It works!!!
I just got my new N channel FETs in, soldered them in place of the other ones I tried, and it works! No delay issues, just smooth power. Bench testing the setup with a Stage 1 motor, I could hold full throttle for a couple of minutes, and the heatsinks on the transistors are still cold to the touch, while the motor is warm.
So here are the transistors you need:
N channel 2x IRL3705N $2.60 each in single quantities
P channel 2x IRF4905N $3.21 each in single quantitiies
Total cost of the parts excluding the small circuit board (about $3) $11.60
Here are the specs...
IRL3705N:
VDS: 55V
ID (Cont): 77A
RDS(on): 0.01 ohms
Max power dissipation: 130W
IRF4905N:
VDS: -55V
ID (Cont): -64A
RDS(on): 0.02 ohms
MAx power dissipation: 150W
Compare these to the numbers on the IRF7389:
VDS: 30V
ID (cont): 7.3A (N channel) -5.3A (P channel)
RDS(on): 0.029 ohms (N channel) 0.058 (P channel)
Mx power dissipation: 2.5W
What does all this mean? Well it means that for the price of a 3x3 stack of IRF7389s, you can get the performance of a 10x10 stack*. Bring on the hottest motor you want, I think you'll find that the batteries will now become the limiting factor in how much current you can draw.
Oh yeah, I got all my FETs at
www.digikey.com
*Max power handling and current capability is 10x that of the IRF7389, but the RDS(on) value is only half to a third of that of the IRF7389. This means that you don't get as much of the benefit of the decreased resistance as you would get with a set of stacked IRF7389s. However, we are talking about really small amounts of resistance here, small enough that you should be worrying about the wires connecting to your motor more than the RDS(on) value.