Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
02/22/11 11:57
Read: times


 
#181238 - Old fabs or old fab equipment
Responding to: ???'s previous message
The big boys sometimes sells their old fabs. And sometimes just sells their fab equipment. But the end result is the same. For a quite low cost, a new company can start producing 8051 chips based on large-geometry technology. A fab originally intended to produce memories or PC-class processors in older geometries suddenly gets the capacity to produce huge numbers of tiny 8051 chips.

Oliver Sedlacek said:
The costs are eye watering, and they only get about 9 months to recoup them! A 0.13um second hand fab is comparatively cheap.

The costs are only eye watering for the current technologies. But the eye watering fabs did leave 0.13u a long time ago. They are using 45nm, 32nm and working with 28nm and smaller. But next question is how many 8051 chips that have reached 0.13u? Think about the costs for a 0.25u or 0.35u fab. 0.35u was "cool" 2003 but such a fab now is way cheaper even if started from scratch with brand-new equipment. It isn't research-level technology anymore. And old fab technology for making 8051 chips don't have any 9 month recovery time. They can use the same technology to bring out cheap chips for 5 or 10 years, which is why it is hard for NXP to keep their existing older fabs running with a reasonable profit margin.

Remember that 1990, 0.8u was quite ok. By todays standards, the larger geometries hardly requires any clean rooms ;) So for 8051 chips that doesn't need really high clock frequencies or really low operating voltages or power consumption, a company with a 0.35u or 0.50u fab can bring out cheap processors from a cheap museum-technology fab and kill most of the profit options for the companies with the 0.18u or 0.13u or 90nm fabs. Unless the guys with the 0.13u fabs makes sure they only produce chips where they take advantage of their ability to throw in 10k extra transistors without significantly changing the yield. Or where they can take advantage of smaller geometries consuming less power or operating at really high frequencies. In the end, low-end 8051 chips gets pushed out because third-world ancient-technology factories can make them cheaper. And with time, smaller and smaller companies will be able to go from 0.35 and 0.25u and instead use 0.18 and 0.13u, making the competition every worse.

There really are big reasons why more "system-on-a-chip" solutions gets introduced. Microcontrollers will not automatically gain large profits from smaller geometries, so companies must add extras to make their chips stand out. So in with mp3 decoding, USB and an LCD interface to allow MP3 players with a chip and a serial flash. Or continue with video decoding to get USB disks cabinets that plays songs and shows photos and video stored on the disk.

Our standard microcontrollers will get more and more mixed up, which will make it harder to create products that are produced for many years without serious component end-of-life problems. All because it is easy to produce chips, while copyrights and marketing policies makes it hard for smaller companies to pick up licenses to continue producing chips after the original manufacturer drops them from from the product line.

List of 65 messages in thread
TopicAuthorDate
NXP suggests 32-bit ARM Cortex-M0 family for 8-bit replaceme            01/01/70 00:00      
   Funny indeed!            01/01/70 00:00      
      Rest of NXP's 8051 line to follow..?            01/01/70 00:00      
         comments            01/01/70 00:00      
            I've been watching them for 20 years now ...            01/01/70 00:00      
               So, what to do?            01/01/70 00:00      
                  Biting the ARM bullet            01/01/70 00:00      
                  SST89E58            01/01/70 00:00      
                  Well, if I had to do something ...            01/01/70 00:00      
                     Linux?            01/01/70 00:00      
                        Just a thought ...            01/01/70 00:00      
                     Leaping to Linux would be ludicrous!            01/01/70 00:00      
                        Unless...            01/01/70 00:00      
                           The target wouldn't necessarily be the host            01/01/70 00:00      
                              I don't ever build on target hardware unless target is a PC            01/01/70 00:00      
                        Supplement - not replace            01/01/70 00:00      
                           I can't disagree            01/01/70 00:00      
                              Competition always needed            01/01/70 00:00      
                              not everyone wants the 805x to survive            01/01/70 00:00      
                                 newer '51 releases            01/01/70 00:00      
                        It depends on your ultimate goal ...            01/01/70 00:00      
                     Real cheap ARM eval boards            01/01/70 00:00      
                        ADuC ARM            01/01/70 00:00      
                           Nearly, ADuC702X            01/01/70 00:00      
                        Yes! Lots of Real cheap ARM eval boards!!            01/01/70 00:00      
                           Why go cheap ...            01/01/70 00:00      
                              Prototyping can interfere with extras            01/01/70 00:00      
                                 "nfity" != "useful" or "helpful" (necessarily)            01/01/70 00:00      
                                 That's often a problem with EvK's            01/01/70 00:00      
                                    50/50 Agree/Disagree            01/01/70 00:00      
                                       Perhaps you're right about the second point            01/01/70 00:00      
      Cortex-M0s            01/01/70 00:00      
         "low end"            01/01/70 00:00      
            How low is "low" ?            01/01/70 00:00      
            money            01/01/70 00:00      
               Depends on view            01/01/70 00:00      
                  the small embedded view            01/01/70 00:00      
                     yes            01/01/70 00:00      
                        applications of low pin count            01/01/70 00:00      
                           Either help with real-time or with wire count/length            01/01/70 00:00      
               Money and technology            01/01/70 00:00      
                  ARM core already tiny enough that you gain no more            01/01/70 00:00      
                     Fab costs            01/01/70 00:00      
                        Old fabs or old fab equipment            01/01/70 00:00      
               Other Meanings            01/01/70 00:00      
   Colonial English            01/01/70 00:00      
      No new models            01/01/70 00:00      
         End of the roadmap            01/01/70 00:00      
   Anachronism            01/01/70 00:00      
      Quite common to extend meaning of old terms            01/01/70 00:00      
   a bit related            01/01/70 00:00      
      Doesn't add any advantage so totally cornered            01/01/70 00:00      
         16-bitters            01/01/70 00:00      
            I think you missed the point            01/01/70 00:00      
   NXP 8051s            01/01/70 00:00      
      you are a bit slow            01/01/70 00:00      
         you are a bit slow            01/01/70 00:00      
            lots of power needed to swing 5V devices            01/01/70 00:00      
               not just the swing            01/01/70 00:00      
                  The area myth gets busted            01/01/70 00:00      
                     Long time since chips started to get different scaling            01/01/70 00:00      
         Wide Vcc is growing trend            01/01/70 00:00      
            no such ceiling, just no avoidance            01/01/70 00:00      
      3V3 or 5V            01/01/70 00:00      
         Oxide thickness            01/01/70 00:00      

Back to Subject List