0
\$\begingroup\$

I'm trying to make a buck converter that should carry a load of up to 12A. Probably no surprise, the n-channel mosfet is getting hot quick, even with a heatsink it is getting to hot. (I have currently changed the Q5 to a IRFP1405PBF)

I have then tried to parallel two mosfets (with wires). But, It seems like only one of the fets gets all the load, and gets hot. How can that be? And are there other things that I could do to get a stable buck converter?

Shematics

\$\endgroup\$
  • \$\begingroup\$ Your schematic appears to have been cropped on the right and bottom sides. We could do with seeing where those connections go. \$\endgroup\$ – Finbarr May 3 '18 at 14:06
  • 1
    \$\begingroup\$ You might need to turn the fet on and off faster. Are you charging the gate via a 1 k resistor? What do the waveforms look like? \$\endgroup\$ – George Herold May 3 '18 at 14:11
  • 1
    \$\begingroup\$ What switching frequency? What is the value of L1? Do you have closed-loop control? How stiff is your 24V input source? \$\endgroup\$ – Adam Lawrence May 3 '18 at 14:39
  • \$\begingroup\$ Connect your oscilloscope to gate-source if Q5 and it will become apparent why you have so high switch losses. \$\endgroup\$ – winny May 3 '18 at 14:42
  • \$\begingroup\$ @Finbarr The cropped out lines in right side is just going to a relay. (to switch direction of current). The bottom line is going to a similar setup (a mirror of the top circuit). \$\endgroup\$ – Rune79 May 3 '18 at 15:55
2
\$\begingroup\$

Add this to your gate drive:

schematic

simulate this circuit – Schematic created using CircuitLab

You need to charge and discharge the gate fast to minimize switching losses. You can also buy a real gate driver that can source or sink amperes of current quickly.

Also, the IRF530 is a bit wimpy for this application. Try getting one with about 1/10 the Rds(on). Remember Rds(on) increases significantly with temperature. For that reason you generally don't want to thermally couple paralleled MOSFETs, but you're better off just getting a better, more modern, part.

\$\endgroup\$
  • \$\begingroup\$ Thank you very much for that schematic, Sperhro Pefhany. It seems like a very likely solution to my problem. I'll start out by trying the Totem pole. Then move on with the integration of a gate driver. Would there be any mosfet that you would recommend for this? I feed the gate with 10v now. \$\endgroup\$ – Rune79 May 3 '18 at 18:20
  • \$\begingroup\$ The one you mention you are using now IRFP1405PBF looks pretty good. \$\endgroup\$ – Spehro Pefhany May 3 '18 at 19:05
1
\$\begingroup\$

Your MOSFETs should probably be thermally coupled in this instance. Any temperature difference can make a positive feedback loop, because the hottest FET tends to have the highest RDS(on). So put both MOSFETs close on the same heat sink.

That being said, you should also consider finding a high current gate driver for your MOSFETs, in order to minimize overall switching losses. Power MOSFETs have significant capacitance at the gate.

If you want to switch them fast enough through BJTs and resistors of your own design, you need to spend some time on designing and testing that. (The totem pole driver in Spehro Pefhany's answer would make a good starting point for that.)

An off-the-shelf driver is by far the easiest solution IMO, and quite possibly better than anything you can easily make from discrete components. For example, the TC4420 has 6A peak current, dual output channels (so one driver chip can drive both your gates independently), and can be bought for less than half a dollar. There are of course many others to choose from, so I suggest you find one that fulfills your particular needs.

\$\endgroup\$
  • \$\begingroup\$ Thanks Dampmaskin, I'll order the TC4420 and make some space in my PCB. But, I'll also give the "oldschool" Totem pole solution a try. Logic-level mosfet, directly feed from the MCU is out of the question in this case, I guess? \$\endgroup\$ – Rune79 May 3 '18 at 18:29
  • \$\begingroup\$ You have to mind the current as well as the voltage. MCUs can typically only provide a few dozen mA of current. A power MOSFET needs much, much more if you want to drive it fast (PWM) and efficiently. \$\endgroup\$ – Dampmaskin May 3 '18 at 19:22

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.