40A H Bridge DC Motor Driver Circuit with PCB

[hesam.moshiri]

An H-Bridge (Full-Bridge) driver is quite popular in driving loads such as brushed DC motors and it is widely used in robotics and industry. The main advantages of using an H-Bridge driver are: high efficiency, rotation direction change, and braking the motor. In this article/video, I have introduced a complete H-Bridge DC motor driver using four IR3205 power MOSFETs and two IR2104 MOSFET drivers. Theoretically, the above-mentioned MOSFET can handle currents up to 80A, however, in practice we can expect to get currents up to 40A if the MOSFET temperature is kept as low as possible, using a big heatsink or even a fan.

[MasterT]

 Theoretically, the above-mentioned MOSFET can handle currents up to 80A, however, in practice we can expect to get currents up to 40A if the MOSFET temperature is kept as low as possible, using a big heatsink or even a fan.

That's right, theoretically.  See figure 8 in the data sheet:
https://www.infineon.com/dgdl/irf3205spbf.pdf?fileId=5546d462533600a4015355defac9190c
unfortunately  DC line is missing on the chart, but I'm sure it 'd roll down below 10A if voltage >10V.
Yesterday I  was testing my new toys, SPD15P10PG rated 100V & 15A. And know what? Tested unit releases magic smoke and failed short circuit drain - source in 3 sec, with only  30V & 5A. Heat think was good to keep 150W for a few minutes.

[langwadt]

 Theoretically, the above-mentioned MOSFET can handle currents up to 80A, however, in practice we can expect to get currents up to 40A if the MOSFET temperature is kept as low as possible, using a big heatsink or even a fan.

That's right, theoretically.  See figure 8 in the data sheet:
https://www.infineon.com/dgdl/irf3205spbf.pdf?fileId=5546d462533600a4015355defac9190c
unfortunately  DC line is missing on the chart, but I'm sure it 'd roll down below 10A if voltage >10V.
Yesterday I  was testing my new toys, SPD15P10PG rated 100V & 15A. And know what? Tested unit releases magic smoke and failed short circuit drain - source in 3 sec, with only  30V & 5A. Heat think was good to keep 150W for a few minutes.

that figure is not very relevant for switching operation, except for the short switch transition the Vds is ~0 or the current is ~0

[Berni]

Yep for switching operation that graph is just to be used to model the transition between the off and on state. By having no DC area on that graph means that you are not supposed to use it in its linear region because this causes only small parts of the transistors structure to turn on, get hot, go into a thermal runaway, get way hotter and blow up the transistor.

These transistors will handle 40A just fine, tho the the pair together will dissipate about 25W of heat, so it does need a fairly chunky heatsink to passively cool, but it is possible. Tho the PCB traces most definitely will not handle 40A. Perhaps up to 15A if you tin them. But you could possibly solder a big thick 4mm2 cable on top of those traces to beef them up.

Where the design might be lacking however are safety features. There is no current sensing of any sort. So if this is powered by a high current capable source such as a large battery then a short circuit will most likely kill two of the transistors in less than a second, if you then attempt to put the motor in reverse this kills the other two causing the whole bridge to die into a short circuit with full battery current flowing trough it until something melts and breaks it. This could be mitigated by sensing current in the ground return, feeding that into a comparator and disabling the bridge in the event of exceeding the current. Similarly trying to hold the MOSFETs permanently on with no PWM signal might cause the bootstrap capacitors to drain down to a point where the FET gets into a linear region, gets into thermal runaway and blows up. This could also be mitigated by a circuit that disables the bridge if the control signal stays high for too long.

[hesam.moshiri]

Tho the PCB traces most definitely will not handle 40A. Perhaps up to 15A if you tin them. But you could possibly solder a big thick 4mm2 cable on top of those traces to beef them up.

Yes, that's why the high current carrying tracks are not covered by the solder mask to be strengthened by thick copper wires and solder.

[hesam.moshiri]

That's right, theoretically.

The RDS(ON) is only 8 milliohm. So it would not generate that much heat

[MasterT]

That's right, theoretically.

The RDS(ON) is only 8 milliohm. So it would not generate that much heat

Correct, my mistake. Didn't realize there is no linear operation area

[Yansi]

8 mOhm .. but at 25 deg C.  Likely 13 mOhm at the typical designed junction operating temperature of ~ 100°C. Less if you put there bigger heatsink, but that of course cost money.  100°C is about the maximum safe long term Tj.

That PCB is very poorly layed out and the traces hilarious for a high current job, tinned or not.

[hesam.moshiri]

8 mOhm .. but at 25 deg C.  Likely 13 mOhm at the typical designed junction operating temperature of ~ 100°C. Less if you put there bigger heatsink, but that of course cost money.  100°C is about the maximum safe long term Tj.

That PCB is very poorly layed out and the traces hilarious for a high current job, tinned or not.

The PCB is not poor. You are hurry to just drop a comment. Tracks are not covered by the solder mask to be strengthened by copper wire and solder. Also, I said to keep the temperature down.

[Yansi]

No, I am not in a hurry.  It is blatantly obvious the person who designed the board didn't even turn his/her brain on.

There is a crapton of empty space on that board for way thicker traces, so it is a nobrainer to use that spaces for thick wide polygon based copper fills, not a 2mm wide tracks and "thick statements" about 40 A which this board will never survive - ever, even with tinned traces.

Not to mention, that any high frequency switching is prohibited on this kind of design that clearly lacks any let alone proper decoupling on each H+L switch pair.

Look, that you have made a video about your board does not make the board magically correctly or well designed. Clearly there is a lot to improve already, without even a need to look at the schematic.

[Yansi]

Look, this is what you should do, if have space on the board. (That below wasn't even a high current inverter to begin with, it is a motor inverter with a phase current of up to a few amps).

You have enough space on that board to make your traces at least three times as thick.