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

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
12/21/06 16:01
Read: times


 
#129911 - I'm a chip provider, not a user
Responding to: ???'s previous message
Kai said:

Ok, that means that you will test each part before using in a project?? Check all temperature, Vcc, capacitive load dependencies?? And you also make an accelerated aging test to see how the parameters drift??


Definitely, users should design to the data sheet. They should also pay their taxes on time, and not run red lights. But many users don't follow the rules. And that is why I have to determine what a part actually does, rather than just match the published spec.

Here is an example. I replaced the NXP 83C751, which has an I2C port on it. My first silicon matched the NXP specs, but was weaker on the I2C pins than the actual NXP parts were. A large customer had a mistake on one of their boards where they had used a 220 ohm pullup rather than a 10K one. As a result, my chip did not work, while the NXP one did. The resistor was on a separate board and in existing systems, so it was not possible to fix the resistor.

If I wanted the sales, I had to modify my part to match the actual NXP characteristics. And so I did.

List of 29 messages in thread
TopicAuthorDate
The 8255A is stronger and faster than reported.            01/01/70 00:00      
   The Data Sheet rules            01/01/70 00:00      
      How can we rely on that specification?            01/01/70 00:00      
   Not proven!            01/01/70 00:00      
   who give s hoot what you have 'found'            01/01/70 00:00      
      Data sheets don' rule in reverse engineering            01/01/70 00:00      
         Does that mean...            01/01/70 00:00      
            I'm a chip provider, not a user            01/01/70 00:00      
               then do notstate "The 8255A is stronger and fast"            01/01/70 00:00      
                  8255 statements            01/01/70 00:00      
         They do determine what you have to deliver ...            01/01/70 00:00      
            What I Did            01/01/70 00:00      
               Hmmmm ...            01/01/70 00:00      
                  Measuring output DC parameters            01/01/70 00:00      
                     What did you really test?            01/01/70 00:00      
                        DC measurements            01/01/70 00:00      
                           That's why measurements are quick and automatic            01/01/70 00:00      
                              My paycheck?            01/01/70 00:00      
   No!            01/01/70 00:00      
   How?            01/01/70 00:00      
      Measuring Specs            01/01/70 00:00      
   Population v sample sigma            01/01/70 00:00      
      Should know better            01/01/70 00:00      
         Dangerous extrapolations            01/01/70 00:00      
      The other side of the looking glass            01/01/70 00:00      
   The timing issue was the delay ....            01/01/70 00:00      
      Thanks for the tip            01/01/70 00:00      
      8255            01/01/70 00:00      
         do these two things            01/01/70 00:00      

Back to Subject List