Max fanout into MOSFET gates

Thread Starter

kyle_em_cee

Joined Aug 19, 2017
9
I am hoping to drive ~100 N-channel MOSFETs (https://www.diodes.com/assets/Datasheets/DMN63D8L.pdf) directly from a 3.3V logic-level PIC pin which can sink/source up to 18mA (PIC24FV32KA304). I would like to find out how many of these particular FETs I can drive with a since 18mA pin. These wouldn't be used as high-speed switches: they'd be ON or OFF for the duration of usage, never switching rapidly.

The gate resistance specified in the datasheet is at 1MHz and looks unhelpful for my purposes, since my switching frequency is effectively 0Hz. I would have expected a gate resistance in the M-ohm range. What I have done is to use the gate-source leakage current (Igss), which is listed as 10uA @vgs = 20V, and used those numbers to calculate a steady-state gate resistance of 2M-ohm. Then driving that gate at 3.3V, gives me a current of 1.65uA, or a maximum fanout of approximately 10909 of these FETs. Is this the correct approach?
 

WBahn

Joined Mar 31, 2012
30,060
For steady state that is probably a reasonable approach. You might want to consider the gate capacitance (which, if not given in the data sheet, you can estimate from the given gate impedance at 1 MHz) to estimate how quickly you can change the state of all of your transistors. Then consider if there are any potentially bad effects due to it switching too slowly.

For perspective, in on-chip CMOS designs, I've driven literally millions of FET gates with a single transistor in cases where the transition speed simply was a non-issue.
 

AlbertHall

Joined Jun 4, 2014
12,346
The MOSFET mentioned has a typical input capacitance of 23.2pF so a total of 2.3nF for 100 of them which is certainly not huge.
 

WBahn

Joined Mar 31, 2012
30,060
The MOSFET mentioned has a typical input capacitance of 23.2pF so a total of 2.3nF for 100 of them which is certainly not huge.
Agreed. That would be a transition time of well under a microsecond. In fact, the distributed R and C of the interconnect wiring might dominate.
 

Thread Starter

kyle_em_cee

Joined Aug 19, 2017
9
Thanks guys! Switching speed is a non-issue for this particular design, but it's nice to know that the transition time is so small.
 
Top