0
\$\begingroup\$

I am working on inverter, MOSFETs in h-bridge inverter goes heat up without any load attached. I also attached pull-up and pull-down resistor on it. here is my circuit duagram. enter image description here

EDIT enter image description here

Sources of PMOS are connected to supply and sources of NMOS are connected to ground.

Output waveform and voltage are all good but have heating issue is not resolve. Any suggestion or modification is very helpful for me.

\$\endgroup\$

2 Answers 2

9
\$\begingroup\$

Your P channel MOSFETs are connected upside down: -

enter image description here

Notice the little diode symbols inside the MOSFETs? They will permanently conduct current and the MOSFETs will not switch off.

\$\endgroup\$
7
  • \$\begingroup\$ Sorry Andy aka I was upload the circuit image with some mistake, but in my actual hardware the circuit is correctly connected so that's why I am uploading the correct circuit diagram. But the problem is another. \$\endgroup\$ Oct 1, 2018 at 18:38
  • 1
    \$\begingroup\$ Edit your question at the end and add the modified (corrected) circuit diagram. Don’t rewrite the question in case it makes my answer look invalid. \$\endgroup\$
    – Andy aka
    Oct 1, 2018 at 18:42
  • \$\begingroup\$ Ok Sir, I have 220 Ohm resistor on optocoupler input. For pull up and pull down 10k Ohm resistor is used. I think but not sure P-channels are getting hot in my circuit because they have lower trans-conductance than the N-channels, and are seeing the majority of the voltage drop. But what can I do ? \$\endgroup\$ Oct 1, 2018 at 19:22
  • \$\begingroup\$ Only PMOS are heat up, NMOS are cool down. \$\endgroup\$ Oct 1, 2018 at 19:31
  • \$\begingroup\$ @AbdulBasit the IRFZ44 has a significantly lower on resistance than the IRF9540 and if there is a shoot-through overlap then this will be the device to dissipate 3 or 4 times more heat than the N channel MOSFET. Shoot-through is caused when one MOSFET turns off and the other one turns on and there is a slight overlap in time when both are on. This takes a pulse of current from the power supply and dissipates heat as \$I^2R\$ loss. Driving with optos is not ideal because they will be slow to turn off the N channel device and not all that quick with the PMOS. \$\endgroup\$
    – Andy aka
    Oct 2, 2018 at 6:54
4
\$\begingroup\$

Can you say "shoot-through"? In addition to the problem that Andy pointed out, your MOSFETs are going to turn off much more slowly than they turn on. Have you allowed enough dead time between pulses to allow for this?

Also, are you really driving the optoisolators through 220k resistors? That's way too much resistance, and this will slow down the turn-on of the MOSFETs as well.

\$\endgroup\$
8
  • \$\begingroup\$ Sorry Dave Tweed I was upload the circuit image with some mistake, but in my actual hardware the circuit is correctly connected so that's why I am uploading the correct circuit diagram. But the problem is another. \$\endgroup\$ Oct 1, 2018 at 18:41
  • \$\begingroup\$ OK, that addresses two of the three issues we raised. What about the 10k turn-off resistors? \$\endgroup\$
    – Dave Tweed
    Oct 1, 2018 at 19:19
  • \$\begingroup\$ I think issue is in pull up and pull down resistance value. here is a link almost same issue and same PMOS and NMOS used but have no idea about the changes of the circuit or/and value of pull up and pull down value. electronics.stackexchange.com/questions/270892/… \$\endgroup\$ Oct 1, 2018 at 19:26
  • \$\begingroup\$ If you want fast switching of the MOSFETs, you really want "totem-pole" (2-transistor) drivers on each one. For example the IRFZ44N has a total gate charge of 63 nC. If you want to switch that in, say 50 ns, you need a peak current of \$\frac{63 nC}{50 ns} = 1.26 A\$ (!). Your resistors only give you \$\frac{12 V}{10 k\Omega} = 1.2 mA\$, about 3 orders of magnitude slower. \$\endgroup\$
    – Dave Tweed
    Oct 1, 2018 at 19:29
  • \$\begingroup\$ Not more fast switching just 50Hz. \$\endgroup\$ Oct 1, 2018 at 19:30

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.