Schmidt

Hi rabaos,

it is quite difficult to do this since there is an extremely high product mix. What would you recommend me as a poka-yoke practice?
I am still thinking which type of sensor would be suitable.

Kind regards,
marcus

Just because of the fact you can not get negative values this is not a one sided application, only if the tolerance allows the radius to go to 0 you have a one sided application. If your tolerance is e.g 5 +0.06/-0 you still have a USL and a LSL….

Excellent! I will send in an hour. I went grocery shopping. I’m sorry for the delay.Marcus

Would you like me to send you the description through e-mail or to this site. I am trying to simulate jewelry boxes into a warehouse. If you have Promodel, I can send you my .mod file.
Appreciative of any help you can give,
Marcus

Unfortunately, the etiquette guidelines prevent me from doing that. I’m not sure if there is a work around. If you know of one, let me know.

marcus makaiwi
Verizon Business
2424 Garden of the Gods Road
colorado springs, co 80919
719-535-5260 work.
[email protected]

What I forgot .. if you have historical data, why do you want to do a machine capability study ? Why not simply calculate your process capability ?

You don’t need subgrouping to do a machine capability study. Sample size should be  > 50 of consecutive parts (100 recommended). But your data has to be normal !
For a preliminary process capability study , you have to use subgroups usually 3-5 of consecutive parts, and at least 20 subgroups.
 
 

If you have a customer using the AIAG Standard you are right…..
If you have a customer using VDA or ANFIA then :
Ppk is Critical preliminary process capability (should be > 1.66)
Cpk is Critical process capability  (should be > 1.33)
So what I learned for calculating process capability .. first ask your customer how he calculates, then do…[Read more]

Gage R&R compares the variation that comes out of your process with the variation that comes out of your measurement system. That means your samples for the gage R&R should be representative for your process, you should have the same(or a similar) standard deviation in your sample parts as in your process. If the variation in your sample lot is sm…[Read more]

Use a response surface design, you will find it in Minitab also under stat, doe, response surface
Marcus

I think the main problem you will have is that you are making different parts on different jigs. What you need to install is a good preventive maintenance of the jigs, and also standards for how jigs are loaded. Make sure that there are always the same amount of the same part per jig, the amount of jigs per carrier. Make sure that jigs are…[Read more]

I think Robert is right, without the assistance of the people working direct in the plating area it will be quite difficult to collect the data. But in my experience. if you talk to the peolple, explain them what you do and why, and that it is only for a defined period of time you want them to do this, most people are willing to help you.
I…[Read more]

If you do plating you probably use jigs with more than one part on it.Measure the rejects and try to stratify the data to jig, position of the part on the jig.Monitor the charge that has been applied to the jig, parameters like temperature, pH .. that can change quickly.
Concentration of your dipping solution should be quite stable, but you…[Read more]

Thanks all for your valuable advise.

Brian,
One thing that I would suggest that has really helped my distribution projects is a process modeling software program like ProcessModel or iGrafx. Usually the purchasing group does their buys on a schedule and you can model that as the input of your process, then specify the different floors/departments and consumption rates (if you have…[Read more]

OK, but what do you do if this produces a balanced scale?
The bad ball must then be found in one weighing from 3 UP balls and 2 DOWN balls.
Weighing 2 v 2 (UP +DOWN vs UP + GOOD) doesn’t work.  If UP + DOWN goes down, its either that DOWN or the opposite UP.
What am I missing?

I don’t thing anyone’s described the full solution yet!  I still haven’t quite got it figured it out.  Here’s the problem:Use Tony / Nayans first weighing (4 vs 4) and assume the pans don’t balance.  Then use their second weighing with 3 of the four “UP” balls and a “GOOD” against 2 “DOWN”s and 2 “GOOD”s.  Now there are only TWO outcomes here:…[Read more]

I think the reason is that GE employees want to be able to “stand out” as change agents in other organizations who are just starting their Six Sigma programs.  They won’t be just a number there, and they will be looked at as an opinion leader among many novices instead of an expert number among many other expert numbers. 
GE, Motorola, and Ho…[Read more]

Just create a pareto chart for each dimension you wish to track. 
Marcus