to find it.
I have a tool that inspects parts for defects. The tool can take up to 10 parts at a time, then inspects each one separately and fails any parts which contain a defect (visual inspection via camera) The parts can only be loaded once, after output it is not possible to load them into the input. However, once they are loaded, I can pass them under the inspection camera as many times as I need to.
As the inspection is a control, it needs to have MSA carried out. I was initially thinking of using Kappa analysis, but as I can only inspect 10 parts for the study, it feels as though there won't be enough data for reliable results. For example, I can run the 10 loaded parts twice round the system and record the pass/fail result for each pass per part. This always returns 100% K score. It just doesn't feel right - in production, the tool processes ~8000 parts per hour, so the 10 parts I load is not really fully representative of the product.
Can anyone offer some advice on a suitable MSA method for this?
Many thanks for reading.
BBz
I'm sure this must have been asked more than once so apologies in advance if the answer is already available - I just haven't yet been able to find it.
I have a tool that inspects parts for defects. The tool can take up to 10 parts at a time, then inspects each one separately and fails any parts which contain a defect (visual inspection via camera) The parts can only be loaded once, after output it is not possible to load them into the input. However, once they are loaded, I can pass them under the inspection camera as many times as I need to.
As the inspection is a control, it needs to have MSA carried out. I was initially thinking of using Kappa analysis, but as I can only inspect 10 parts for the study, it feels as though there won't be enough data for reliable results. For example, I can run the 10 loaded parts twice round the system and record the pass/fail result for each pass per part. This always returns 100% K score. It just doesn't feel right - in production, the tool processes ~8000 parts per hour, so the 10 parts I load is not really fully representative of the product.
Can anyone offer some advice on a suitable MSA method for this?
Many thanks for reading.
BBz