- This topic has 6 replies, 5 voices, and was last updated 16 years, 5 months ago by Anonymous.
April 12, 2004 at 5:08 am #35203
AshishParticipant@Ashish Include @Ashish in your post and this person will
be notified via email.
Is anyone using Rolled Thruput Yield for reporting manufacturing performance at his plant. If yes please let me know what problems they faced(if any) in convincing Managers that RTY is the right way of depicting Plant Yield.
Thanks in adavance
ash0April 12, 2004 at 3:01 pm #98360
DarthParticipant@Darth Include @Darth in your post and this person will
be notified via email.
We are using FPY, RTY and Traditional Yield. Of course, some leadership love the concept of Traditional Yield since there is no penalty for rework and hidden factory. Nobody likes FPY because it makes them look really bad. So the compromise is RTY which gives them credit for reworked items but also penalizes them and accounts for the hidden factory.0April 29, 2004 at 11:52 am #99398
We have moved some of our lines to RTY. It can be challenging however. I think there are two aspects to the challenge. The first is that RTY is harder to understand than other measures. We used to measure defects as an absolute measure. This was easy but obviously not all that useful as volume fluctuated. Nevertheless, the average person can understand the concept of a defect easier than the concept of RTY. We have started to move through this by explaing RTY and its relation to defects. Today, we post both. I suspect however that most still don’t understand RTY. It just takes time and repetition. Even today, I repeat the definition every time I present it to other managers.
The second problem is the RTY makes things look bad. While it makes inherent mathematical sense, the result is often lower than what people expect or ‘want’ to see. As a result it is hard to paint a good picture from early RTY attempts. People do not want to present data that makes them look bad and RTY, when it is not well understood, is perceived to do that.
Hope that is of some value. Good luck.
Ty0April 29, 2004 at 12:37 pm #99403
Andrew BrodyParticipant@Andrew-Brody Include @Andrew-Brody in your post and this person will
be notified via email.
What is “RTY” and how does it work. I have been given significan responsibiity for our company’s yield hence inventory levels,
Andy0April 29, 2004 at 12:40 pm #99405
High first-time rolled yield is what made Motorola waferfabs great during the 1980s.
In 1990, A MOS 8 manager hired a device engineer from a well-known company in the South West and he told him that MOS 8 made four times as many microprocessors – same complexity device, same facility capacity – as his previous company.
One of the problems with counting defects is that they are not often random and independent, and finding ‘common causes,’ or ‘systematic cause’s of defects’ can be very difficult unless one has the right approach.
Dr. Harry’s version of Six Sigma does not have the right approach.0April 29, 2004 at 1:02 pm #99407
Andy U, am I to understand from your post that we should
not count defects unless they are random and fully
independent? My understanding is that -ln|RTY|=TDPU.
This seems to make perfect mathematical sense. Can
you explain further? Seems if we take your approach
nothing will ever get reported in terms of defects. Seems
that such a purist approach will shut things down. So
what is the alternative if we should not count defects?0April 29, 2004 at 2:47 pm #99418
Good counterpoint Ted … my answer is ‘count defectives’ …
If your mathematical formula makes perfect sense to you, then by all means keep on using it. However, I would’nt for two reasons:
– My first reason is that it is not even a good model for random defects.
– My second reason is that it does not include a systematic portion.
If you doubt this you can easily test this yourself, and I have already provided Matt with an explanation how to do it.
Why is this important? Well what happens if a manufacturer of PCBs runs the wrong solder flow program and creates thousands of bridging areas on a PCB? How do you know what caused the defects? Do you count all of them? How can this situation be reported honestly?
The previous cavity example is another process that suffers the similar problems. What if a clamp is loose and most of the parts have a ‘flash’ defect?
What are you going to do when the CEO looks at your DPU and tells everyone that it makes more economic sense to manufacture in China because your Process Z is so bad. Think it can’t happen? I know of a case where it did.
The forum ‘General’ is closed to new topics and replies.