Good day,

I have duplicated this circuit using a 40106 hex inverter to drive a motor using pwm. The only difference in my design was to switch out the transistor stage for an Irfz34n MOSFET. The motor is a 12v 2amp drill motor. When I tested it, I received decent pwm, adjusting the motor speed a little, however the MOSFET almost burned up immediately. I have bought a heatsink for test #2, however I'm concerned I am missing a crucial variable regarding thermal conductivity. Although MOSFET gate impedence is high, should I have a small resistor to check current along with a pulldown resistor? The invertor going low seems to me would take care of floating charge but I am still learning. I'm also concerned a resistor to the gate would slow the Gate charge time and mess with the pulsed switching speed. Any thoughts would be appreciated.

Edit: I should add, the motor stage is powered independently with a 12v lead acid battery.

enter image description here

  • \$\begingroup\$ There may be very little gate-source conductance, but the capacitance is quite high. Your 40106 likely just can't charge that parasitic capacitor fast enough. You need a FET driver. \$\endgroup\$ – Hearth May 20 '17 at 19:23
  • 1
    \$\begingroup\$ Checking datasheets and doing the calculations (correct me if I'm wrong), a ballpark figure for maximum PWM frequency with this configuration is somewhere around 1-2kHz. At 20kHz, the maximum output current of the 40106 takes fully half the PWM period to charge the capacitor, and then the other half will be consumed discharging it. At duty cycles other than 50%, the capacitor will never get fully charged or fully discharged, further messing things up. This means your transistor will always be in linear mode, and thus always dissipating a significant amount of heat, above this frequency. \$\endgroup\$ – Hearth May 20 '17 at 19:32
  • 1
    \$\begingroup\$ My recommendation of 1-2kHz is simply an order of magnitude lower than the maximum. Slower frequencies will cause less stress on the transistor. But all of this is assuming you're getting a constant optimistic output of 1.4mA from the 40106, which is not actually going to happen. (incidentally, values used for this analysis were 34nC gate charge from this datasheet, and 1.4mA maximum high-state current from this datasheet. \$\endgroup\$ – Hearth May 20 '17 at 19:36
  • 1
    \$\begingroup\$ You probably could, but you'd need a push-pull configuration; you have to discharge that gate as well as charge it! If you want an actual FET gate driver, Digikey (link leads to their gate driver catalog) is a good source for parts, if you're okay with ordering them. \$\endgroup\$ – Hearth May 20 '17 at 19:43
  • 1
    \$\begingroup\$ With a 12 volt supply the 40106 has higher drive strength, I estimate your PWM frequency to be in the 100's of Hz so you should be able to make this circuit fit your needs \$\endgroup\$ – sstobbe May 20 '17 at 20:21

If the only change you made was exchange Q1 for a mosfet (Irfz34n).

There are two issues you are facing.

One reason your switching element is burning up is slow turn on/off times. With a 22k gate resistor your fet is spending a long time transitioning from Rds_on to being turned off. Same is true for turning on. During these transitions because the FET is current limited by Vgs there is an appreciable power dissipated (Vds*Id). Approximating the gate as 1 nF, 1 RC time constant is 22 uS. Your first step is to lower R1 to say 100 ohms.

The second issue is with having separate supplies, D1 clamps the inductive kick-back across the motor, however it does not clamp kick-back from the lead inductance in your hookup wiring. One solution is to add an RC snubber at the drain terminal, such as,


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ Thank you. I will try out the rc snubber. I apparently did not make it clear that I also did not include the resistor along with replacing the transistor for the MOSFET. Currently there was no resistor when I tested. Should I have one regardless at around 100ohms or is it even necessary with outputs from the hex inverter? Seems switching speed is my issue and I will look there. Most valuable help. Thanks again. \$\endgroup\$ – Archaeus May 20 '17 at 20:05
  • \$\begingroup\$ The recommendation is for Rg is more anecdotal than theoretical with good tight layout it can be reduced, but for a bread-boarded type construction I would personally add an Rg right at the gate to minimize parasitic oscillation during switching. \$\endgroup\$ – sstobbe May 20 '17 at 21:11
  • \$\begingroup\$ Perfect. Cheers. \$\endgroup\$ – Archaeus May 20 '17 at 21:56
  • 1
    \$\begingroup\$ Update: I made the modifications and it's purring like a kitten, not even warm. The knowledge base here is admirable. Thanks again for helping a hobbyist learn. \$\endgroup\$ – Archaeus May 20 '17 at 22:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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