Hi Michael, Eric and all, Now that I've stopped rolling on the floor laughing and wiped the tears from my eyes, I'll continue. Laughing is better than crying. Do they an English translation for that book? A SPC book used the word 'management', Oh boy. If you subscribe to the theory that SPC is a means to end, a tool by which to improve processes and capabilities, putting up the control chart, tracking yields and responding to the out of controls is probably adequate to accomplish your goal. If you aren't going to make money using SPC why use it? If you have to have a Cpk number, why not reverse engineer it? After all, the Cpk value is only telling you what the expected yield of the process is. As an example, a process capability of 1.00 means that you have an expected defect level of 1,350 ppm on a one-sided specification (which is applicable to a yield attribute). Thus if your attribute control chart showed you had an average yield of 99.865%, your Cpk would be 1.00. Here's how I understand it, A high capability process should detect all process mean changes which produce out of spec. product. Otherwise, you will sell the customer stuff that doesn't meet her requirements and even worse not know which ones are bad. Cpk measures the distance between the process mean and both the upper and lower specification limit on a mean chart at half width. Cpk = lesser of (Upper Specification Limit - Process Mean)/3 Sigma or (Process Mean - Lower Specification Limit)/3 Sigma I think this makes the assumption that the random variations range only six standard deviations (Sigma) but it's been a while. Questions for Eric, Which processes are we talking about here? Are you already using SPC? Is the process in control? What are your spec. limits or are you trying to establish them? Hans Time to go home and get ready for the Hurricane! ~~~~~~~~ Hans Michel Hinners WR-ALC/LYPME Bldg. 640 Materials Engineer 380 Second Street, Suite 104 Manufacturing Eng. Sec. Robins AFB GA 31098-1638 912-926-1970 (Voice) 468 - 1970 (DSN) 912-926-7974 (Fax) [log in to unmask] ---------- From: Hinners Hans CIV WRALC/LYPME [SMTP:[log in to unmask]] Sent: Thursday, September 09, 1999 2:55 PM To: [log in to unmask] Subject: Re: [TN] Process capability Eric and all, How about tracking the yield through the process/machine over time - per shift, daily, weekly, whatever makes sense? Example, I have 95% yield so I know for every 100 scheduled/produced I can expect, on average, to get 95 good ones to go to the next step. I can't tell you which ones are good until I measure/inspect something. I think of process capability (or the index), Cp or Cpk, as a means to compare the ability of a process/machine to produce something within tolerance. It tells me whether I will go broke trying to make something. Hans ~~~~~~~~ > > What is the best way to determine the process capability of a machine in > terms of defective units produced? How can I say that a certain machine is > capable only of producing a certain no of good units everytime it is being > run? Any inputs will be highly appreciated. > > Eric > FAAI ############################################################## TechNet Mail List provided as a free service by IPC using LISTSERV 1.8c ############################################################## To subscribe/unsubscribe, send a message to [log in to unmask] with following text in the body: To subscribe: SUBSCRIBE TECHNET <your full name> To unsubscribe: SIGNOFF TECHNET ############################################################## Please visit IPC web site (http://www.ipc.org/html/forum.htm) for additional information. If you need assistance - contact Gayatri Sardeshpande at [log in to unmask] or 847-509-9700 ext.5365 ##############################################################