Opinion on H-Bridge Circuit

I'm working on an H-Bridge to control a drill motor that requires 18V and 50A of power. I found and example of an H-Bridge at hvlabs.com - This website is for sale! - hvlabs Resources and Information. . I have studied the schematic, and from my knowledge it should work, but I would like to get a 2nd opinion on it before I buy the parts.

The schematic can be found at http://www.hvlabs.com/Images/dcmotorcontroller.jpg and more info on the HIP4081 can be found at hvlabs.com - This website is for sale! - hvlabs Resources and Information. . I

tecno40:
I'm working on an H-Bridge to control a drill motor that requires 18V and 50A of power. I found and example of an H-Bridge at http://www.hvlabs.com/hbridge.html . I have studied the schematic, and from my knowledge it should work, but I would like to get a 2nd opinion on it before I buy the parts.

The schematic can be found at http://www.hvlabs.com/Images/dcmotorcontroller.jpg and more info on the HIP4081 can be found at http://www.hvlabs.com/files/HIP4081.pdf . I

That's 50A of current.

Typical implementation using that HIP4081 IC.

Ever made or done anything like this before?

That's 50A of current.

I took a more "holistic" view to his post which said "18V and 50A of power" which is correct in the sense that V and A together are power when mulitplied :smiley:

It would probably be best to look on ebay for an appropriate sized h-bridge. The below is good reading if you are thinking of a DIY h-bridge.

That circuit has gate resistors that seem very large (100 ohms). That would slow the switch-on to perhaps 10us,
probably preventing PWM operation altogether. Suggest more like 10 ohms is a better choice to allow high speed
switching (thus reducing losses in PWM). The 18V supply will need good ceramic decoupling right at the H-bridge to
prevent severe bounce on the power rails, and more protection circuitry is really needed to reduce the risk of system failure.

Firstly the logic signals into the HIP4081 must go through resistors, perhaps 4k7's, to protect the microcontroller
when the thing fails - during testing expect problems! The HDEL and LDEL resistors need to be chosen to
ensure enough dead-time to prevent shoot-through conduction - this is vital, failure to prevent shoot-through
will usually result in MOSFETs getting extrememly hot and failing if PWM is applied.

Careful attention to PCB layout is needed in with the HIP4081 and the MOSFETs as described in the datasheet.

Go an have a search for "open source motor controller" project too.

The good news is that 18V is quite low and lower supply means fewer failure modes to worry about in an Hbridge.

The high currents will pose some issues, in particular you will need to detect overcurrent and shutdown the bridge
very quickly - I-squared-R losses in the MOSFETs are important, as are switching losses.

You must expect to test the thing first with a low current load and supply, gradually building up to higher values
if all is well - its surprisingly easy to blow up all the the FETs and the 4081 if you don't proceed very cautiously.

50A really is pushing the devices to their limit (forget what the max I specification is, calculate the power
dissipation - 10W per device (not including PWM switching losses, which may double that figure). Consider 2 FETs
per leg of the bridge, a fan-cooled heatsink is a given too.

Good point about making sure to test the device at a low current first. I will probably buy 10 and 100Ohm resistors and then test them at a lower amperage to see how much they decrease the PWM.

As for the ceramic decoupling, would putting some capacitors in parallel with the H-Bridge work for this?

I will definitely make sure to put resistors between the H-Bridge and board, and I will probably also use a relay to prevent shoot through from a programming error

A few things about that circuit look odd to me. First, the rather high value (100 ohm) gate series resistors with dioes in parallel. Looks like the designer was trying to achieve slow turn-on but fast turn-off. I can't see why that was done, given that the HIP4081A provides adjustable dead-time. I would ditch the diodes (as in the schematic of the evaluation board on the datasheet) and use lower value resistors, around 10 ohms or even less. Second, I can't see that the 10K + 330nF snubber network in parallel with the motor serves any useful purpose, because 10K is far too high a value.

tecno40:
Good point about making sure to test the device at a low current first. I will probably buy 10 and 100Ohm resistors and then test them at a lower amperage to see how much they decrease the PWM.

As for the ceramic decoupling, would putting some capacitors in parallel with the H-Bridge work for this?

I think that's a good start - but watch out for the high ripple currents involved overheating capacitors - low
ESR is important here

I will definitely make sure to put resistors between the H-Bridge and board, and I will probably also use a relay to prevent shoot through from a programming error

Relays unnecessary - the HIP4081 prevents static shoot-through (it would be a crazy H-bridge controller that allowed it),
but you have to set the dead-time adequately to allow for the time taken by your MOSFETs to switch off.
The datasheet shows the deadtime against resistor value on HDEL and LDEL. Even shoot-through for 100ns
can blow up MOSFETs if being PWM'd since all the power of the supply is available (augmented by decoupling
caps).