Posted by Bill Lash on 22, 2000 at 7:30 PM:
In Reply to: Keyboard Blocking posted by Malberg on 22, 2000 at 1:47 PM:
: I noticed while I was laying out my matrix for my keyboard hack that it was a 16x8 matrix. Just for the sake of testing the matrix for ghosting, with my second identical keyboard from compaq, I picked any 3 points on the matrix that form a rectangle or square and pressed them. I was using the Ghostkey program which I downloaded from this site to test for ghosting. As I was pressing the keys one at a time and holding the previous key down, I was only able to get 2 keys of the rectangle to only show. And I tried many combinations and different rectangles, anything I tried I only got 2 keystokes to show. But If I pressed down any keys that didn't form a rectangle then I was able to hold down 8 to 10 keys and all would show in the Ghostkey program.
: Can anybody give me any suggestions why this is so???
: I thought that if you hold down 3 keys in a rectangle the forth key would automatically be triggered which is called ghosting!!! I can't give get 3 to show in a rectangle, can someone please explain!!!
: Regards -
This sounds like they are using a software workaround for the ghosting problem. The keybord chip is basically a little processor that scans the matrix and sends the data to the PC. Someone got clever several years ago, and decided that you don't need diodes in the matrix (the classic way of solving the n-key rollover or "ghosting" problem), if you detect that 2 keys of the rectangle have been pressed, you can have the software ignore any new corners of the rectangle that get connected until you drop back to only one or zero corners being detected. I see this method mentioned in an application note from National Semiconductor from 1989 for using one of their COP microcontrollers as a keyboard encoder. It seems that a lot of newer keyboards use this method, which makes diodes useless. I think there are some older keyboards that don't do this, and they are probably better to use for a hack.
You may have found a good way to determine if a keyboard encoder could benefit from diodes. If you actually see ghosting, and not the 2 key only behavior you see now, the ghosting can be taken care of with proper use of diodes. If you see the 2 key behavior, diodes won't help. You may be able to come up with a clever map for your controls to get enough free inputs though. Good luck.