-----Original Message----- From: Alexandre Guimaraes To: PICLIST@MITVMA.MIT.EDU Date: Quarta-feira, 1 de Abril de 1998 19:05 Subject: Re: Overclocking PICs ie 16F84 >> From the manufacturer's point of view, it seems that as their >>process improves, ALL their chips would be able to do 20 MHz. If they >>mark them all for 20 MHz, there could end up being a shortage of the >>cheaper 4 MHz parts, possibly increasing their price to above that of the >>more abundant 20 MHz parts. I wonder if they just set all the ones that >>pass 20 MHz in a bin and mark them for whatever they have orders for. >> >>Harold > > > > How do they do this kind of tests on OTP parts that come programmed with >blank memory and RC oscilattor ?? They have a different point to inject the >clock ? Would just that be enough to check that the oscilattor would >start-up at that frequency ? That got me curious about the selection >process. I have used 4mhz parts ate 16 mhz also and had no apparent >problems. I prefer to use the correctly rated parts anyhow, if it fails it >is not my fault :-) > >Best regards, >Alexandre Guimaraes >alexg@iis.com.br > All ICs must have "testability" systems. Sometimes, they're internal pads that you inject some signal and then watch the signal flow; mostly, they use the exteriorized pads (aka the pins!). You can do it by "measuring" the critical path and creating some signals that excite them. The next step is to increase the frequency until the signal becomes unstable or wrong. The signal may, or may not, have some correlation with IC functions, since you're only trying to excite the critical paths. 4 MHz chips really come from the same waffer of the 10MHz ones. It's just some statistical testings that says "This lot works on 10MHZ, all the rest works 4MHz". The untested lots also become "4MHz" chips, since it is the minimum that the production process guarantees. Rafael Pinto