View Full Version : 100% continuous inspection question
elsmadnar 10th June 2009, 07:22 PM Hello,
I am new to the quality control industry, but I have a scientific background. We are running a plastic rod product with tight tolerances. The customer is demanding 100% inspection of all product. The product is wound onto various spools. We are collecting continuous sample data (2 diameter measurements per foot) and storing data for every spool that goes out the door.
Is 100% inspection QC referring to the fact that every product has been inspected or sampled to a degree, or does it mean that within each reel, 100% percent of the product is measured?
The former sounds correct to me, while the later sounds like increasing the sampling resolution (which is what the customer says is 100% inspection). Is this just semantics or is he correct?
He is now demanding a continuous light source to measure the product. What could he be talking about?
Thanks in advance!
Coury Ferguson 10th June 2009, 08:04 PM Hello,
I am new to the quality control industry, but I have a scientific background. We are running a plastic rod product with tight tolerances. The customer is demanding 100% inspection of all product. The product is wound onto various spools. We are collecting continuous sample data (2 diameter measurements per foot) and storing data for every spool that goes out the door.
Is 100% inspection QC referring to the fact that every product has been inspected or sampled to a degree, or does it mean that within each reel, 100% percent of the product is measured?
The former sounds correct to me, while the later sounds like increasing the sampling resolution (which is what the customer says is 100% inspection). Is this just semantics or is he correct?
He is now demanding a continuous light source to measure the product. What could he be talking about?
Thanks in advance!
Welcome to the Cove.
My interpretation of 100% inspection means that all of the manufactured product is inspected. Sampling is not allowed.
The Light Source they might be talking about is there are companies that make Laser Measuring Inspection equipment. They use laser to measure dimensions. If this is what the customer is requiring than I would think that they should either provide the Laser measuring equipment or add the cost/value to purchase the equipment as part of the Contract.
Miner 10th June 2009, 08:18 PM 100% inspection of a continuous product such as wire or cable is not a clear cut as it is for discrete product. In your case it needs to be defined in writing with your customer.
One statistical argument that you could make to define a part as something other than 100% coverage of the surface is to perform an auto-correlation analysis. Many continuous processes such as drawing or extrusion are auto-correlated, which means that observations are dependent on the observation immediately preceding it. The analysis will define how many time periods are correlated with each other. This time period could then be defined as a part.
However, if you can use continuous monitoring such as a laser mike with minimal cost impact, it may be an easier path to take.
Stijloor 10th June 2009, 08:21 PM 100% inspection of a continuous product such as wire or cable is not a clear cut as it is for discrete product. In your case it needs to be defined in writing with your customer.
One statistical argument that you could make to define a part as something other than 100% coverage of the surface is to perform an auto-correlation analysis. Many continuous processes such as drawing or extrusion are auto-correlated, which means that observations are dependent on the observation immediately preceding it. The analysis will define how many time periods are correlated with each other. This time period could then be defined as a part.
However, if you can use continuous monitoring such as a laser mike with minimal cost impact, it may be an easier path to take.
Could something like this (http://news.thomasnet.com/fullstory/501820)be applied?
Stijloor.
Miner 10th June 2009, 08:33 PM Could something like this (http://news.thomasnet.com/fullstory/501820)be applied?
Stijloor.
"Providing ±.1 mm accuracy, PM-300 Extrusion Profile Monitoring System checks multiple complex feature measurements at rate of 12 times/sec on inline extrusion production lines. Continuous monitoring occurs in real-time during manufacturing process, so every extruded portion is verified to be within spec."
For extruded products that I had dealt with, it definitely would suffice. Whether the OP's customer will be satisfied, is another question.
The interesting thing here is that it is really sampling every 0.083" (for a line speed of 60 FPM), so it is not really continuous monitoring.
elsmadnar 10th June 2009, 08:34 PM @Coury
Thanks for the reply. We are using a 2-axis laser micrometer 1200 scans/sec/axis (lump/neckdown detector) to QC reels when they are re-spooled from production reels to smaller reels, and a 3-axis laser micrometer on the production line (300 scans/sec/axis).
I am currently collecting serial data output from the 2-axis at a resolution of 2 scans/foot.
I do not batch sample the reels, meaning I don't scan one or two reels from a batch and then let the rest pass. Every reel is inspected using the 2-axis laser throughout its length. Isn't this 100% inspection? Increasing the resolution of samples throughout a reel (or product item) should be a different thing, right?
As for the continuous light source, I think he is just yanking our chain. He should know that we are already inspecting our product using laser micrometers, as we have told him.
Thanks!
elsmadnar 10th June 2009, 08:45 PM @stijloor + @minor
This is a neat tool as it would help with some profile issues that we have been having with our rod. Thanks.
Continuous sampling should mean that samples are taken at a regular interval (second, foot, etc). The smaller the interval, the greater the sampling resolution. Transducers, from what I know, provide the only means to provide continuous real-time analog data, however they suffer from noise and calibration issues. Correct?
Thanks
Stijloor 10th June 2009, 08:54 PM @stijloor + @minor
This is a neat tool as it would help with some profile issues that we have been having with our rod. Thanks.
Continuous sampling should mean that samples are taken at a regular interval (second, foot, etc). The smaller the interval, the greater the sampling resolution. Transducers, from what I know, provide the only means to provide continuous real-time analog data, however they suffer from noise and calibration issues. Correct?
Thanks
I have to pass that question on to my Fellow Participants here who are more familiar with that type of device. :(
Stijloor.
elsmadnar 10th June 2009, 09:36 PM Sorry, I didn't mean to scare you away with the word transducer:o
Transducer is just a generic term that converts one type of energy to another (light->electric, etc.). A laser mike has a transducer to covert light into electricity, which is then digitally sampled to a reading.
Laser micrometers based on their technology must sample at a certain rate. 100% inspection (in my customers eye's) means that every last drop of information (diameter) at every point along the rod would be examined. This is virtually impossible. It is like examining the rod at the atomic scale. I have been telling my boss that 100% inspection must mean that we examine every product that we produce as exhaustive as possible (in our case 2 samples per foot).
Thanks again.
Jim Wynne 10th June 2009, 10:02 PM I have been telling my boss that 100% inspection must mean that we examine every product that we produce as exhaustive as possible (in our case 2 samples per foot).
Welcome to the Cove. :D
Why would you not just ask the customer what they want?
elsmadnar 10th June 2009, 11:00 PM He wants 100% inspection with absolutely NO defects at ANY length on a plastic rod- absolute perfection. We can only sample a moving line so many times per second. He is telling my boss that we need to install a "continuous light source" to inspect the rod. We are currently using 2 different lasers, so I am interested in what he actually means.
Truth is we finally got a reel returned and I pulled up the data on the reel and it was right on spec. I re-sampled the reel and got the same results. Due to the sampling rate, there is a possibility that a small defect got past without being defective.
Before I started working for this small business, they did not keep records of their data and did no analysis. They just had an operator watch numbers fly by on a screen and if he saw any out of spec they stopped and started a new reel. I had an opportunity to sample reels that were produced in this manner and there were HUGE defects that were being released. No complaints. Though there is a small chance that a defect caused the problem for the client, it seems like user error. Of course, my boss doesn't want to hear that.:bonk:
harry 10th June 2009, 11:11 PM He is telling my boss that we need to install a "continuous light source" to inspect the rod.
Are they referring to: Automated optical inspection
elsmadnar 10th June 2009, 11:20 PM I did think that something like that would help detect inclusions (carbon deposits inside the plastic) and color defects with the rod. But would it not be a big investment? And he is mainly concerned with an absolute outer diameter. The rod rotates as is is pulled down the line, requiring at least 2 2D cameras.
thanks
brahmaiah 11th June 2009, 02:31 AM Hello,
I am new to the quality control industry, but I have a scientific background. We are running a plastic rod product with tight tolerances. The customer is demanding 100% inspection of all product. The product is wound onto various spools. We are collecting continuous sample data (2 diameter measurements per foot) and storing data for every spool that goes out the door.
Is 100% inspection QC referring to the fact that every product has been inspected or sampled to a degree, or does it mean that within each reel, 100% percent of the product is measured?
The former sounds correct to me, while the later sounds like increasing the sampling resolution (which is what the customer says is 100% inspection). Is this just semantics or is he correct?
He is now demanding a continuous light source to measure the product. What could he be talking about?
Thanks in advance!
AchievING 100% acceptability(About PPM=0<500) Is impossible through inspection only.There is nothing called 100% Inspection in practice.Human efforts however hard they are, they will be imperfect.Automatic inspection as used in bearing industry is the solution for your problem.To get the right answer to your question, following information is required:
1.What is the size and tolerance of the part.
2.What is the process chart and mfg.equipment and methods.
3.What PPM rejection OK for the customer
4.What PPM inplant rejection is accetable to the supplier.
Onthe outset I can suggest following method:
1.Before quantity production Follow correct set-up procedure with large samples.
2.Follow SPC and introduce control chart after training operators on SPC.
2.Carry out 1/2 hourly inspection by operator who will enter results on control chart and also correct the process when warranted.For this a thorough understanding of SPC is required by the operator.Frequent and unwarranted adjustments to process should be avoided.
4.Introduce hourly patrol inspection by q.c inspectors.
5.Carry out final inspection with large sample size.
In any case Process process control is the key for your problem.The solution is not 100% Inspection.
V.J.Brahmaiah
elsmadnar 11th June 2009, 05:17 PM AchievING 100% acceptability(About PPM=0<500) Is impossible through inspection only.There is nothing called 100% Inspection in practice.Human efforts however hard they are, they will be imperfect.Automatic inspection as used in bearing industry is the solution for your problem.To get the right answer to your question, following information is required:
1.What is the size and tolerance of the part.
2.What is the process chart and mfg.equipment and methods.
3.What PPM rejection OK for the customer
4.What PPM inplant rejection is accetable to the supplier.
Onthe outset I can suggest following method:
1.Before quantity production Follow correct set-up procedure with large samples.
2.Follow SPC and introduce control chart after training operators on SPC.
2.Carry out 1/2 hourly inspection by operator who will enter results on control chart and also correct the process when warranted.For this a thorough understanding of SPC is required by the operator.Frequent and unwarranted adjustments to process should be avoided.
4.Introduce hourly patrol inspection by q.c inspectors.
5.Carry out final inspection with large sample size.
In any case Process process control is the key for your problem.The solution is not 100% Inspection.
V.J.Brahmaiah
There are two reasons for a 100% inspection on this product: we have not yet obtained a high enough Cpk (around 1) and any slight defect can cause catastrophic problems. There is a big problem with our customer as none of that is set in stone yet, but my boss' have decided to run the product anyways. I am also the only person employeed to both respool and qc the product as we are a small business - I am the QC department:lol:
We are attempting to improve the process, but we have been supplying the product to the customer for about a year with few complaints (He is the distrubtor of the product). As far as he is concerned, as it seems, he wants ABSOLUTELY NOTHING to pass, and has not supplied a PPM. Wouldn't that be great. :bonk:
:mad:He said we should be using a "continuous light source" to measure our product. I discovered today he means a Keyence single-axis micrometer (80us scan time). We are examing the product using both a 3-axis and a 2-axis laser, and he knows that we have them. He should just say we need to be sampling at a higher resolution (we are currently at a 416 scan rate w/ our 2 axis). I think he thinks we are stupid. I believe he is also using the term 100% inspection incorrectly, as is the reason for this thread. I would like to think that 100% inspection - inspecting every product instead of sampling just some of the product - and sampling rate are independent.
Thanks and sorry for the rant.
|
|