H-bridge design- I keep shorting out my power mosfets and I am unsure of how else I could drive them so that I stop shorting them out.

Thread Starter

rfengineer28

Joined Apr 28, 2021
79
pcb.png

I supply this circuit 120V and I am having a hard time understanding why I am burning my transistors when I try to drive a motor.
 
Last edited:

dendad

Joined Feb 20, 2016
4,452
The proper drivers incorporate a "dead band" that has a defined off time for all FETs to prevent just what your problem may be.
Just for a test, add a mains incandescent lamp in line with the incoming HVDC to limit the current pulse.
And then you may see the lamp flash as the direction switches. Try with no load.
A lamp is sometimes a really handy debug tool!
 

Papabravo

Joined Feb 24, 2006
21,159
Without taking a deep dive into the circuitry, my first guess is that you have not made sufficient provision for "dead-time" between switching the high and low sides of the bridge. Even a very small overlap can allow huge currents to flow with disastrous results. If you are certain that this base is covered then we must look deeper. The next potential culprit would be the switching speed of the FETs being affected by an unexpectedly long "Miller Plateau". If you are certain that this base is covered, then it is time for a deep dive into the circuit, but it's late.
 

crutschow

Joined Mar 14, 2008
34,285
Why are you using slow analog op amps to drive the digital signals to the MOSFET gates?
You should be using a much faster device as drivers.
 

Thread Starter

rfengineer28

Joined Apr 28, 2021
79
Why are you using slow analog op amps to drive the digital signals to the MOSFET gates?
You should be using a much faster device as drivers.
the reason I used the lm318 is because I set the frequency of my signals to only 4kHZ, therefore I though those op amps would be fast enough. what's a faster device you suggest I use for this application? I really think that the issue is that the corresponding mosfets are not getting turned on/off fast enough.
 

DickCappels

Joined Aug 21, 2008
10,152
As stated earlier, use proper MOSFET gate drivers. Most likely the switching time for your P-channel MOSFETs is too slow, which if not the cause of the dead transistors, certainly does not help the efficiency. See MaxHeadRoom's comment in post #4.

You have two, now make that three people concerned about dead time. The slow speed of your P-channel device may be part of the reason. Do you even have dead time in the drive waveforms?
1650093313377.png
 

Thread Starter

rfengineer28

Joined Apr 28, 2021
79
As stated earlier, use proper MOSFET gate drivers. Most likely the switching time for your P-channel MOSFETs is too slow, which if not the cause of the dead transistors, certainly does not help the efficiency. See MaxHeadRoom's comment in post #4.

You have two, now make that three people concerned about dead time. The slow speed of your P-channel device may be part of the reason. Do you even have dead time in the drive waveforms?
View attachment 265232
Thank you for your reply. If you don’t mind me asking, how could I possibly address the speed concern? Would I need dedicated high side and low side drivers rather than the op amps?
 

MrSalts

Joined Apr 2, 2020
2,767
The Gate of your P-channel Mosfets charge/discharge through 11k resistors. The Gate is like a capacitor that takes time to charge and discharge let's look at how much time that is in your circuit. The datasheet says the Gate charge is 200nC at 10v = 30nF. This will take just over one time constant to drop the -10v Gate voltage below threshold of -3v. Let's simplify and call it one time constant (a time constant is a 63% change from initial voltage). The time constant of your RC discharge circuit is 11k ohms x 20nF = 220 microseconds. Therefore, it will take more than 220 microseconds to discharge the MOSFEt Gate below the threshold voltages (point where it no longer conducts).

I set the frequency of my signals to only 4kHZ
"Only 4kHz" should be rephrased for your circuit as, "the very high frequency of 4kHz".

Here's why,
If it takes 220 microseconds to discharge your Mosfet gate, it is not even full discharged by the time the next ON signal comes. 4000 Hz means only 250 microseconds and your current setup would require a dead-band of 220 microseconds to discharge your P-channel Mosfet.


A guess at your problems
***!,!,!
Option 0) **** this is certain ***
IF YOU ARE DRIVING YOUR MOTORS WITH 120v, your p-channel Mosfet Gates have a 120/2 = 60v on them (look at your two 11k voltage dividers). The datasheet says max (transient of -30v and typical max -20v).

these are other possible design/control problems...
Option 1) You are controlling both P-channel and N-channel at 4kHz. The P-channel is not turning off fast enough snd you are getting "shoot-through" voltage down one side of your H-bridge (that is, the current is not going though your motor. It is just shorting to ground).

Option 1 solution) keep the P-channel on and modulate your PWM only though your N-channel when going in one direction. Turn off all Mosfets for more than 400 microseconds before changing directions.

Option 2) you are driving your gates well below the desired 10v and the Mosfets are getting hot. Or, you are switching the Mosfets way too slowly and the Mosfet is in between full off snd full on for too long and getting hot.
Option 2 solution) use a Mosfet driver. You can get multichannel drivers to replace the op amps and the extra n-channel Mosfet that you use to drive the p-channel Mosfet
 

DickCappels

Joined Aug 21, 2008
10,152
Thank you for your reply. If you don’t mind me asking, how could I possibly address the speed concern? Would I need dedicated high side and low side drivers rather than the op amps?
Thank you for asking. To turn a MOSFET on or off it is necessary to put a charge into to take a charge out of the MOSFET's gate. This often done with driver integrated circuits because to switch quickly you need amps of drive. The LM318 is a nice and pretty fast op amp, but if falls way short of being able to drive any but the smallest MOSFETs.
 

Papabravo

Joined Feb 24, 2006
21,159
You will need the oscilloscope for two distinct reasons:
  1. You need to VERIFY the actual amount of dead-time between switching signals that could potentially short your HVDC supply to ground.
  2. You need to verify the amount of time your rising and falling gate drive waveforms are spending on the "Miller" plateau where the Rds(on) of the MOSFET is not at it's minimum value cause high power dissipation for as long as the gate drive waveform remains on the plateau.
The following article may be of interest to you:
https://itectec.com/electrical/elec...e-miller-plateau-of-mosfets-dependent-on-vds/
 

crutschow

Joined Mar 14, 2008
34,285
the reason I used the lm318 is because I set the frequency of my signals to only 4kHZ, therefore I though those op amps would be fast enough. what's a faster device you suggest I use for this application?
Using an op amp as a comparator is sort of like using a screwdriver as a hammer.
It may work but you don't always get the desired results. :rolleyes:

An LM339 comparator would be much faster, although you likely need more than that to get the desired switching sequence from your circuit, such as a generated non-overlap signal (do you understand what that is?).
 
Top