4

I'm working on designing a half H-bridge circuit around an IR2184 Mosfet driver chip. I've had limited success with it. Ideally it will allow me to drive a 12V/4A bilge fan when I get it working.

One iteration had the fan wanting to kick over but never full starting. Something I've attributed to bootstrapping on my HS driver. The larger issue as I've iterated over it is that I keep killing the driver chip completely with only an oscilloscope on the output.

To drive it, I am using an Arduino Micro 5V, configured to output a PWM signal on pin 10 running at 32kHz. For bootstrapping I am using a US1M Diode (Basically a SMD version of the UF4007). I have 10Ω gate resistors with US1M bypass diodes in parallel. I also have 15V Zenners with 10kΩ Gate-Source resistors on both MOSFETS.

I'm using PCB's instead of breadboards as the currents I am using are more than a breadboard can handle. Plus they are cheap(ish) enough to waste several. Either way I have destroyed enough of these chips to seek some professional advice. Without any significant load, what can be causing my driver to fail?

Schematic: Half Bridge Schematic

PCB: Half Bridge PCB

JYelton
  • 34,119
  • 33
  • 145
  • 265
Laveur
  • 71
  • 6
  • The faliure can be due to zener diodes used, remove them. You may remove also 10k resistors from gates. Try with lower PWM frequency. – Marko Buršič Jul 19 '19 at 15:20
  • You don't need gate pulldowns since the gate driver IC should have been designed to handle all that. I think 10 ohms for MOSFETs with that gate charge is too slow for 32kHz. Maybe consider ferrite beads instead. – DKNguyen Jul 19 '19 at 15:29
  • 1
    This is just a pump right? Why do you even need all this? What's wrong with just a low side MOSFET and a low side gate driver? – DKNguyen Jul 19 '19 at 15:42
  • The zener diodes limit the voltage between gate to source to 15V reducing an spikes that may be caused due to the inductive nature of the output. The 10k resistors also help to pull the gate voltage completely off. The Gate drive won't guarantee that for me. I could due this with just a low side driver but that would be inefficient. An H-Bridge helps smooth things out. – Laveur Jul 19 '19 at 15:49
  • First, your gate driver already handles the pull-down issue. Look at the block diagram of pg 4 on the datasheet. Second, you said you are using an H-bridge but your schematic and layout only show a half-bridge. Did you only show us half your circuit? Third, H-bridges don't improve efficiency or anything of the sort if you only need unidirectional rotation. A half-bridge can improve efficiency over just a low-side switch by using synchronous rectification instead of a flyback diode. You need to provide some clarifications here since it's no longer certain what you have or are trying to say. – DKNguyen Jul 19 '19 at 15:58
  • Maybe this is responsible for the confusion...but H-bridge and half-bridge are not the same. The "H" in H-bridge doesn't stand for "half". The "H" is the shape of the circuit which is an H. An H-bridge is four transistors. A half-bridge is half of that (two transistors, one HI and on LO). Everything you have shown so far is a half-bridge and it takes two of those (one on each terminal of a motor) to make an H-bridge which allows rotation in both directions. – DKNguyen Jul 19 '19 at 16:01
  • Thanks DKNyguyen, I am not confused. I am well aware. I should have been more specific that this is a half H-Bridge and not an H-Bridge. – Laveur Jul 19 '19 at 16:08
  • I would like to add that this design has worked for me. But more often then not it fails with basically no output connected. – Laveur Jul 19 '19 at 16:08
  • 2
    Sorry to say, but your layout is scrap. MOSFET gate drivers (IR in particular, ask me how I know) are notoriously sensitive to negative voltage. You must minimize the stray inductance between U1, Q1 and Q2. Start by placing them as close to each other as possible and then fill in with the passives around it. What’s your deadband time? – winny Jul 19 '19 at 16:27
  • why the thermal relief of the heatsink pads ? – Jasen Слава Україні Jun 05 '21 at 23:08

1 Answers1

3

So after many hours of testing, and rebuilding it all on a breadboard I think I found my mysterious issue. Seems that my reflow oven is the culprit. The data sheet specifies that the leads should not exceed 300℃ for longer than 10 seconds. It doesn't appear to heat up the board that high during the reflow process. But hand soldering the chip on by hand at a much lower temp made everything work fine.

Laveur
  • 71
  • 6