9
\$\begingroup\$

H-Bridge circuit

In this circuit I designed, I use a TC4427 as driver and a dual package FDS8858CZ as MOSFET. I'm feeding the driver 12 V and the MOSFETs 18 V. I am using the C9, D3, and C10 and D4 components you see in the schematic to increase the applied PWM signal from 0-12 V to 6-18 V.

When I apply the signals you see in the schematic to run the motor in one direction, the P-channel MOSFET without PWM gets broken and the system becomes short-circuited.

I control the duty cyle with a potentiometer. When the duty cycle I apply is 0% (the motor is running at the highest speed) there is no problem. When I start it at 100% (the motor stops) and try to reduce the duty cycle with the potentiometer, the P-MOSFET without PWM is broken. When I feed the drivers and MOSFETs with the same voltage, there is no problem.

I think the problem is related to the C9, C10, D3, and D4 components I mentioned earlier, or the frequency. I tried to run the system between 1-10 kHz and I used different motors, but the result did not change. Can you help?

\$\endgroup\$
4
  • \$\begingroup\$ Welcome! Show Vgs and Vds waveforms. \$\endgroup\$
    – winny
    Apr 4, 2023 at 11:58
  • 4
    \$\begingroup\$ P channel MOSFET without PWM gets broken seeing AC coupling (C9 & C10), I am not surprised. \$\endgroup\$
    – greybeard
    Apr 4, 2023 at 12:02
  • 1
    \$\begingroup\$ How far away is the +18V bypass capacitor(s)? How are you creating dead time in the drive signals? \$\endgroup\$ Apr 4, 2023 at 12:16
  • \$\begingroup\$ It may be helpful when designing and prototyping circuits like this, to add a current limiting device such as an incandescent lamp. It won't much affect a properly operating circuit with light loads, but will limit current and glow brightly when there is shoot-through or other fault. \$\endgroup\$
    – PStechPaul
    Apr 4, 2023 at 19:00

2 Answers 2

7
\$\begingroup\$

You are powering the the MOSFET bridge from +18V, but the gate driver is only producing 0V/+12V. Considering that you require +18V at the gates of the upper (P-channel) FETs to switch them off, this is a problem.

It seems that you have attempted to solve this by AC coupling the gate driver output to the gate, with capacitor C10, and clamping to +18V maximum with diode D4. This will indeed "level shift" the 0V/+12V gate driver output upwards to +6V/+18V, but there's a problem.

It takes time for this DC "restoration" to establish its final offset of +6V, due to the long time constant C10×R4 (about half a second). We can simulate this:

schematic

simulate this circuit – Schematic created using CircuitLab

enter image description here enter image description here

The above graphs show gate potential, for two different duty cycles of gate driver output. As you can see, from the moment the signal starts oscillating, it takes 200ms at least for the upper peaks of the gate waveform to reach a point near enough to the +18V necessary to switch off the MOSFET. This delay is even dependent upon duty cycle.

During this initial delay, the lower MOSFET is switching on and off as required, but the upper MOSFETs are permanently on, obviously causing massive shoot-through.

The solution is simple; power the gate drivers from +18V, and remove C10, R4 and D4.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ I didn't know about the other CSV interpolation modes like PWLREPEAT and PWS. Very good to know - thanks. \$\endgroup\$
    – PStechPaul
    Apr 4, 2023 at 18:55
  • \$\begingroup\$ @PStechPaul Glad you found something new in my answer. Let me ask you a question: have you had any trouble with the DC operating point analysis in CircuitLab, like the DC and DC sweep simulation tabs? Lately I've found them to be often completely wrong in simple circuits with negative feedback. Just want to make sure it's not just me, before I report something to them. \$\endgroup\$ Apr 4, 2023 at 19:18
  • \$\begingroup\$ I've only used time domain simulation. I've had problems with some components like voltage controlled PWM. So I usually use LTspice. \$\endgroup\$
    – PStechPaul
    Apr 4, 2023 at 20:24
6
\$\begingroup\$

You are on the right track concerning which parts. When you switch on each PWM cycle, the 18-volt line will have noise spikes or ringing. The gate threshold on the FET U5B can be as little as minus one volt, so when the voltage spike goes low enough, C10 is discharged quickly through D4 to a lower value, but stays low because it must be re-charged to 18 volts through R4 (10K). So the combination of these three parts acts like a negative peak detector, and any perturbation on either the 12- or 18-volt supply will cause U5B to turn on and shoot through.

You might try switch to IX4427 for U1 and U2 - it is pin compatible and has a higher voltage rating (35 volts). Then you could drive the entire circuit directly from 18V without need for the capacitor network.

\$\endgroup\$
1
  • 3
    \$\begingroup\$ Or better still -- ditch the PMOS and improve performance and cost by going all NMOS, using bootstrap drivers e.g. IR2110 or better. These have the minor limitation that they don't work at 100% duty cycle (high side), but likely 95% would be fine here. (100%-capable parts are also available e.g. HIP4081, or by adding an external charge pump circuit or DC-DC converter.) \$\endgroup\$ Apr 4, 2023 at 14:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.