Reflow Oven Evaluation Using Cpk

Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: 1. Cpk's don't apply to ovens. 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. 3. Operating people never paid attention to Cpk's anyhow. What are your thoughts on this?

reply »

| Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | 1. Cpk's don't apply to ovens. | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | 3. Operating people never paid attention to Cpk's anyhow. | What are your thoughts on this? What else is new? Who should be responsible? Buyers and users of course. Oven, wave solder, and other solder, and some other, equipment and tools suppliers want to sell stuff. You're right, they don't want to get involved. They simply leave it up to users to discover good and bad, after the fact. They make it very hard to use SPC, let alone the capabilities profile part of the equation. Ironic they should recommend thermal profiles without knowing process capabilities. They also make it hard to manage processes instead of having to react to results as defect. How about that for concurrent engineering? capabilities. So on it goes.

reply »

| Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | 1. Cpk's don't apply to ovens. | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | 3. Operating people never paid attention to Cpk's anyhow. | What are your thoughts on this? Depends on the solder paste you use. Some solder paste is sensitive to changes in the oven parameters, some are not. The oven's job is to properly melt the solder and it's capability should be measure to what tolerance the paste can handle. It the solder paste can handle a 100% variance in the nominal profile parameters, why should the oven be held to a Cpk of 1 when a 2+ is just as adequate? If you properly profiled the product, the study should be what is upper and lower limits required to be out of control -- based on how the paste behaves, not the oven. There are some pastes I don't use just for this reason... Good Luck, Scott

reply »

| | Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | | 1. Cpk's don't apply to ovens. | | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | | 3. Operating people never paid attention to Cpk's anyhow. | | What are your thoughts on this? | What else is new? Who should be responsible? Buyers and users of course. | Oven, wave solder, and other solder, and some other, equipment and tools suppliers want to | sell stuff. You're right, they don't want to get involved. They simply leave it up to users | to discover good and bad, after the fact. They make it very hard to use SPC, let alone the capabilities profile | part of the equation. Ironic they should recommend thermal profiles without knowing process | capabilities. They also make it hard to manage processes instead of having to react to results | as defect. How about that for concurrent engineering? | capabilities. So on it goes. In defense of suppliers - how can they always know what screwy requirements we need. It's just that I believe they should be more involved at the design (prevent defect) level so it is easier to statistically characterize tools and equipment. This means capabilities as well as long term reliability so buying is better and easier justified.

reply »

| Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | 1. Cpk's don't apply to ovens. | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | 3. Operating people never paid attention to Cpk's anyhow. | What are your thoughts on this? Dave, You're wailing me with a college nightmare. I know you know what CpK is but for those looking at the forum who don't, it's a means of measuring how well your process is actually perform vs. what it is capable of. The formula for Cp is + or - 3sigma over the spec limit. CpK is the minimum value derived from one of the following two equations: + or - 3sigma over UpperSpecLimit - xbar OR + or - 3sigma over xbar - LowerSpecLimit. Sorry about the "sadistics" lesson everyone. If we think about what these formulae are implying there is no real way to use them on an oven, with any accuracy. We as buyers are going to determine our spec limits. The wider we set these limits the better our oven appears to perform. Ah, Motorola statistical magic at its' finest. That's not meant to be a dig against a fine company by any means. Every statistician designs their systems to say what they want them to say. The same applies Cp / CpK studies on ovens. Sure, you can create a test vehicle and profile that vehicle 100 times and you may get pretty close to the same result every time but how do you set a spec limit on this. Is + or - one degree a good spec limit, + or - 5 degrees, etc. Who knows? It's up to you and your process. If your processed materials can handle your ovens' drift, you will be fine. If not, you're in big trouble. I use this methodology as a means of determining when an oven needs PM. Flux build up will affect convection currents and thus product profiles. When the oven starts to drift, PM it. I wouldn't worry too much about zone temperature drift, this can be measured by profiling your product every so often. I'd worry more about the smoothness of conveyance. Vibrations in your oven can wreak havok on your process. Small vibrations can cause shorts up the wahzoo if they occur at the right point in an oven. I've lived through that one and it's no fun. Just make sure you profile product with temperature sensitive components frequently. We've all dealt with the parts that can only see 220 C for a max temp. Make sure any drift doesn't compromise that and you should be all set. I hope that helped a little. If not, sorry. Maybe it will be good for someone else. Good Luck and best regards, Justin Medernach Flextronics Int'l PIC East

reply »

| | Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | | 1. Cpk's don't apply to ovens. | | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | | 3. Operating people never paid attention to Cpk's anyhow. | | What are your thoughts on this? | Dave, | You're wailing me with a college nightmare. I know you know what CpK is but for those looking at the forum who don't, it's a means of measuring how well your process is actually perform vs. what it is capable of. The formula for Cp is + or - 3sigma over the spec limit. CpK is the minimum value derived from one of the following two equations: + or - 3sigma over UpperSpecLimit - xbar OR + or - 3sigma over xbar - LowerSpecLimit. Sorry about the "sadistics" lesson everyone. If we think about what these formulae are implying there is no real way to use them on an oven, with any accuracy. We as buyers are going to determine our spec limits. The wider we set these limits the better our oven appears to perform. Ah, Motorola statistical magic at its' finest. That's not meant to be a dig against a fine company by any means. Every statistician designs their systems to say what they want them to say. The same applies Cp / CpK studies on ovens. Sure, you can create a test vehicle and profile that vehicle 100 times and you may get pretty close to the same result every time but how do you set a spec limit on this. Is + or - one degree a good spec limit, + or - 5 degrees, etc. Who knows? It's up to you and your process. If your processed materials can handle your ovens' drift, you will be fine. If not, you're in big trouble. I use this methodology as a means of determining when an oven needs PM. Flux build up will affect convection currents and thus product profiles. When the oven starts to drift, PM it. I wouldn't worry too much about zone temperature drift, this can be measured by profiling your product every so often. I'd worry more about the smoothness of conveyance. Vibrations in your oven can wreak havok on your process. Small vibrations can cause shorts up the wahzoo if they occur at the right point in an oven. I've lived through that one and it's no fun. Just make sure you profile product with temperature sensitive components frequently. We've all dealt with the parts that can only see 220 C for a max temp. Make sure any drift doesn't compromise that and you should be all set. I hope that helped a little. If not, sorry. Maybe it will be good for someone else. | Good Luck and best regards, | Justin Medernach | Flextronics Int'l PIC East

Justin, Outstanding reply. However, how can we improve getting closer together so these questions become less frequent and important. Earl Moon Proof Of Design

reply »

| | | Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | | | 1. Cpk's don't apply to ovens. | | | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | | | 3. Operating people never paid attention to Cpk's anyhow. | | | What are your thoughts on this? | | Dave, | | You're wailing me with a college nightmare. I know you know what CpK is but for those looking at the forum who don't, it's a means of measuring how well your process is actually perform vs. what it is capable of. The formula for Cp is + or - 3sigma over the spec limit. CpK is the minimum value derived from one of the following two equations: + or - 3sigma over UpperSpecLimit - xbar OR + or - 3sigma over xbar - LowerSpecLimit. Sorry about the "sadistics" lesson everyone. If we think about what these formulae are implying there is no real way to use them on an oven, with any accuracy. We as buyers are going to determine our spec limits. The wider we set these limits the better our oven appears to perform. Ah, Motorola statistical magic at its' finest. That's not meant to be a dig against a fine company by any means. Every statistician designs their systems to say what they want them to say. The same applies Cp / CpK studies on ovens. Sure, you can create a test vehicle and profile that vehicle 100 times and you may get pretty close to the same result every time but how do you set a spec limit on this. Is + or - one degree a good spec limit, + or - 5 degrees, etc. Who knows? It's up to you and your process. If your processed materials can handle your ovens' drift, you will be fine. If not, you're in big trouble. I use this methodology as a means of determining when an oven needs PM. Flux build up will affect convection currents and thus product profiles. When the oven starts to drift, PM it. I wouldn't worry too much about zone temperature drift, this can be measured by profiling your product every so often. I'd worry more about the smoothness of conveyance. Vibrations in your oven can wreak havok on your process. Small vibrations can cause shorts up the wahzoo if they occur at the right point in an oven. I've lived through that one and it's no fun. Just make sure you profile product with temperature sensitive components frequently. We've all dealt with the parts that can only see 220 C for a max temp. Make sure any drift doesn't compromise that and you should be all set. I hope that helped a little. If not, sorry. Maybe it will be good for someone else. | | Good Luck and best regards, | | Justin Medernach | | Flextronics Int'l PIC East | | Justin, | Outstanding reply. However, how can we improve getting closer together so these questions become less | frequent and important. | Earl Moon | Proof Of Design Earl, That's a great question. These posts only stay up for 3 months. This is going to cause a cyclical trend of questions. Can we stop that from happening? I don't know. I use this thing to learn. I respond so frequently because people tend to respond back. I've only been in this business for two years. I worked at a couple of Coops when I was in college. I graduated last May and have been doing this full time since then. I don't have the experiential advantage of most of the guys that jump on this forum. By responding to the routine, sometimes I learn new things from other people's replies. I'm not really tired of responding, yet. I may get to that point some day but hopefully by then I'll have learned as much as I can. It is awfully frustrating sometimes to respond to the same type of questions, IE the people that grab you for your knowlege of PWBs. But you're extremely knowlegable in this area so you're a valuable resource. Not many engineers know half of what you do about the fabs and they can contribute tremedously to product defects. I guess we could write a book and tell everyone who jumps on this thing to buy it but then the knowlege exchange stops. I wouldn't learn anymore if I told people to just go out and by our book. Maybe more people should post technical papers on this site about the "heavy hitters." That would be one feasible way around all the repititious questions. For now though, I'm content in responding in hopes that I may gain something I am unfamiliar with. It happens all the time for me. Please keep responding, you are a very good resource for this community and your not participating would be sorely missed. thanks for all the info. Sincerely, Justin Medernach

reply »

| | | | Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | | | | 1. Cpk's don't apply to ovens. | | | | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | | | | 3. Operating people never paid attention to Cpk's anyhow. | | | | What are your thoughts on this? | | | Dave, | | | You're wailing me with a college nightmare. I know you know what CpK is but for those looking at the forum who don't, it's a means of measuring how well your process is actually perform vs. what it is capable of. The formula for Cp is + or - 3sigma over the spec limit. CpK is the minimum value derived from one of the following two equations: + or - 3sigma over UpperSpecLimit - xbar OR + or - 3sigma over xbar - LowerSpecLimit. Sorry about the "sadistics" lesson everyone. If we think about what these formulae are implying there is no real way to use them on an oven, with any accuracy. We as buyers are going to determine our spec limits. The wider we set these limits the better our oven appears to perform. Ah, Motorola statistical magic at its' finest. That's not meant to be a dig against a fine company by any means. Every statistician designs their systems to say what they want them to say. The same applies Cp / CpK studies on ovens. Sure, you can create a test vehicle and profile that vehicle 100 times and you may get pretty close to the same result every time but how do you set a spec limit on this. Is + or - one degree a good spec limit, + or - 5 degrees, etc. Who knows? It's up to you and your process. If your processed materials can handle your ovens' drift, you will be fine. If not, you're in big trouble. I use this methodology as a means of determining when an oven needs PM. Flux build up will affect convection currents and thus product profiles. When the oven starts to drift, PM it. I wouldn't worry too much about zone temperature drift, this can be measured by profiling your product every so often. I'd worry more about the smoothness of conveyance. Vibrations in your oven can wreak havok on your process. Small vibrations can cause shorts up the wahzoo if they occur at the right point in an oven. I've lived through that one and it's no fun. Just make sure you profile product with temperature sensitive components frequently. We've all dealt with the parts that can only see 220 C for a max temp. Make sure any drift doesn't compromise that and you should be all set. I hope that helped a little. If not, sorry. Maybe it will be good for someone else. | | | Good Luck and best regards, | | | Justin Medernach | | | Flextronics Int'l PIC East | | | | Justin, | | Outstanding reply. However, how can we improve getting closer together so these questions become less | | frequent and important. | | Earl Moon | | Proof Of Design | Earl, | That's a great question. These posts only stay up for 3 months. This is going to cause a cyclical trend of questions. Can we stop that from happening? I don't know. I use this thing to learn. I respond so frequently because people tend to respond back. I've only been in this business for two years. I worked at a couple of Coops when I was in college. I graduated last May and have been doing this full time since then. I don't have the experiential advantage of most of the guys that jump on this forum. By responding to the routine, sometimes I learn new things from other people's replies. I'm not really tired of responding, yet. I may get to that point some day but hopefully by then I'll have learned as much as I can. It is awfully frustrating sometimes to respond to the same type of questions, IE the people that grab you for your knowlege of PWBs. But you're extremely knowlegable in this area so you're a valuable resource. Not many engineers know half of what you do about the fabs and they can contribute tremedously to product defects. I guess we could write a book and tell everyone who jumps on this thing to buy it but then the knowlege exchange stops. I wouldn't learn anymore if I told people to just go out and by our book. Maybe more people should post technical papers on this site about the "heavy hitters." That would be one feasible way around all the repititious questions. For now though, I'm content in responding in hopes that I may gain something I am unfamiliar with. It happens all the time for me. Please keep responding, you are a very good resource for this community and your not participating would be sorely missed. | thanks for all the info. | Sincerely, | Justin Medernach Yeh, I know. Keep it up. Someday we'll all benefit, but this must continue to improve. Is there a better, faster forum, or educational system? Earl Moon

reply »

| | | | | Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | | | | | 1. Cpk's don't apply to ovens. | | | | | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | | | | | 3. Operating people never paid attention to Cpk's anyhow. | | | | | What are your thoughts on this? | | | | Dave, | | | | You're wailing me with a college nightmare. I know you know what CpK is but for those looking at the forum who don't, it's a means of measuring how well your process is actually perform vs. what it is capable of. The formula for Cp is + or - 3sigma over the spec limit. CpK is the minimum value derived from one of the following two equations: + or - 3sigma over UpperSpecLimit - xbar OR + or - 3sigma over xbar - LowerSpecLimit. Sorry about the "sadistics" lesson everyone. If we think about what these formulae are implying there is no real way to use them on an oven, with any accuracy. We as buyers are going to determine our spec limits. The wider we set these limits the better our oven appears to perform. Ah, Motorola statistical magic at its' finest. That's not meant to be a dig against a fine company by any means. Every statistician designs their systems to say what they want them to say. The same applies Cp / CpK studies on ovens. Sure, you can create a test vehicle and profile that vehicle 100 times and you may get pretty close to the same result every time but how do you set a spec limit on this. Is + or - one degree a good spec limit, + or - 5 degrees, etc. Who knows? It's up to you and your process. If your processed materials can handle your ovens' drift, you will be fine. If not, you're in big trouble. I use this methodology as a means of determining when an oven needs PM. Flux build up will affect convection currents and thus product profiles. When the oven starts to drift, PM it. I wouldn't worry too much about zone temperature drift, this can be measured by profiling your product every so often. I'd worry more about the smoothness of conveyance. Vibrations in your oven can wreak havok on your process. Small vibrations can cause shorts up the wahzoo if they occur at the right point in an oven. I've lived through that one and it's no fun. Just make sure you profile product with temperature sensitive components frequently. We've all dealt with the parts that can only see 220 C for a max temp. Make sure any drift doesn't compromise that and you should be all set. I hope that helped a little. If not, sorry. Maybe it will be good for someone else. | | | | Good Luck and best regards, | | | | Justin Medernach | | | | Flextronics Int'l PIC East | | | | | | Justin, | | | Outstanding reply. However, how can we improve getting closer together so these questions become less | | | frequent and important. | | | Earl Moon | | | Proof Of Design | | Earl, | | That's a great question. These posts only stay up for 3 months. This is going to cause a cyclical trend of questions. Can we stop that from happening? I don't know. I use this thing to learn. I respond so frequently because people tend to respond back. I've only been in this business for two years. I worked at a couple of Coops when I was in college. I graduated last May and have been doing this full time since then. I don't have the experiential advantage of most of the guys that jump on this forum. By responding to the routine, sometimes I learn new things from other people's replies. I'm not really tired of responding, yet. I may get to that point some day but hopefully by then I'll have learned as much as I can. It is awfully frustrating sometimes to respond to the same type of questions, IE the people that grab you for your knowlege of PWBs. But you're extremely knowlegable in this area so you're a valuable resource. Not many engineers know half of what you do about the fabs and they can contribute tremedously to product defects. I guess we could write a book and tell everyone who jumps on this thing to buy it but then the knowlege exchange stops. I wouldn't learn anymore if I told people to just go out and by our book. Maybe more people should post technical papers on this site about the "heavy hitters." That would be one feasible way around all the repititious questions. For now though, I'm content in responding in hopes that I may gain something I am unfamiliar with. It happens all the time for me. Please keep responding, you are a very good resource for this community and your not participating would be sorely missed. | | thanks for all the info. | | Sincerely, | | Justin Medernach | Yeh, I know. Keep it up. Someday we'll all benefit, but this must continue to improve. Is there a better, faster forum, or educational system? | Earl Moon None that I know of but I'll let you know if I find something. Justin Medernach

reply »

| Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | 1. Cpk's don't apply to ovens. | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | 3. Operating people never paid attention to Cpk's anyhow. | What are your thoughts on this? Running a Cpk on a reflow oven may not be the best way to assess your reflow furnace. A typical temperature measuring device has a resolution of 1C (or F), and a stated accuracy of +/- 1C. The thermocouples used are also not extremely accurate. However, I would assume that you are using a tolerance level around +/-5C. The resolution of the measuring system is 10% of your tolerance range. A large sample size is necessary to gather meaningfull data, because your standard deviation is going to be inflated if for instance the actual max temp on run 1 was read as 217.4C (217C), and the second reading was 217.6C (218C) etc... We continue to record Cpk on the belt speed of the oven, but have moved to a X bar and R chart because the Cpk on the oven never seemed to show a pattern, or gave us reliable data on the performance of the heating mechanism. One could run a Cpk on the process, but I could see why the equipment vendor might shy away from putting their stated capabilities to a Cpk. Hope this sounds reasonable. Steve Abrahamson MCMS

reply »

| Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | 1. Cpk's don't apply to ovens. | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | 3. Operating people never paid attention to Cpk's anyhow. | What are your thoughts on this? Monitoring a solder reflow oven by running profiles is tedious and time consuming. Each profile is like taking a picture of the state of the oven. It is assumed the oven hasn't changed between pictures, but you never really know. Our company makes a tool that installs in the oven permanently and automatically records a continuous video of the process. In this way we can effectively record the thermal profile for every product manufactured. In a typical reflow oven we install two 1/4" diameter tubes that run the length of the conveyor. Each tube has 15 evenly spaced thermocouples. The tubes are usually mounted above the conveyor just high enough so the boards can pass beneath them. The idea is that these 30 thermocouples are continuously monitoring the process temperature. You then run one product profile, and our software calculates the relationship between process temperature and product temperature for this "baseline profile". During production, the software continuously calculates how this product profile has changed based on measured changes in process temperature. We call this calculated product profile a "virtual profile". The virtual profile is typically updated every 30 seconds and is recorded permanently. The software allows the data to be replayed both forward and backward at different speeds with "VCR buttons". This system makes it much easier to calculate CPKs and other statistics on oven stability by doing much of the work for you. Check out our web site at www.kicthermal.com.

reply »

| | | Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | | 1. Cpk's don't apply to ovens. | | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | | 3. Operating people never paid attention to Cpk's anyhow. | | What are your thoughts on this? | Monitoring a solder reflow oven by running profiles is tedious and time consuming. Each profile is like taking a picture of the state of the oven. It is assumed the oven hasn't changed between pictures, but you never really know. Our company makes a tool that installs in the oven permanently and automatically records a continuous video of the process. In this way we can effectively record the thermal profile for every product manufactured. | In a typical reflow oven we install two 1/4" diameter tubes that run the length of the conveyor. Each tube has 15 evenly spaced thermocouples. The tubes are usually mounted above the conveyor just high enough so the boards can pass beneath them. The idea is that these 30 thermocouples are continuously monitoring the process temperature. You then run one product profile, and our software calculates the relationship between process temperature and product temperature for this "baseline profile". During production, the software continuously calculates how this product profile has changed based on measured changes in process temperature. We call this calculated product profile a "virtual profile". | The virtual profile is typically updated every 30 seconds and is recorded permanently. The software allows the data to be replayed both forward and backward at different speeds with "VCR buttons". This system makes it much easier to calculate CPKs and other statistics on oven stability by doing much of the work for you. Check out our web site at www.kicthermal.com. I like the thought of a permanent mole. Please send information. 385 Weddell Drive Sunnyvale, CA 94089

reply »

| | Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | | 1. Cpk's don't apply to ovens. | | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | | 3. Operating people never paid attention to Cpk's anyhow. | | What are your thoughts on this? | Running a Cpk on a reflow oven may not be the best way to assess | your reflow furnace. A typical temperature measuring device has | a resolution of 1C (or F), and a stated accuracy of +/- 1C. The | thermocouples used are also not extremely accurate. However, | I would assume that you are using a tolerance level around +/-5C. | The resolution of the measuring system is 10% of your tolerance | range. A large sample size is necessary to gather meaningfull data, | because your standard deviation is going to be inflated if for | instance the actual max temp on run 1 was read as 217.4C (217C), | and the second reading was 217.6C (218C) etc... | We continue to record Cpk on the belt speed of the oven, but have | moved to a X bar and R chart because the Cpk on the oven never seemed | to show a pattern, or gave us reliable data on the performance of | the heating mechanism. | One could run a Cpk on the process, but I could see why the | equipment vendor might shy away from putting their stated | capabilities to a Cpk. Hope this sounds reasonable. | Steve Abrahamson | MCMS This is very reasonable and should be passed on to those of us who did not completely understand. I will do this, but with the hope more suppliers do become more involved to make this explanation and provide better SPC.

reply »

| Cpk's on top flight ovens (e.g., Conceptronics, Heller, Electrovert, Vitronics) vary wildly, from 1 to 2+, between different supplers. The suppliers don't seem to understand Cpk's. A supplier's rep whose oven has a Cpk of 1.0 told me: | 1. Cpk's don't apply to ovens. | 2. Measuring them was just some rigamaroll that the QA people made the operating people do in evaluating ovens. | 3. Operating people never paid attention to Cpk's anyhow. | What are your thoughts on this? ************** Dear Dave, A capability analysis can be used for a reflow process (or another soldering process) but you need to have a good understanding of what you are doing. I agree completely with remarks made in some other replies about evaluating the capability of your measurement process rather than your reflow process. A lot of users tend to use the analysis without having made sure that the process is under statistical control. Also, many people evaluate the process variation against specification limits which are not in accordance with the basics of a capability analysis and its statistical theory. I have recently written a paper about this subject (you can have a copy if you send me an email) because there is a lot of misunderstanding.

regards, Frank J. de Klein, CQE,CQM

reply »