Posted by Bill Lash on August 07, 1999 at 14:28:12:
In Reply to: Diod Method A Myth? posted by Rob Caskey on August 06, 1999 at 23:32:12:
: me and a friend tried to get the diod method to work and it works just the same as without diods. We both think that the diods really dont help or hurt anything at all but add more expense & effort & that you need to try something else. Anyone here claim to have gotten diods to work?
There are really 2 things that could cause "masking" problems in a matrix. One is the external switch matrix, and the other is the chip in the keyboard controller, wich is usually a small microprocessor running some code to scan the matrix.
Without diodes, the processor doesn't have a chance, when it turns on one of the scan lines, and the right set of keys is pressed, it will see "confusing" information coming back. As saint said, see the writeup about this that he referenced.
With diodes, the information passed to keyboard processor chip is sufficient for it to do the correct thing. The question is, was the code on the keyboard controller written in such a way that it does the right thing. The other issue here is
that the code was written for application in a keyboard environment where you very seldom want multiple keys pressed at the same time (other than a few like shift, control, and alt). The designers of the code may be trying to make a good decision
based on the application. For example, in a tech note from National Semiconductor (see link below in PDF format), about how to make a keyboard using their COP-800 microcontroller, they actually have code to detect when a ghost key press would be generated, and then not generate
it. This is fine for a keyboard, but if you put the diodes in to get it right, this is bad behavior. Now I doubt that all keyboard decoders work this way, but it could be that the ones that you have tried do something like this.