Mosfet channel P: Speed enhancement circuit

Thread Starter

Deivi

Joined Nov 10, 2016
5
Hi everybody,

I am working in a proyect and i'm having difficulties with a part of it.

Long story short, i have to make a pulse generator of 0 to -100V with a variable duty cicle
The idea is to control a mosfet (like a switch) with a microcontroller to generate the pulse.

My issue is that i need a rise time of 25ns for the pulse, so i need the mosfet to switch very fast. I use a totem pole configuration to "speed" my mosfet (in other words, to reduce the turn-off time )

This is the circuit that i am using:

upload_2016-11-17_21-22-15.png


I am simulating in LTspice and the problem is that i dont have a reduction of the turn-off time with the totem pole and i cant see where the problem is.

Thanks in advance
 

#12

Joined Nov 30, 2010
18,224
There seems to be come confusion about polarities. +9v on the collector of a PNP?
No resistor to bring the base of Q1 to ground?
After checking 5 gate drivers, most of them can't do a 25 us shut-off.
How fast do you really need to go?
 

Thread Starter

Deivi

Joined Nov 10, 2016
5
There is -9V in the collector of the PNP
Yes, i need 25ns, the transition time of rise and fall (10% to 90% of the signal) can not exceed 25 ns.
That is why i'm using the totempole, to reduce time as much as possible
 

#12

Joined Nov 30, 2010
18,224
I'm not the best guy here on this sort of thing, and I can't simulate at all, but the impression I get is: lower the impedance of the totem pole driver to 1K or less and DC couple it. Add 1k (or less) to ground at the gate of your M1. The 2N2222 can move at 4 nanoseconds, so that's good. The 2N2907 seems a bit slow at 30 ns to 80 ns...if I'm reading that correctly.



 

Attachments

crutschow

Joined Mar 14, 2008
34,285
Okay, here's my take on the circuit with some modifications.
Since a microcontroller doesn't have a lot of output drive I used six 74HC04 high speed CMOS inverter gates (one hex inverter chip) in parallel to provide some drive boost. In practice you may find fewer than six works better but that worked best in the simulation.
I also used high speed saturated switch type BJTs in the gate drive circuit.
(You can generally identify high speed BJTs by their relatively low Vce rating.)

The LTspice simulation below of this shows an output rise and fall time of about 25ns.

Note that the MOSFET I used is only rated for 100V which is the best I had in my available models.
You need one rated for a least a 120V. It should also be a logic-level device with a low gate capacitance.

Achieving these rise/fall times in practice will be tricky. The circuit will need to be on a copper ground plane with short leads and good decoupling (0.1μF ceramic caps directly to the ground plane at critical points such as the 74HC04 power pin, Q2's collector, and the top of R1). A vector board with a ground plane should work if the layout is carefully done.

upload_2016-11-17_22-51-5.png
 

Thread Starter

Deivi

Joined Nov 10, 2016
5
Thank you all for the contribution I will be testing the proposed solutions to see how they work

Achieving these rise/fall times in practice will be tricky. The circuit will need to be on a copper ground plane with short leads and good decoupling (0.1μF ceramic caps directly to the ground plane at critical points such as the 74HC04 power pin, Q2's collector, and the top of R1). A vector board with a ground plane should work if the layout is carefully done.
I think these issues will be less problematic if i design this part with SMD
 

Thread Starter

Deivi

Joined Nov 10, 2016
5
Hi guys, its me again
I had abandoned this project but now I'm taking it back....And the old problems reappeared.

The things that you recommended to me were very useful, so I decided to go back here
Here is the new schematic:


upload_2017-8-18_0-34-9.png

The circuit is working fine but im stil having a problem with the rise time. The turn-on time is 13ns but the turn-off time is 52ns
Here is an image of the simulation:

upload_2017-8-18_0-45-39.png


Do you know how to solve this problem?
Any idea is welcome, thanks!!
 

ebeowulf17

Joined Aug 12, 2014
3,307
Hi guys, its me again
I had abandoned this project but now I'm taking it back....And the old problems reappeared.

The things that you recommended to me were very useful, so I decided to go back here
Here is the new schematic:


View attachment 133164

The circuit is working fine but im stil having a problem with the rise time. The turn-on time is 13ns but the turn-off time is 52ns
Here is an image of the simulation:

View attachment 133166


Do you know how to solve this problem?
Any idea is welcome, thanks!!
What is C2 for?
 

Alec_t

Joined Sep 17, 2013
14,280
A couple of points.
The IRF740 needs 10V Vgs to turn on fully.
Q3 and Q4 (200mA rated) are restricting the charge/discharge current of M1 gate, hence slowing the switching.
 

Thread Starter

Deivi

Joined Nov 10, 2016
5
What is C2 for?
Is a decoupling capacitor. At the out of the driver i have a square signal of 0 to 5V and i need that the Vgs voltage go from 2 to 7 (Vgs threshold is 2v minimun and 4V maximum), so putting C2, R8, and R6 make that signal in the gate of mosfet be a square -18V to -13V and the voltage in the source is -20V, so Vgs is 2 to 7


Q3 and Q4 (200mA rated) are restricting the charge/discharge current of M1 gate, hence slowing the switchin
I already tried with more current tbj (I tried 800mA) and even with Mosfet and I cant reach the 25 ns.
 

ebeowulf17

Joined Aug 12, 2014
3,307
Is a decoupling capacitor. At the out of the driver i have a square signal of 0 to 5V and i need that the Vgs voltage go from 2 to 7 (Vgs threshold is 2v minimun and 4V maximum), so putting C2, R8, and R6 make that signal in the gate of mosfet be a square -18V to -13V and the voltage in the source is -20V, so Vgs is 2 to 7




I already tried with more current tbj (I tried 800mA) and even with Mosfet and I cant reach the 25 ns.
I see. I didn't realize that 2V offset was deliberate. Now I get it. Thanks for the explanation.
 
Top