When to use diodes with switches?

I opted to move away from the MB6582 button-matrix design and, instead, use standard shift-registers mounted on my control surface boards. Less modules, less wires to connect between boards, etc. This is primarily because I am limited in Eagle Standard and can’t create a single control surface board (where a button-matrix would certainly make more sense).

Thing is, I’m so used to using diodes for buttons but with just using shift-registers, I assume that they are no longer required (but probably don’t hurt anything either)? I know the pull-up resistors are required to prevent phantom button presses and things but I assume the diodes are optional?

Also, when does an isolation plane make sense? How about a power plane? The bottom plane is for ground but most of the top wires are for power, so I was curious if I should use an isolation plane, or if a power plan would even make sense. I hate wasting copper :slight_smile:

For the curious, I’ve included the first board I switched over to using on-board SRs. There’s quite a few SMD parts on it but I figured that would make the layout easier, which seems to be the case (though I have more vias than I’d like).

EDIT: For to ask about grounding unused pins for the input SRs? Yes?

using signal diodes (1n4148) permits use of a matrix , i.e 64 buttons with only 8 outputs (1x74HC595) and 8 inputs (1x74HC165)

some complexe example here

more simple http://www.midibox.org/forum/index.php/topic,10476.msg79411.html

the idea is the following :

A column is ON, the others 7 are OFF. then we scan what Row is ON, corresponding to a button press.

the do it again with next Column, etc…

Though so, thanks for clarifying! This means, in my case, I won’t need diodes since I won’t be using a button matrix.