0
\$\begingroup\$

I have a HIP4082 full bridge MOSFET driver driver driving a full bridge. After some experimenting with the circuit prototype i found out that the driver heats up to over 60 °C after a short while of running, which concerned me but it worked fine. However as I decreased impedance across the load (which was originally connected to primary coil of a transformer) the driver started acting in a weird way and i found out that it has blown out. This is already the second driver i destroyed this way and they're expensive as hell, so I need a solution.

I think what's causing the driver to blow out is that when I decrease impedance across the load I basically create a short circuit between the driver's bootstrap pin and ground, which kills it. By adding a resistor across the load or the whole bridge and ground, I could easily solve the issue, however I do need low impedance on the load because I need high current (up to 20A).

I thought about adding a resistor across the driver's bootstrap line, but I have concerns about it affecting the bootstrapping functionality.

EDIT: I'm actually using IGBTs in place of MOSFETs (specifically IRGPS4067DPBF) Also I'm not posting a layout because the full bridge is not on a PCB but it's simply bridge-soldered to the driver circuit. The full bridge operates at 150 kHz square wave.Both circuit and load voltage is 12v

Also here's my circuit schematic: enter image description here The full bridge is connected as in the driver's datasheet, except the feedback loop and shunt resistor: enter image description here

Here's the control circuit layout: enter image description here

And here's the picture of the bridge: enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ Time to measure with an oscilloscope \$\endgroup\$ – PlasmaHH May 18 '17 at 13:20
  • 1
    \$\begingroup\$ IMO the 150kHz is a huge for such bridge, not a good starting point for DIY. Also note that the schematics in datasheet is just a sketch, not the real life one. \$\endgroup\$ – Marko Buršič May 18 '17 at 13:32
  • 1
    \$\begingroup\$ Start with 100mV current shunt on gnd at low frequency and measure deadtime (if poss.) then Idc vs f switch rate for different loads and wiring ESL and C or Z(f) It may be 1st, 2nd or 3rd order effect with frequency depending on root cause. \$\endgroup\$ – Sunnyskyguy EE75 May 18 '17 at 14:06
  • 1
    \$\begingroup\$ Rg * Ciss is a significant value in bridge circuits as is L/RdsOn and RdsOn/load ESR \$\endgroup\$ – Sunnyskyguy EE75 May 18 '17 at 14:11
  • 1
    \$\begingroup\$ "And here's the picture of the bridge:" - what a mess! Long wires = high inductance = voltage spikes and excessive ringing. \$\endgroup\$ – Bruce Abbott May 18 '17 at 19:28
2
\$\begingroup\$

If your load is a transformer, some inductance will be involved. The power leads also have inductance. Switching will create voltage spikes on the main power rail, which can be high enough to blow your driver.

These spikes are normally absorbed by decoupling caps on the power rail, but there is no decoupling on your schematic... and you didn't show your layout.

This is really a hunch. You should show your layout.

\$\endgroup\$
  • \$\begingroup\$ The bridge part is just bridge soldered. And there is no decoupling capacitor, I'm still a bit new to this. The thing is though, that the driver blow out when I bypassed the transformer and I suspect it will also happen if the load current goes sufficiently high. \$\endgroup\$ – DELTA12 May 18 '17 at 16:04
  • 1
    \$\begingroup\$ Show us a picture, please. There was a recent case (maybe 2-3 weeks ago) of a guy blowing his bridge driver... so I told him to put decoupling caps on the power rail and that solved it. Fast switching of a high current combined with supply wire inductance to create a nasty L*di/dt spike which blew the chips... It wasn't load inductance in this case, simply the power supply wires. The nastiness (di/dt) is proportional to the current, so I'm not surprised it works at low current then blows when current gets high. \$\endgroup\$ – peufeu May 18 '17 at 16:38
  • 1
    \$\begingroup\$ OK... the pictures are explicit. This cannot work at 150kHz. Wiring has way too much inductance. I'm even surprised the MOSFETs don't explode due to oscillation induced by gate inductance. \$\endgroup\$ – peufeu May 18 '17 at 22:24
  • 1
    \$\begingroup\$ You need a tight layout, on a PCB, with short connections, low parasitic inductance, good decoupling, ground and power planes, etc. High switching frequency and low switching losses needs fast switching, which means high di/dt, which is not compatible with high inductance (ie, L*di/dt). Inductance in the gate/base lead of a power device slow down switching and makes oscillations likely. You cant use flying wires as shown in the picture. Even 0.1µH inductance is wayyyy too much for this to work. \$\endgroup\$ – peufeu May 19 '17 at 17:17
  • 1
    \$\begingroup\$ You can get a 4 layer board for 50$ these days. Your design most likely does not need 4 layer though, a good double sided layout should do the trick. Think of inductance as surface area in a current loop, this is what it is. Your wires make loops. Check loop surface area. Compare a wire loop like a rectangle of 100mm x 20mm = 2000 mm2 area with a 1.6mm thick PCB with short traces, 10-20mm long. Imagine the traces on both layers (1.6mm apart) as two wires, now you can see the surface area is very small. 1.6mm x 20mm = 32mm2, which is 60x lower inductance than the wires. \$\endgroup\$ – peufeu May 19 '17 at 17:28
2
\$\begingroup\$

Your layout may be sloppy. Blowing the driver out at high load current (as opposed to shoot through, which is independent of current) usually indicates that you have stray inductance which is causing excursions at the driver output that are unacceptable. Try adding some reasonable series gate resistance (15 or 20 ohms) and clean up your layout to minimize loop areas that carry load current. Make sure you have bypass capacitors on the 80V and 12V buses.

Generally 150kHz is quite a high switching frequency for IGBTs, also IGBTs are not a very good choice at low voltage- they have a lot of voltage drop. MOSFETs are typically a superior solution when voltages are low- their big advantage is that they are available and are inexpensive (small die size) with very high voltage ratings (eg. 1200V) and the voltage drop does not get worse.

\$\endgroup\$
  • \$\begingroup\$ I'm pretty sure that what causes the driver to blow out is overcurrent at bootstrap supply as I explained in the original post. \$\endgroup\$ – DELTA12 May 18 '17 at 16:46
  • \$\begingroup\$ I might be wrong though \$\endgroup\$ – DELTA12 May 18 '17 at 16:58

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.