This entry was posted on Monday, January 23rd, 2012 at 11:55 pm and is filed under Social. You can follow any responses to this entry through the RSS 2.0 feed.
Both comments and pings are currently closed.
Yes! Preferrably to some thru holes that you can solder headers onto! My current boss makes boards with exactly the IOs he thinks he needs and then acts all surprised when he needs two more IO pins!
(Then he spins a new board with the two extra IOs and acts all surprised when it happens again! Argh! The fricking chip has hundreds of IOBs!).
Even if the pin has a dedicated use, like, say, it’s for the on-chip ethernet MAC that your project doesn’t need, it’d sure be nice if you broke it out.
This reminds me of the “wiggle room” that John R. Barnes recommends to be designed into circuits, in his book Robust Electronic Design ( http://www.dbicorporation.com/book-out.htm ). His goal is to help you get your board production-ready and robust in only 1 or two board spins. The idea is to make sure you _think ahead_ and design to allow for later fixes and other alterations. Kind of interesting. (I was lucky and got the big book for $8.00 from Biblio.com ( http://www.biblio.com/books/432689686.html ).
we used to have a standing rule that every board needs at least one spurious LED we don’t stuff them in production but the sure look purty – adding an LED to a gpio is sort of that last resort debug feature you need to get a working system up to the point here a debug serial console is working
Yup, pads and LEDs. I’ve been there many times. Sometimes not because I need one more I/O pin to drive something in my circuit, but to trigger a logic analyzer or somesuch. If there’s room, I’ll often include a DIP pattern labeled OSIF, for “Oh, S*** I Forgot”. In one board, I populated that with a GAL, and then proceeded to reprogram that poor GAL dozens of times and move the wires around nearly as often. Saved me dozens of board spins. And Colin, thanks for the pointer to the book!
Yes! Preferrably to some thru holes that you can solder headers onto! My current boss makes boards with exactly the IOs he thinks he needs and then acts all surprised when he needs two more IO pins!
(Then he spins a new board with the two extra IOs and acts all surprised when it happens again! Argh! The fricking chip has hundreds of IOBs!).
!
LOL
My boss stopped asking me what all these unpopulated headers are for. :) Well I guess sometimes even he uses them for his projects.
Greetings,
LOOOL
Nice one!
I’d go one step further.
Even if the pin has a dedicated use, like, say, it’s for the on-chip ethernet MAC that your project doesn’t need, it’d sure be nice if you broke it out.
This reminds me of the “wiggle room” that John R. Barnes recommends to be designed into circuits, in his book Robust Electronic Design ( http://www.dbicorporation.com/book-out.htm ). His goal is to help you get your board production-ready and robust in only 1 or two board spins. The idea is to make sure you _think ahead_ and design to allow for later fixes and other alterations. Kind of interesting. (I was lucky and got the big book for $8.00 from Biblio.com ( http://www.biblio.com/books/432689686.html ).
we used to have a standing rule that every board needs at least one spurious LED we don’t stuff them in production but the sure look purty – adding an LED to a gpio is sort of that last resort debug feature you need to get a working system up to the point here a debug serial console is working
Yup, pads and LEDs. I’ve been there many times. Sometimes not because I need one more I/O pin to drive something in my circuit, but to trigger a logic analyzer or somesuch. If there’s room, I’ll often include a DIP pattern labeled OSIF, for “Oh, S*** I Forgot”. In one board, I populated that with a GAL, and then proceeded to reprogram that poor GAL dozens of times and move the wires around nearly as often. Saved me dozens of board spins. And Colin, thanks for the pointer to the book!
Join the EFF.org bunny, and let them champion digital rights.