Hi, I am just trying to get my head around the current drawn from the ATMega2560, and the time this current is drawn for.
Using the RAMPS1.4 as an example, the digital pin is connected to the gate of a STP55NF06L via a 10ohm resistor. The data sheet says the input capacitance is 1700pF. Is the input capacitance referring to the capacitance of the gate? If so, that would make the tau constant for this pairing 17 microseconds, and the time to fully charge the gate as 85 microseconds. My issue is that for the initial start of the gate charging, the gate will draw 500mA from the 2560, although this will drop significantly within the first 20 or so microseconds. Is this really safe for the microcontroller, and if so, why is it ok to draw such current initially?
Obviously, if this 10 ohm resistor is increased, the peak current drawn will reduce, at the cost of longer switching time and more wastage via the junction. Clearly this is a trade off, but I'm not sure why this particular balance has been chosen.
17ns, not 17us. And in practice it would be about 85ns as the Arduino output driver has about 40 ohms
impedance, so about 50 ohms in total driving that 1.7nF. Using a 10 ohm here is dumb, its not doing
anything much and anyway the Arduino pin is being overloaded with more than 40mA flowing...
Normally for MOSFETs you get several capacitances listed such as gate-source, gate-drain. You will also
see something called "total gate charge" - this is the most useful for switching purposes, as it allows
for the highly non-linear behaviour as the channel forms. Say the total gate charge is 4nC at Vgs=5V and
Ids = 10A, then you can assume it will switch on in 4us with a gate current of 1mA (assuming
constant current). You could also divide the total gate charge by the gate voltage to get the effective
capacitance of the gate, here 800pF.
If you drive a lot of voltage and power (50V, 20A say) with a MOSFET then you must drive the gate
voltage with a low impedance source to prevent the gate-drain capacitance and the rapid dV/dt at
the drain terminal from back-driving the gate voltage out of spec. You would use a MOSFET gate driver
chip (that's what they are for after all!).
Gate resistors are usually unnecessary except to protect the driving circuitry from over current(*). However
if you parallel several MOSFETS, some gate resistance on each device protects against oscillation modes.
In old MOSFET circuits you'll see gate resistors everywhere, but back then MOSFETs had much higher
on-resistances and were commonly paralleled up for more current handling.
(*) and reducing EMI by slowing down switching. Yes, its a compromise.
Thanks, Mark, that all makes good sense to me. Two schoolboy errors that I need to catch myself on are forgetting to factor in the output impedance (embarrassing really, I had that drilled into me regarding op amps at uni last year) and using 10^-9 when calculating pF, not nF. I also now understand how a gate driver gives much better control regarding switching speeds which I can now research further.
MarkT did an excellent job of explaining the various dynamic parameters. The only thing I can add is this reference which has a deeper explanation of the dynamics of a Mosfet. If you read nothing else you should look at figure 13 to see the interaction of the charges with the gate voltage.
