Multiplexing, making sure only one transistor is on

[Ontaelio]

Hullo there!

I guess this stuff is common knowledge, but I wasn't able to find it googling.
Is there any _hardware_ way to make sure only one P-Channel MOSFET is conducting in any given moment?
Say, I have an LED matrix. Grounds are connected to the LED drivers, Vccs to P-Channel mosfets. Is it possible to design a board that will prohibit two or more of the MOSFETs to open up simuntaneously because the programmer screwed it? I know I can protect the LED drivers with N-mosfets, but this seems to be a bit excessive when talking about fool-proof. I have an itch there is a solution involving diodes or something, but am not able to nail it down. Please, anyone?

Thanks in advance!

[Signal32]

I can't think of any simple solution using diodes or similar, but one thing you can do is use an analog demultiplexer like CD4067 ro CD4097 and control that with the micro.
You select which of the 16 outputs is active on the CD4067 with 4 inputs. You control the 4 inputs using the micro. No matter what the micro will output, only one of the outputs of the CD4067 can be active.
Disadvantage is 80ohm internal resistance of the CD4067 + perhaps some latency = slower MOSFET switching speed.
Perhaps someone can suggest a digital equivalent to the CD4067 .

[Ian.M]

Sure.  Instead of running one P-MOSFET from each I/O line and requiring the programmer to only set one pin low out of N, use one or two 74HC138 active low 3 to 8 line decoder chips, with your MCU providing a binary MOSFET number and a strobe.  They have two active low enables (/G2A, /G2B) and one active high (G1), so you can use two chips for 16 lines, feeding an extra high address line to G1 on the high chip and G2B on the low chip.  However you do need an enable/strobe signal from the MCU to both chips /G2A pins so you can shut off all the MOSFETs to avoid glitches when changing the selected MOSFET number.

You could also do something with diodes and series resistors summing into an analog comparator.  If more than one gate is low, the comparator input would be pulled below its setpoint, and the comparator output could be used to reset the MCU, clearing the too many gates low condition.
 

[Ontaelio]

Thanks a lot, people! I'm on my way to dig the datasheets.

Just to clarify, though, I actually want to drive the matrix with the LED driver, no input from the MCU. Both X and Y axis; to make sure the switching takes place simuntaneously. Seems like the 74HC138 will not only do that, but also multiply Vcc sources, good. Or multiply LED driver outputs... maybe good too.

Anyways, on to datasheets. Thanks again!