Multi-fixtures in one CNC lathe - How to do Cmk study?

[Larry.Jiang]

Hopefully you will help me , thank you in advance

we have a CNC lathe within 8 fixtures, we can load 8pcs at one time , but there is some variations between fixtures.

now we are told to do cmk study, i machining 10 times , got 80pcs, then measure the 80pcs then calculate the Cm/Cmk, but the result is very bad, from the graph, we can see there is a big variation between fixture3 and fixture8.

now my question is : is the calculation strategy resonable?

if not , what is resonable calsulation strategy?

shoud we calsulate it by separate fixture?

 

[bobdoering]

Re: multi-fixtures in one CNC lathe-how to do Cmk study?

Hopefully you will help me , thank you in advance

we have a CNC lathe within 8 fixtures, we can load 8pcs at one time , but there is some variations between fixtures.

now we are told to do cmk study, i machining 10 times , got 80pcs, then measure the 80pcs then calculate the Cm/Cmk, but the result is very bad, from the graph, we can see there is a big variation between fixture3 and fixture8.

now my question is : is the calculation strategy resonable?

if not , what is reasonable calculation strategy?

should we calculate it by separate fixture?

I would calculate it by separate fixture. In essence, they act as individual "machines". Attempting to lump them altogether will give you a combined multimodal distribution, that has little value.

 

[Miner]

Re: multi-fixtures in one CNC lathe-how to do Cmk study?

bobdoering's response is correct. You can also take it one step further by performing a multiple comparison of means test such as Tukey's for all of the fixtures.

Fixtures that show a statistically significant difference are studied separately per bob's suggestion. Fixtures that do not show statistically significant differences may be pooled and analyzed together.

 

[bobdoering]

Re: multi-fixtures in one CNC lathe-how to do Cmk study?

I would be remiss not to mention the techniques described in Statistical process control for precision machining, which I would also review on a by-fixture manner. It will change the capability calculation approach you will end up using, however.

 

[bobdoering]

Re: multi-fixtures in one CNC lathe-how to do Cmk study?

bobdoering's response is correct. You can also take it one step further by performing a multiple comparison of means test such as Tukey's for all of the fixtures.

Fixtures that show a statistically significant difference are studied separately per bob's suggestion. Fixtures that do not show statistically significant differences may be pooled and analyzed together.

That may be good for snapshot, but until one knows the nature of the ongoing wear characteristics of I would analyze them individually. I would definitely do them individually as an ongoing control methodology, to catch special causes attributed to the individual fixture, and due to the variation of the fixture wear characteristics. I think it will end up being cheap insurance, especially for tight tolerances. You also need to be able to separate tool wear from any fixture effect, and doing the ongoing analysis by fixture will allow that.

 

[Larry.Jiang]

Re: multi-fixtures in one CNC lathe-how to do Cmk study?

That may be good for snapshot, but until one knows the nature of the ongoing wear characteristics of I would analyze them individually. I would definitely do them individually as an ongoing control methodology, to catch special causes attributed to the individual fixture, and due to the variation of the fixture wear characteristics. I think it will end up being cheap insurance, especially for tight tolerances. You also need to be able to separate tool wear from any fixture effect, and doing the ongoing analysis by fixture will allow that.

thanks first

but there is also another question if analysis each fixture or ongoing control for each fixture, the problem is :

when we load the pieces, we must number each part, then we can know this part was fixed on which fixture, it will waste lots of time to do this .

 

[bobdoering]

Re: multi-fixtures in one CNC lathe-how to do Cmk study?

thanks first

but there is also another question if analysis each fixture or ongoing control for each fixture, the problem is :

when we load the pieces, we must number each part, then we can know this part was fixed on which fixture, it will waste lots of time to do this .

It is not a waste of time if you get real data. I agree - it is not easy, but that does not make it a waste of time. Composite data of all fixtures will have so many variations mixed in (tool wear, fixture to fixture, part to part, etc.) that you can not make good decisions. And, good decisions are the point of the exercise.

BUT - the good news is if you control the process correctly (as in the previous link) you will have a very good idea what the frequency of these checks should be. That way even though it may be difficult, you will not have to do it any more frequently that absolutely necessary. The uniform distribution allows you to develop the most sane sampling frequencies of any distribution, and you can take advantage of that!