H-Bridge IC which can handle high peak currents

[T_guttata]

Hi

I'm looking for an H-bridge IC which can handle high DC peak currents (at least 30 amps). On time and mean power will be very low, therefore I don't expect any thermal issues. I have found one called DMHC3025LSD https://www.diodes.com/assets/Datasheets/DMHC3025LSD.pdf.

I could drive this with an arduino and some transistors. But because I might use many H-bridges, maybe 10 or more, this will require a lot of components. For a low RDs_on gate voltage should be 10V. Is there any easier way to drive those ICs with 5V of an arduino? Maybe a gate driver IC with many channels? Switching time is not very important, 1ms would be enough.

Thanks for your help.

[lucazader]

You are probably better constructing the h-bridge out of discrete fets (all n channel) and then use some half h-bridge driver chips.

Half H-bridge drivers are super common and can be has in easy to use packages for hand soldering, as well as super small packages if you want to get some boards assembled.

Something like the LM5109 or DGD0506 as the driver, and then whatever fets will work for your voltage and current requirements.

If you chose say the dgd0506 and the aon7804 fets, for your 10 h-bridge example it would cost (prices from digikey):
2x 10 x dgd0506 - $0.457ea = $9.14
2x 10 x aon7804 - $0.446ea = $8.92
So a total of $18.06 for 10 channels
Obviously getting cheaper if you start to buy in volume.

[eliocor]

VNH7040AY ?

• Output current: 35 A
• 3 V CMOS compatible inputs
• Undervoltage shutdown
• Overvoltage clamp
• Thermal shutdown
• Current and power limitation
• Very low standby power consumption
• PWM operation up to 20 KHz
• Max working voltage: 38V

Very cheap evaluation board: EV-VNH7040AY :
https://www.digikey.com/en/products/detail/stmicroelectronics/EV-VNH7040AY/7691037
https://eu.mouser.com/ProductDetail/STMicroelectronics/EV-VNH7040AY?qs=YCa%2FAAYMW012cf7rG264jw%3D%3D

Cheap and yet available even if it is an automotive component!

[ajb]

What sort of voltage, load (inductive or?) frequency, and duty cycle?  30A peak, but what about average current?  You said that 1ms switching time is sufficient, but the part you linked has an absolute maximum pulsed drain current of 30A at 10us and 1% duty cycle for the P-channel transistors, which is quite a ways off from what you describe.  Even so, at 30A you're going to be dissipating 60-80W (guessing because the P-channel charts don't go to 30A). 

10x H-bridges with 30A peak each is a shitload of power, even at only 12V you're talking about ~3kW.  That's going to take some care to build no matter what you use for the bridge.  Some sort of overcurrent/short protection would be a good idea, which might lean you toward an integrated bridge like the one eliocor suggests.  Those often have a number of protection features built-in, including shoot-through protection (which ensures that the top transistor is fully off before turning the bottom one on and vice versa) which you will have to manage yourself in a discrete solution.  Many modern MCUs have timers designed for driving bridges that can do dead-time insertion and have other useful features, but even if you're using an Arduino board with such an MCU I don't expect the Arduino libraries expose that functionality.  With a discrete solution you can optimize things a bit more, though, and in some cases it can be easier to lay out.  If not all of your ten bridges need to be active at the same time then depending on the load you may be able to do some multiplexing that would reduce the total component count with a discrete solution.

[PCB.Wiz]

I'm looking for an H-bridge IC which can handle high DC peak currents (at least 30 amps). On time and mean power will be very low, therefore I don't expect any thermal issues. I have found one called DMHC3025LSD

You asked for > 30A peak, and then linked to a part rated at 6A and 4.2A.
Their peak spec is for very short transients, of  (10μs Pulse, Duty Cycle = 1%), so you need to be very precise on how long your > 30A peaks last for.

The other important spec to consider is if you need 100% duty cycle (hi side always on)
If yes, then you need either charge pump drivers with N-FET upper side, or P-FET upper side.

Parts like these have driver+fets in one protected package, easily logic driven, and give current sense output too - if you need 10 or more, overall size can matter.

https://www.infineon.com/cms/en/product/power/motor-control-ics/brushed-dc-motor-driver-ics/single-half-bridge-ics/ifx007t/
https://www.infineon.com/cms/de/product/power/motor-control-ics/brushed-dc-motor-driver-ics/single-half-bridge-ics/btn7030-1epa/

[MarkR42]

As far as I can see, there are several options:

1. Use discrete mosfets, with gate driver chips and other passives as required - this will give maximum flexibility but have quite a high part count - 4 mosfet for a h-bridge and driver chips, several passives etc, depending on how many channels you have it starts getting complicated quickly.
2. Use the high-current "integrated half bridge" chips such as the Infineon BTN8962 - which looks amazing but due to chipgeddon is now unobtainable.
3. Integrated full-bridge chips - which do exist but probably not with the required current,

I'm also looking at doing this right now and it seems I have to use discrete mosfets. There are packages with 2 fairly high current mosfets in a half-bridge (or independent) setup which can be used to save a bit of board space / part count.

But it all depends on how many volts you are using too, if it's lots, then I'm not sure - might need to use igbts or something?!?

[T_guttata]

You are probably better constructing the h-bridge out of discrete fets (all n channel) and then use some half h-bridge driver chips.

Half H-bridge drivers are super common and can be has in easy to use packages for hand soldering, as well as super small packages if you want to get some boards assembled.

Something like the LM5109 or DGD0506 as the driver, and then whatever fets will work for your voltage and current requirements.

If you chose say the dgd0506 and the aon7804 fets, for your 10 h-bridge example it would cost (prices from digikey):
2x 10 x dgd0506 - $0.457ea = $9.14
2x 10 x aon7804 - $0.446ea = $8.92
So a total of $18.06 for 10 channels
Obviously getting cheaper if you start to buy in volume.

Are there any ICs which will control more than just 2 MOSFETs? 4 channel (1 IC per H-bridge) would be nice.

VNH7040AY ?

• Output current: 35 A
• 3 V CMOS compatible inputs
• Undervoltage shutdown
• Overvoltage clamp
• Thermal shutdown
• Current and power limitation
• Very low standby power consumption
• PWM operation up to 20 KHz
• Max working voltage: 38V

Very cheap evaluation board: EV-VNH7040AY :
https://www.digikey.com/en/products/detail/stmicroelectronics/EV-VNH7040AY/7691037
https://eu.mouser.com/ProductDetail/STMicroelectronics/EV-VNH7040AY?qs=YCa%2FAAYMW012cf7rG264jw%3D%3D

Cheap and yet available even if it is an automotive component!

Voltage and current looks ok, but the data sheet says nothing regarding pulse current? In general, what will be the limiting factor, the current itself or the thermal issues resulting from the high current?

What sort of voltage, load (inductive or?) frequency, and duty cycle?  30A peak, but what about average current?  You said that 1ms switching time is sufficient, but the part you linked has an absolute maximum pulsed drain current of 30A at 10us and 1% duty cycle for the P-channel transistors, which is quite a ways off from what you describe.  Even so, at 30A you're going to be dissipating 60-80W (guessing because the P-channel charts don't go to 30A). 

10x H-bridges with 30A peak each is a shitload of power, even at only 12V you're talking about ~3kW.  That's going to take some care to build no matter what you use for the bridge.  Some sort of overcurrent/short protection would be a good idea, which might lean you toward an integrated bridge like the one eliocor suggests.  Those often have a number of protection features built-in, including shoot-through protection (which ensures that the top transistor is fully off before turning the bottom one on and vice versa) which you will have to manage yourself in a discrete solution.  Many modern MCUs have timers designed for driving bridges that can do dead-time insertion and have other useful features, but even if you're using an Arduino board with such an MCU I don't expect the Arduino libraries expose that functionality.  With a discrete solution you can optimize things a bit more, though, and in some cases it can be easier to lay out.  If not all of your ten bridges need to be active at the same time then depending on the load you may be able to do some multiplexing that would reduce the total component count with a discrete solution.

Well, I would like an IC with higher current, but options seem quite limited. However, there is again the question regarding what is limitting the current. Is it the current itself or the thermal issues that come along? Basically A single pulse shall have the energy of approx. a 1000 uF capacitor (25V). Duty cycle will be <<1%. Important: the H-Bridges don't need to be activated simultaneously.

I'm looking for an H-bridge IC which can handle high DC peak currents (at least 30 amps). On time and mean power will be very low, therefore I don't expect any thermal issues. I have found one called DMHC3025LSD

You asked for > 30A peak, and then linked to a part rated at 6A and 4.2A.
Their peak spec is for very short transients, of  (10μs Pulse, Duty Cycle = 1%), so you need to be very precise on how long your > 30A peaks last for.

The other important spec to consider is if you need 100% duty cycle (hi side always on)
If yes, then you need either charge pump drivers with N-FET upper side, or P-FET upper side.

Parts like these have driver+fets in one protected package, easily logic driven, and give current sense output too - if you need 10 or more, overall size can matter.

https://www.infineon.com/cms/en/product/power/motor-control-ics/brushed-dc-motor-driver-ics/single-half-bridge-ics/ifx007t/
https://www.infineon.com/cms/de/product/power/motor-control-ics/brushed-dc-motor-driver-ics/single-half-bridge-ics/btn7030-1epa/

If I understand it correctly the BTN7030-1EPA is a half-bridge with driver IC included? The data sheet states 17A, but nothing regarding peak current?

As far as I can see, there are several options:

1. Use discrete mosfets, with gate driver chips and other passives as required - this will give maximum flexibility but have quite a high part count - 4 mosfet for a h-bridge and driver chips, several passives etc, depending on how many channels you have it starts getting complicated quickly.
2. Use the high-current "integrated half bridge" chips such as the Infineon BTN8962 - which looks amazing but due to chipgeddon is now unobtainable.
3. Integrated full-bridge chips - which do exist but probably not with the required current,

I'm also looking at doing this right now and it seems I have to use discrete mosfets. There are packages with 2 fairly high current mosfets in a half-bridge (or independent) setup which can be used to save a bit of board space / part count.

But it all depends on how many volts you are using too, if it's lots, then I'm not sure - might need to use igbts or something?!?

At the moment I will tend towards 2). I found an IC which sounds even more interesting: https://www.infineon.com/cms/de/product/power/mosfet/20v-800v-automotive-mosfet/20v-40v-n-channel-automotive-mosfet/iauc45n04s6n070h/. Tiny package, pulse current of 120A and it is cheap. The Infineon OptiMOS line looks very interesting. They even have a single MOSFET with a ridiculously high pulse current of 1731 A: https://www.infineon.com/dgdl/Infineon-BSC005N03LS5-DataSheet-v02_00-EN.pdf?fileId=5546d46272aa54c00172b72c00d54a41

The IAUC45N04S6N070H looks promising. I could use two of them with a gate driver. Are there any suitable multi channel gate drivers around?

[ajb]

In general, what will be the limiting factor, the current itself or the thermal issues resulting from the high current?

In many of the integrated bridge ICs the load current is limited by the protection features of the device.  These may be based on the junction temperature, directly on the current through the bridge transistors, or a combination of both.  When overload protection is based on current, they often use the channel resistance of the transistor as the current sense element, so the point when the protection kicks in isn't always particularly well controlled as it would be with a dedicated sense resistor.  The protection system may actively limit current by operating the switch transistors in their active region which will increase resistance (and also power dissipation in the part, which will eventually lead to a thermal cutout), or the output may be just turned off.  You will need to look at the datasheet to understand what sort of protection scheme the device uses and how that might affect your application. 

[PCB.Wiz]

Are there any ICs which will control more than just 2 MOSFETs? 4 channel (1 IC per H-bridge) would be nice.

Yes, you can find 3 phase designs, with 6 mosfets or 6 drivers, but you need to check those to see if 2 can drive 3 full bridges.

Voltage and current looks ok, but the data sheet says nothing regarding pulse current? In general, what will be the limiting factor, the current itself or the thermal issues resulting from the high current?
Well, I would like an IC with higher current, but options seem quite limited. However, there is again the question regarding what is limitting the current.
 Is it the current itself or the thermal issues that come along?
Basically A single pulse shall have the energy of approx. a 1000 uF capacitor (25V). Duty cycle will be <<1%. Important: the H-Bridges don't need to be activated simultaneously.

What does that mean, exactly ? It is a strange way to spec a bridge current load.
Normally you have running current and stall currents, and those are long-duration effects, as far as silicon goes. One stall a day may be << 1%, but bonding wires and die warm up in microseconds.
It is more common for MOSFET circuits to be thermally designed, the peaks they can manage these days are often way above real circuit values.
There is a strong incentive to avoid heatsinks, so SMD packages and RDS values are chosen to make the thermal loading tolerable on a PCB.

If you really are dumping "the energy of approx. a 1000 uF capacitor (25V)" in some sort of impulse circuit, you need to worry about the under voltage trips as what is the end-voltage of your pulse dump ?

What is the actual load you are driving here, with what waveforms, and what are its specs ?