3
\$\begingroup\$

As I'm a noob in this field I’m experimenting with a half bridge circuit to run a buck converter to learn about it. First I tested my circuit by putting everything together on a breadboard and it worked great. After soldering it together on a piece of perfboard I ran into a problem. When I apply a voltage to the circuit, the high side MOSFET runs very hot within seconds. I found that the high output of the gate driver doesn´t provide enough voltage to the gate, it outputs around 17V instead of 25V as it does on the breadboard (VCC is 15V). So I think something is wrong with my bootstrap circuit.

PWM is created by an ESP32 and SD is pulled to 5V.

my schematic

Things I tried so far:

  • Doubled the bootstrap cap (ceramic) to 200 nF.
  • Increased C4 from 100 nF to 10 uF (thought 100 nF might not be enough to charge the bootstrap cap).
  • Tried several PWM frequencies from 20 to 75 kHz and duty cycles from 20 to 80%.
  • Replaced the gate driver IC and MOSFETs.

Still the high side MOSFET is running hot. Again, this design runs perfectly fine on the breadboard. I checked the connections on the perfboard more than once and it´s soldered together like in the schematic. I´m happy about any help!

Update: Here the VS and VB pins as requested in the comments, unfortunately I only have a potato China oscilloscope available. But I think what can be observed is that the V_min on the VB pin on the perfboard is way higher than on the breadboard. Can anyone observe what could be wrong with the circuit? Also note that the perfboard has a 1N5817W Schottky diode installed instead of the 1N4007 now.

enter image description here

\$\endgroup\$
15
  • 1
    \$\begingroup\$ Welcome to EE.SE! How does the VS and VB pins look on your scope? Have you tried a faster diode for D1? \$\endgroup\$
    – winny
    Commented Apr 23, 2021 at 13:56
  • \$\begingroup\$ Thank you! :) I didn´t try another diode yet, but also already thought about changing it. I just have a 1N5817W laying around, do you think that might be better? Or is it too weak? I can post a picture of VS and VB later on. \$\endgroup\$ Commented Apr 23, 2021 at 14:05
  • \$\begingroup\$ Check the trr time in ns for each bootstrap diode in the datasheet. You need a very fast one which can withstand Vcc+Vb+margin. \$\endgroup\$
    – winny
    Commented Apr 23, 2021 at 14:11
  • 1
    \$\begingroup\$ I'm thinking that 22uF cap may require a lot of current to charge up before your boost kicks in substantially. Winny is right, look at VS and VB on the scope and post them. What frequency are you running at? \$\endgroup\$ Commented Apr 23, 2021 at 14:16
  • 1
    \$\begingroup\$ Whoa! That doesn’t look even remotely correct. Any LP filter activated in the potato? It may in the end be down to that you need a real PCB, two layer minimum with good ground plane and proper decoupling to make it work. \$\endgroup\$
    – winny
    Commented Apr 24, 2021 at 17:53

0

Browse other questions tagged or ask your own question.