Mosfet PWM Switch Problem

[VanillaIceCream]

I'm playing around with Mosfet switches in LTspice and trying to get one working which should switch relatively high frequency PWM signals from 1- 150 Khz.
My supply voltage is (-)80 V, the Mosfet I'm using is a P-channel enhancement type IRFP9240, RGate is 10 kOhms, the pull down resistor is 10 kOhms and RLoad is 100 kOhms.

The switch works fine for a 1 Khz 50% duty cycle PWM signal and the waveform/amplitude at the output is also ok, but at higher frequencies like 50 Khz 50% (PWM) the signal at the Output gets completely distorted and the amplitude goes from
 (-)3 V - (-)17 V, at 80 Khz 50% duty cycle (PWM) the same thing happens but even worse.
 
I read the Mosfets datasheet and the mosfet should be able to handle these frequencies without problems I think.
The Mosfets VTh is (-)2 V,  (-)10 V are required to fully turn the Mosfet on, the maximum VGS is (-)20 V so I'm using (-)12 V at the gate.

I've tried everything with different values without any success, I just cant get it working at higher frequencies.

The pictures show the output at 50Khz 50% duty cycle PWM and the input PWM signal.
I don't know what I'm missing or doing wrong.
Help would be appreciated.




50Khz 50% duty cycle pwm pulse source :   PULSE(0 -12 0 0 0 0.01m 0.02m)
80Khz 50% duty cycle pwm pulse source :   PULSE(0 -10 0 0 0 0.00625m 0.0125m)

[Zero999]

Your gate resistor is far too higher value. Try 100R.

[fourtytwo42]

A few other problems
Your load resistor (100K) is so high it will interact seriously with the Drain capacitance, try 1K.
This is a P channel device so the gate voltage should be negative wrt source as should your drain supply.

[Zero999]

A few other problems
Your load resistor (100K) is so high it will interact seriously with the Drain capacitance, try 1K.
This is a P channel device so the gate voltage should be negative wrt source as should your drain supply.

Good point about the drain resistance being too high. I missed that.

The gate and drain voltages are negative, with respect to the source. The orignal poster has confusingly drawn all the voltage sources, as positive and put a negative sign in the properties.

[Cerebus]

Nobody has explained to the poor chap/chapess why it's going tits up.

That FET has a gate capacitance of 1200pF, so with the gate resistor forms a low pass filter (that's 6dB down to start with because of that unnecessary resistor to ground), and is 9dB down at 28 kHz (the 3dB point of the low pass part). That square wave has a lot of high frequency information in it and you're throwing it all away by putting 6dB of DC attenuation and then -6dB/octave AC attenuation on top of it starting to roll off well before you get to your fundamental at 50kHz which is suffering over -12dB of attenuation (a factor of four) and it just gets worse as the frequency goes up.

[IDEngineer]

It also doesn't help that the pulldown is positioned to create a voltage divider. The gate only ever sees 50% of the applied signal, no matter what else happens, further slowing the charge rate of the gate capacitance.

[jenniferkim]

Which software you have used to design this transistor simulation & graph?

[ledtester]

Which software you have used to design this MOSFET simulation & graph?

It looks like LTspice - freely available for Mac and Windows.

[VanillaIceCream]

A few other problems
Your load resistor (100K) is so high it will interact seriously with the Drain capacitance, try 1K.
This is a P channel device so the gate voltage should be negative wrt source as should your drain supply.

Thanks, it works this way but what do I have to do to be able to run the switch with a Lower load (+100kohm) ?

[Zero999]

If you need to switch such a high impedance load, use a good old BJT. Another option is a P-channel J-FET, but you'll need to drive the gate with a positive voltage the switch it off.

[Cerebus]

Thanks, it works this way but what do I have to do to be able to run the switch with a higher load (+100kohm) ?

Hold on, what exactly do you mean here? Most people when talking about a higher load mean a lower resistance/impedance for the load.

[Terry Bites]

As mentioned before, you have insuffcient base drive current. As you go up in frequency you have to charge and discharge the gate capacitance more rapidy. So you need more current. Rg is way too high. With Rg and R2 at 10k you've only got 6V of gate drive making things still worse and you just get no gate drive see To drive 1200pF of gate capcitance at this speed, you need a very low source impedance. CV=iT (near enough) so (12*1200e-12)/10e-9 =1.44A  peak drive current. An active driver circuit is required. See https://www.eevblog.com/forum/projects/driving-a-p-channel-mosfet/

for a detailed calculation see https://www.taiwansemi.com/assets/uploads/productcategoryfile/AN-1001_A1611.pdf

[VanillaIceCream]

Thanks, it works this way but what do I have to do to be able to run the switch with a higher load (+100kohm) ?

Hold on, what exactly do you mean here? Most people when talking about a higher load mean a lower resistance/impedance for the load.

Yes my bad, I'm asking about how to run the switching circuit with a lower/smaller load (+100kohm).

[Zero999]

Thanks, it works this way but what do I have to do to be able to run the switch with a higher load (+100kohm) ?

Hold on, what exactly do you mean here? Most people when talking about a higher load mean a lower resistance/impedance for the load.

Yes my bad, I'm asking about how to run the switching circuit with a lower/smaller load (+100kohm).

Replace the MOSFET in the original circuit, with a >80V BJT, such as the 2SA1579.

[Terry Bites]

Yeah, thats a monster MOSFET for less than 1mA of load current.
A small signal device will do.

[Zero999]

Yeah, thats a monster MOSFET for less than 1mA of load current.
A small signal device will do.

The problem is MOSFETs tend to be power devices, especially those rated to above 80V. This is why I suggested a BJT. I mentioned P-channel J-FETs above, but I've just realised you won't get one rated to above 80V.

[Cerebus]

Yeah, thats a monster MOSFET for less than 1mA of load current.
A small signal device will do.

The problem is MOSFETs tend to be power devices, especially those rated to above 80V. This is why I suggested a BJT. I mentioned P-channel J-FETs above, but I've just realised you won't get one rated to above 80V.

You're not wrong, but that statement needs qualifying to avoid making it misleading to the uninitiated. One needs to be careful about making generalisations, unqualified, in the beginners topic because beginners tend to soak up these casually made unqualified statements from more experienced people and treat them as lore.

The vast majority of MOSFETs in existence are very low voltage and low power devices, sitting inside all those microprocessors out there. For every power MOSFET in existence there are probably a million or more teeny tiny very low power integrated MOSFETs. Yes, the majority of discrete MOSFETs around nowadays are relatively high power devices but there are a fair few small discrete signal MOSFETs around, RF circuits abound with them, and you'll find a fair number of small discrete MOSFETs doing things like rail switching in lots of devices.

[VanillaIceCream]

Thanks, it works this way but what do I have to do to be able to run the switch with a higher load (+100kohm) ?

Hold on, what exactly do you mean here? Most people when talking about a higher load mean a lower resistance/impedance for the load.

Yes my bad, I'm asking about how to run the switching circuit with a lower/smaller load (+100kohm).

Replace the MOSFET in the original circuit, with a >80V BJT, such as the 2SA1579.

Thanks, but I really wanna do it with Mosfets even if it sounds stupid. And I also wanna stick to my IRFP9240 if possible. Can you please explain to me how I can get it working again with a much lower/smaller load and without changing the transistor?

[Cerebus]

Your IRFP9240 has a C_oss (output capacitance) of 370pF, because it's comparatively physically large. So the time constant of your 100k load with that capacitance is going to be 37 microseconds A typical small signal MOSFET, a 2N7002 has a C_oss of 25pF, nearly 15 times smaller. If you use a large MOSFET at low current then its output capacitance is going to take time to charge and consequently it's gonna be slow.

Speed tends to go hand in hand with expending lots of current to overcome junction and stray capacitances no matter what flavour of semiconductor you're using.

[Terry Bites]

Off state leakage with this Vgs=0 is 100uA or 10 volts across your 100k, do you need that? Its beyond your control. Start with the spec. not the components.  You need a rethink and a give candid desription of why you need this inappropriate curcuit to work. 8uA at 80Khz.  FYI a small signal mosfet is not a jfet. Also note you can stack Jfets to get the voltage required and get a reduction in switch capcitance at the same time- cascode style.