Hey all,
So I'm working on a device that's job is to measure a volume. For this evaluation we have known (calibrated/confirmed) volumes and we use the device we produce to measure the volumes (device reports a numerical volume measurement). We would like to use this test method to evaluate different configurations/versions of our device, so I now need to validate this method. First instinct is to throw the process and operators into a crossed Gage R&R however a number of things jump out at me.
1) With a Gage R&R this effectively makes our device the caliper in the caliper/gage block example. But I'm making the calipers, not the the blocks. It feels like the device you are evaluating shouldn't be part of the measurement system loop.
2) there are really 2 options for the test config.
3 operators, 5 devices, 3 reps and re-evaluate for each volume we measure.
3 operators, 1 device, 3 reps on 5 known volumes.
The first feels like the better option, as the second is really just evaluating one system, which works great in the caliper example, if you will use that caliper forever more, but our intent is to change the device.
If you remove our device from the problem completely you have a known volume, and an operator who has to read a numerical output off of a screen. Not much to validate there.
Is it still worth going down the road of Gage R&R? Or is there a different method by which we can evaluate our measurement system?
So I'm working on a device that's job is to measure a volume. For this evaluation we have known (calibrated/confirmed) volumes and we use the device we produce to measure the volumes (device reports a numerical volume measurement). We would like to use this test method to evaluate different configurations/versions of our device, so I now need to validate this method. First instinct is to throw the process and operators into a crossed Gage R&R however a number of things jump out at me.
1) With a Gage R&R this effectively makes our device the caliper in the caliper/gage block example. But I'm making the calipers, not the the blocks. It feels like the device you are evaluating shouldn't be part of the measurement system loop.
2) there are really 2 options for the test config.
3 operators, 5 devices, 3 reps and re-evaluate for each volume we measure.
3 operators, 1 device, 3 reps on 5 known volumes.
The first feels like the better option, as the second is really just evaluating one system, which works great in the caliper example, if you will use that caliper forever more, but our intent is to change the device.
If you remove our device from the problem completely you have a known volume, and an operator who has to read a numerical output off of a screen. Not much to validate there.
Is it still worth going down the road of Gage R&R? Or is there a different method by which we can evaluate our measurement system?