Qg(tot) max 33nC, at 4.5V. Which is equivalent to 7.3nF. 74HC logic is roughly 100 ohms output resistance, so will drive the gate in about 730ns. I wouldn't recommend a pin toggle shorter than about 30 microseconds.
Tim
That is fantastic!! Thanks very much!! To be 100% honest with you, I hadn't even considered the RC time constant at all! Very very helpful thank you! Can I please ask where the 30 micro seconds has come from?
1*RC is only 63% (assuming a real capacitor and resistor), so that's not enough. The gate is nonlinear -- at that point, it'll probably be stuck in the miller plateau, not actually 63% done. 2*RC is a good rough idea for "most of the way" -- it'll probably be fully on by then (past the miller plateau), if not at lowest Rds(on) yet. This is well enough the rise/fall time. Take this up by 10-20x to figure the maximum switching frequency with reasonable losses. More of a switching converter SWAG, but should be useful enough here.
Remember to estimate Cgss from Qg(tot) max. The value given for Cgss is usually under bias, and usually underestimates the charge by four times or so. The equivalence is: Q/V = C.
Downsides to shorter pulses: particularly if you're doing PWM, finite rise/fall time causes rounding off of very short pulses, skewing the visual response curve at the end points. So, a #010101 color might be fully black because it doesn't manage turn on at all, and the next few shades above that aren't spaced evenly because it's turning on slowly. Because the transistor doesn't switch fully on during these pulses, power dissipation is much higher -- again, a bigger hazard in switching converters, where the power transfer is typically many times the power dissipation capability of the transistor. (Case in point: I once had a 1kW bench supply, one of those $300 Chinese junkers. They used massive 50A IGBTs in the power stage, which were woefully underdriven by a few small transistors -- 2SC1815s no less! Needless to say, the pulse width at low voltage output was so narrow relative to the switching speed that the transistors never fully saturated, and BANG, your waste of $300 is now verified.) Since we're talking LEDs here, unless you plan on driving a massive number of them at once, or using very small transistors (like a SOT-23 for amperes!), this probably isn't going to be a concern -- you may even *need* some linear range for current control or something (though probably not on the row drivers..).
Tim
Home
Help
Search
Login
Register
