Speeding of MOSFET turn-off

[uglyduck]

Guys, I need some help with my circuit which I have attached below...

My circuit is intended to be a very cheap solution of a transmitter-receiver channel - the receiver is comprised of R1 and D1 and the rest components are the transmitter.

The problem is with the gate voltage of the p-channel mosfet. It seems to me that I cannot turn off fast the N-MOS in front of it. When there are only the N-MOS (BSS138 and not 123 as in the schematic) and the R1 resistor everything looks fine - sharp edges, no delays. But when I add the P-MOS then things go wrong - simulations attached.

10kHz


100kHz
[url=http://obrazki.elektroda.pl/6087576300_1398939803.png]

1Mhz
[url=http://obrazki.elektroda.pl/1883380800_1398939831.png]

I need to switch this circuit with up to 1MHz. As can be seen the N-MOS is ON very fast and than with the increase of the frequency it can be turned off much and much slower. At first I had doubts on the G-S capacitances, but BSS138 has ~25pF and BSS84 has ~45pF that is not as much. I tried to vary with that R1 value and also tried to pull down the N-MOS gate with resistors between 1k and 100k but no luck. What can I do about speeding up this circuit? Any ideas?

And one more question - is the output voltage (after the zener) always high because of the missing load - how can it be like that?

 

[FvM]

You want to flip the PMOS transistor (source to left side).

 

[uglyduck]

You want to flip the PMOS transistor (source to left side).

Flipping the PMOS not solved the problem (that was a mistake in the simulation, but in the real circuit it's right - Drain is the output, Source is to the left). In simulation when flipped the result is the same. I have some success only if R1 is way too low - about 430 ohms. What is the explanation of that, what is the physical reason? Are there other solutions?

 

[Audioguru]

The gate of a Mosfet has a very high capacitance to the source and the "Miller effect" adds boosted capacitance from the drain. Therefore to switch quickly the signal to the gate must be a low resistance with a fairly high current to quickly charge and discharge these capacitances.

Frequently a complementary pair of emitter-followers is used to drive the gate.

 

[dick_freebird]

Big PMOS wants more gate drive than a 10K resistor can
source. Consider making another, smaller version of the
circuit as the main PMOS shunt, driven antiphase.

All you really need is an "on" resistance well lower than
ZL of the PA, right? You could get that for about 50 cents
from any power MOSFET driver, bare output, and hardly
any RF circuitry operates at drain voltages greater than
you can find a power MOSFET driver or purpose-made
drain switch / load switch IC for.


A proper level shifting power MOSFET driver for PMOS
high side, might not cost you much more than your low
side NMOSFET. And you often get two per package, so
you could (say) have independent Tx / Rx enable.