Help requested MOSFET driving "N" led array

[LabVIEWguru]

I was playing with the idea of using MOSFETS to drive a 2 or 3 dimensional array of LEDs. I may turn on 1 LED or "N" LEDs in different rows and columns. My concerns are physical size and cost. I can get a device from Maxim that is driven by SPI that will run 64 or 128 LEDs, but they are fairly expensive. I was thinking I could use a very inexpensive micro and a handfull of 8 cent small signal MOSFETS.

As long as I select my MOSFETS to turn on 100% at the gate drive voltage (3.3 or 5 volt) and the current capability of the (sot-23) MOSFET is high enough,

(1) is there any reason this won't work?

(2) Is there a better way to do it? (there is ALWAYS a better way....)

[solutions]

It'll work - sorta...just make sure you are able to turn your p-channels off (may not be possible if you use 3.3V gate drive and 5V LED supply). 

Also, go through what you want to do with various on/offs...for instance upper pchannel and left nchannel to turn on third led, AND lower pchannel and right nchannel to turn on LED7...as turning on only those two LEDS as your only intent ALSO happens to turn on two LEDs that I don't think you want on :-)

You are also wasting power in the resistors, BUT it is the simplest and cheapest way to do it IMO.  You could just use a 1 of 8 selector (TTL) and a counter (via GPIO or a chip/FPGA/ucontroller) and just crank your clock rate (and LED current) up to where it looks like LEDs are always on, when all you are doing is cycling through 8 of them round robin, turning on the CS- when you want it ON.

[tedz]

I am wondering a bit the "upper" transistors: In this configuration they are used as "source-followers", which significantly limits the output voltage - approximately by amount of their gate threshold voltage. By using p-channel transistors instead, they would be working better.  However, as solutions points out, gate drive for those upper MOSFETs in that case have to reach very close to the LED supply (+5V on the schematics) to secure turning off them.

I have in another project constructed something similar with bipolar transistors. In that case you would need PNP on top, and NPN on lower ones + proper resistors on bases. Again, driving the upper transistors need most thought even in that case.

[solutions]

LOL - I didn't notice the transistor polarity and just assumed they were p-channels.

What tedz said....

[pl4tonas]

Hi,
As mentioned by the other two guys, the upper Mosfet connected to the PAx inputs has to be a P-type one.  With logic level 1 you turn it off and with 0 on.
I suggest to build only one unit consisting of one P and one N mosfet in order to check if it works ok.  Possible problems could arise on th P-type mosfet where the on-off levels supplied by the micro controller might not be within the appropriate range (voltage levels for logic 1 and 0) that the mosfet needs.  

Also, take into consideration the fact that you may have all the PBx mosfets on an one PAx mosfet on.  In that case the current of all the LEDs from each PBx mosfet will sum to a single PA site mosfet.  Can the mosfet being able to handle the current for all these LEDs?

[Kombinator]

try IRLMLxxxxx