Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
12-11-2010, 07:40 AM - 1 Like   #61
Veteran Member
Tom S.'s Avatar

Join Date: May 2008
Location: S.E. Michigan
Photos: Albums
Posts: 4,317
QuoteOriginally posted by falconeye Quote
Interesting aspect.

But I think it is more simple. After every so many manufacturing steps, the probablility of a fault has multiplied enough to be significant. And if you don't inspect at that stage, you risk to produce faulty end products.

The most cost effective was is to inspect when probablilities have risen enough.

And there is no way to get rid of the inspections. Just add enough manufacturing steps ...

E.g., in chip production, yield management is at the core of the process. I.e., you optically search for faults on the wafer before packaging into chips.

I cannot imagine that there is a Japanese and American way to it. The money counts the same both sides of the Pacific

A quick lesson in history! After WWII, the US began in the rebuilding of both Europe and Japan. When Japanese manufactures began producing, they quickly got a reputation for two things: cheap prices and junk (ring a bell with China today?). The Japanese turned to Dr. Deming to learn how to stop making shoddy products. Deming taught them that quality doesn't cost, it pays. The way it pays is two fold: it actually lowers manufacturing costs by eliminating needless inspections and waste, plus it enhances the public product perception and garners more sales. The Japanese literally took this to heart, designing their processes to eliminate the need for inspections by incorporating systems that made defective items so infrequently, inspection was not needed.

Think of it in these terms: what is better, a system that makes 5 defects out of every 100 parts, requiring constant monitoring and inspections, or a system that makes 5 defects out of every 100,000 parts. In the later system, the end user becomes the 'inspector' at the extremely low cost of the replacement of the 5 defective items.

Whether you agree with the philosophy or not, it's how things operate there, and has been in the process of being adopted worldwide for at least 30 years.


All of this is not to say the Japanese do no inspections, because they do. However instead of inspecting individual parts, the inspect the end result. In our conversation / regard, that would be the K5. So, they pull one off the line and check it thoroughly for all aspects of function. If it passes, they may not check another one for the next 1,000 (or more) units. Playing the lottery for a minute, if they check a bad one, then inspection intensifies to see if there are more or if they merely got the "one out of 100,000". If on the other hand, they pull a good sample from the que and it checks out OK, they don't check another for 1,000 units. That means it's conceivable a run of at least 2,000 could have undiscovered issues, and this could escalate to higher numbers if by chance the next inspection reveals another good unit. At some point however, end user reports start to surface indicating an undiscovered problem. Those end user reports are what we are seeing now. The same thing applied to the Toyota throttle issue, and many other problems that lead to the product recalls we have all become familiar with.

Did I say 'quick'? Sorry for being so long winded, but having worked in manufacturing for almost 40 years, I often take things for granted that others may not fully understand, and I wanted to make sure you understood the point I was trying to make!

12-11-2010, 08:06 AM   #62
Veteran Member
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,871
QuoteOriginally posted by NaClH2O Quote
the sensor subassembly.
Sony provides the sensor which is a ball grid ceramic chip with a cover glass on the upper side. No sensor subassembly. Before anything else, Pentax must soldier it onto a PCB which happens to be Pentax own SR daughterboard.

QuoteOriginally posted by Tom S. Quote
a system that makes 5 defects out of every 100,000 parts.
Tom, I think I get the idea. Nevertheless, we all speculate about the fault rates at every manufacturing step.

If y_i is the individual yield rate of manufacturing step i, the total yield y rate (after skipping n inspection steps) is
y = y_1 * y_2 * ... y_n
and in order to keep y high you must not increase the number of steps n you skip in inspection beyond a certain threshold. I agree, a low y_i eliminates the need to inspect every step. But it doesn't eliminate the need to inspect some intermediate steps.

And without any hard data, we are speculating here. Also, it has nothing to do with cultural differences or what one American consultant once said. After all, it is a simple optimization problem and money urges to find a good solution to it.

I only wanted to say in my earlier post that the final assembled part "SR unit" would make for an ideal candidate for an automated assembly line inspection. And the string of pearls should have been detected at that stage as dust contamination is a likely fault at that stage. If you ask me. What obviously nobody does

Last edited by falconeye; 12-11-2010 at 08:12 AM.
12-11-2010, 07:28 PM   #63
Veteran Member




Join Date: Sep 2007
Photos: Gallery
Posts: 969
while i am no manufacturing guru, i do notice that this "lean" concept has bean making a big impression in the west (and in corporate america, most predominantly). my experience is that it's generally misunderstood and grossly oversimplified (it does often come down to "we should not need to test if we do things right"). it is also my observation that the result of such utopian simplifications can have catastrophic (if at times funny) results. so i would urge people not to be naive about this.

i think falk put it very clearly: it is at the end of the day a mathematical truth: once you process is complex enough, the variation from the required end result can be huge due to very small variations in initial "input" (it's a bit simplified, but should get the message through). once you understand this, you will understand that, unless what you're making is something like spoons, you must assume things will go wrong. when some of the testing becomes redundant, you must be doing something right, but redundancy is good, so you won't stop testing, you'll keep the feedback loop running.

imho, the "let's just get it right, and not bother testing after" (blind faith) only works for cheap, disposable consumer items, which cost more to ship than they cost to produce. for those, you can just overstock your distributors by, say, 10 percent (at no extra cost to them), and accept "no questions asked" replacements, on the spot. it doesn't work so well if you try to apply it to 1k+ bucks dslr's and such, though
12-11-2010, 09:58 PM   #64
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 11,250
Also, let's not forget that cameras need quite of calibration work done to them. Doing some tests on the side over and above what needs to be calibrated, should almost be free.

12-11-2010, 10:14 PM   #65
Veteran Member
johnmflores's Avatar

Join Date: Apr 2009
Location: Somerville, NJ
Photos: Gallery | Albums
Posts: 5,361
Original Poster
Interesting discussion, fellas, but you've completely hijacked the thread LOL!

I wish there was a way to break off a thread into a new thread when the discussion diverges in an interesting but different direction...
12-12-2010, 03:19 AM   #66
Veteran Member
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,871
QuoteOriginally posted by johnmflores Quote
Interesting discussion, fellas, but you've completely hijacked the thread LOL!
No hijacking. We're just sitting in the thread's waiting room and have a small talk. Until the "real photos" start to come in
12-12-2010, 08:49 AM   #67
Veteran Member
Tom S.'s Avatar

Join Date: May 2008
Location: S.E. Michigan
Photos: Albums
Posts: 4,317
QuoteOriginally posted by johnmflores Quote
Interesting discussion, fellas, but you've completely hijacked the thread LOL!

I wish there was a way to break off a thread into a new thread when the discussion diverges in an interesting but different direction...
Yes, part of growing old means your mind wanders off topic a lot! I apologize!

12-12-2010, 09:13 AM   #68
Veteran Member
johnmflores's Avatar

Join Date: Apr 2009
Location: Somerville, NJ
Photos: Gallery | Albums
Posts: 5,361
Original Poster
QuoteOriginally posted by falconeye Quote
No hijacking. We're just sitting in the thread's waiting room and have a small talk. Until the "real photos" start to come in
Good one Falk! LOL!
12-12-2010, 11:37 AM   #69
Site Supporter
Site Supporter




Join Date: Sep 2006
Location: Bronx NY
Photos: Gallery | Albums
Posts: 5,631
Nice summation Tom

QuoteOriginally posted by Tom S. Quote
A quick lesson in history! After WWII, the US began in the rebuilding of both Europe and Japan. When Japanese manufactures began producing, they quickly got a reputation for two things: cheap prices and junk (ring a bell with China today?). The Japanese turned to Dr. Deming to learn how to stop making shoddy products. Deming taught them that quality doesn't cost, it pays. The way it pays is two fold: it actually lowers manufacturing costs by eliminating needless inspections and waste, plus it enhances the public product perception and garners more sales. The Japanese literally took this to heart, designing their processes to eliminate the need for inspections by incorporating systems that made defective items so infrequently, inspection was not needed.

Think of it in these terms: what is better, a system that makes 5 defects out of every 100 parts, requiring constant monitoring and inspections, or a system that makes 5 defects out of every 100,000 parts. In the later system, the end user becomes the 'inspector' at the extremely low cost of the replacement of the 5 defective items.

Whether you agree with the philosophy or not, it's how things operate there, and has been in the process of being adopted worldwide for at least 30 years.


All of this is not to say the Japanese do no inspections, because they do. However instead of inspecting individual parts, the inspect the end result. In our conversation / regard, that would be the K5. So, they pull one off the line and check it thoroughly for all aspects of function. If it passes, they may not check another one for the next 1,000 (or more) units. Playing the lottery for a minute, if they check a bad one, then inspection intensifies to see if there are more or if they merely got the "one out of 100,000". If on the other hand, they pull a good sample from the que and it checks out OK, they don't check another for 1,000 units. That means it's conceivable a run of at least 2,000 could have undiscovered issues, and this could escalate to higher numbers if by chance the next inspection reveals another good unit. At some point however, end user reports start to surface indicating an undiscovered problem. Those end user reports are what we are seeing now. The same thing applied to the Toyota throttle issue, and many other problems that lead to the product recalls we have all become familiar with.

Did I say 'quick'? Sorry for being so long winded, but having worked in manufacturing for almost 40 years, I often take things for granted that others may not fully understand, and I wanted to make sure you understood the point I was trying to make!
Function testing is a big part of our mfg process. Of course since our end product carries human passengers, it is a bit different than Tom's example above. It has to be because if a K-5 malfunctions you've got an angry customer, if a subway train malufnctions you can have dead passengers and crew. But the process is similar. We will take subassemblies and make sure they POST (power on self test, it can be done with a laptop and a power source) and then install them on the train, no exhaustive complete function testing is done. It is assumed that they will "work as designed". When a 4 or 5 car unit is finished the whole thing is basically function tested to make sure that it starts, stops, the doors work, the safety interlocks work, and the communications systems work. But the fancier stuff is not tested, it is assumed to work. Any problems with acceleration, dynamic braking, doppler sensing, FIND (fancy signs that tell you where you are and where you are going) and other communication systems are expected to be revealed during burn in testing where we basically dirve the train around the system with no passengers. After burn in the train is inspected again, downloads taken of the electronic systems (propulsion, braking, doors and communications). The train is also measured for possible misalignments (the NYC subway system is old and bumpy). If all those pass the train is put in service.

NaCl(the only truly exhaustive testing happens during burn in)H2O
12-12-2010, 03:24 PM   #70
Senior Member




Join Date: Oct 2010
Location: Adelaide, Australia
Photos: Albums
Posts: 250
To add my two cents...

When I was working in an Australian manufacturing plant (which shall remain nameless), my manager told us about Nokia's QC (Quality Control) process. Apparently (according to him) they do NO testing during manufacturing to any of their phones. Not sure whether they do a quick power-on test to some of the final products before they ship though. Our manager was trying to get us to go more down this track. More on that later. It seemed to me that this manager was saying they ship without ever having been tested in any way. Which is quite surprising considering the complexity of a modern phone (although it doesn't quite compare to a DSLR which still has more mechanical parts). Of course, as someone else pointed out, the result is a few out-of-box failures. These are quickly swapped. The problem is that now EVERYONE is inconvenienced by the failures (end-user, retailer, etc). The consumer becomes the QC technician. And we as consumers are fast becoming conditioned to expect this.

The above leans weight to what some have been saying about Japanese manufacturing philosophy. Make the product and process good enough and QC becomes irrelevant. But the question remains: Is it possible to get it right 100% of the time? I don't think so, when imperfect humans always have parts to play along the way. With this philosophy mistakes, i think, would tend to be big and expensive (as we have started to see from some of the big manufacturers).

Where I worked, on the other hand, we had almost the opposite philosophy. Every sub-board was tested before assembly, then the full assembly (every single one!) was tested to extremes of temperature. We manufactured commercial products which needed to be high reliability.

Unfortunately, there actually was a high failure rate of subassemblies and sometimes complete products as well. This necessitated the extensive QC. It became a continuing cycle. The products and manufacturing process were not up to par, so much QC was needed. But because the QC was in place Engineering didn't bother to fix the problems. It ends up with a lot of waste in both man-hours and materials.

So I see how the Japanese model of perfecting the product and process is beneficial, but i'm a bit sceptical about whether we can really expect to get so good that QC is irrelevant. I think the odds are stacked against us (which is also why I find it hard to believe in macro-evolution *puts on flame-proof pants*). I guess I tend to be over-conservative and tend toward perfectionism, but that doesn't lead me to believe in myself so much that I think I can create a perfect product like the Japanese seem to think. Redundancy and self-checking are my friends

Last edited by secateurs; 12-12-2010 at 03:37 PM.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, dslr, k-5, k-5 ii, k-5 iis, k5, pentax k-5, photos, shots, stains, test, videos
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Showing "Lens" Info in EXIF Data twokatmew Digital Processing, Software, and Printing 4 02-24-2010 01:27 PM
"Test" shots with my new DA*60-250 f4 Lance B Pentax SLR Lens Discussion 15 06-18-2009 12:58 AM
Vivitar "Series1" 70-210 f/3.5 Test shots hinman Pentax SLR Lens Discussion 11 02-06-2008 09:04 PM
First "real" photos taken with the K100D seymop Post Your Photos! 2 11-28-2007 07:19 PM
String stabilizer "monopod" ve2vfd General Talk 3 10-16-2007 07:20 PM



All times are GMT -7. The time now is 01:41 PM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top