deglitching a clock

I agree, but the design team insisted that they needed the ability to reprogram the CPU's clock rate, "Just in case." Then to add insult to injury, they mounted the watch type tuning fork crystal vertically, and soldered the bottom edge of the crystal's case to a large pad rather than lay it down and use a drop of RTV to shock mount it. I saw a lot of dead boards with dead crystals. There were so many that I kept 10 spare crystals on my bench. They mounted the less sensitive baud rate crystal the right way, and soldered the top of the can to a pad. All of this was done a year before I was hired so I was stuck with getting the things out of production, and into customer's hands.

Hello John,

Yes, generally. But you need a negative supply and ideally a +8V or higher as well (really, really good bypasses or it'll oscillate). This is usually the case for the systems I work on but the analog folks are often suspicious about me tapping into "their" supplies for lowly digital purposes. Once I tell them that I am an analog guy myself that relieves them from their worries. Now, about that big honking stepper motor...

Regards, Joerg

We once soldered tiny wires to the vias on the *top* of a BGA package to get at some signals we'd thought we wouldn't need to get to on the prototype... that was fun :-)

Try XAPP609 and Answer Record 17697 for a description of local clock placement and routing restrictions (although I think XAPP609 only describes V2, the structure is similar but not identical in S3).

Brian

Thanks! I'll give it a thourough reading.