I just built a high power H-Bridge for my inverted pendulum project and I thought I would share it.

I was running it at 45VDC into a 12 ohm, 200W resistor with no problems (thats over 150W). I was alittle over ambitious with this H-Bridge, it turns out my servo motor for the pendulum draws <50W when running. The motor itself is capable of 200W.

There's a few things that need to be noted:
1. High voltage is lethal. attempt at your own risk.

2. This circuit generates alot of electrical noise. Opto-isolation of all logic inputs is definitly recommended. Add additional stiffening/bypass capacitors across all voltage supplies.

I'm thinking about making up a PCB for this circuit over winter break. I'll be sure to post it if I do.

Attached Images

h_bridge.JPG (138.1 KB, 1068 views)

__________________
Jeff
To the optimist, the glass is half full.
To the pessimist, the glass is half empty.
To the engineer, the glass is twice as big as it needs to be.

 

That's a bit over-engineered but there's nothing wrong with that.

I wouldn't worry about working with 45V, it's still considered to be a non-lethal ELV (extra low voltage).

__________________

I do not answer private messages asking for help because no one else can: benefit from advice I may give or correct me if I'm wrong.

Please ask on the open forum if you have a question and I'll be happy to help, if I know the answer.

 

No flyback diodes?

How do those MOS drivers work? Does the driver activate both high and low sides when IN = HI? Or does IN=HI only activate the high side and IN = LO only activate the low-side? Because if it's the former it would create a short and if it's the latter a PWM = HI would make the motor go forward and PWM=LO would make the motor go backwards and jerk the motor fwd/bwd and PWM rates and damage it? Maybe I just don't know how the driver works.

 

The flyback diodes are pictured across the mosfets. Unless those are the body diodes?

__________________

 

I think they're body diodes, because they have the little ticks at the end which make them look like zeners.

 

As long as the switching speed isn't high I've seen some people rely on the body diode for flyback protection, don't know how wise this is.

__________________

 

Well if my understanding is right, the voltage spike will exceed the supply voltage by the diode's voltage drop plus the amount the spike continues to rise before the reverse recovery time of the diode kicks in, so even if the body diode can still handle the current (which it should easily be able to do since the FET can), if the current being switched is high enough it could still cause the voltage spike to get too high before the diode reacts.

 

I was just relying on the internal diodes. I read somewhere that the internal diode will recover much faster than any external diodes.

When you put in a logic 1, it'll drive the highside fet. So by inverting the input of the second driver, it'll drive the lower fet on the other side. A 50% duty cycle will hold the motor still.
The chip also has a built in dead time to prevent shoot-through, which is adjustable on the IR21844's via pin 4. It works very well.

Running at 25kHz, I dont think switching directions that fast will cause any damage.

__________________
Jeff
To the optimist, the glass is half full.
To the pessimist, the glass is half empty.
To the engineer, the glass is twice as big as it needs to be.

 

Other way around.

I see. So your control scheme is more of a "complimentary" nature? PWM<50% causes the motor to spin at various speeds in one direction and PWM>50% causes it to spin in the other? That's the locked anti-phase drive right? It reduces the current flowing through the diodes, but current isn't really the problem when you are using body diodes (and at 48V, 200W the current is relatively measly anyways). I dunno...I just don't trust body diodes.

 

They are both....All FETs have the intrinsic diode. Many have the full spec including switching speed defined in the data sheet. Some are avalanche rated and also have the diode speced as a zener.

Normally what he is doing is not a problem and there are FETs designed just for it.

D.

 

The zener diode is like a relief bypass valve. It only kicks in when the voltage reaches high enough (usually around the voltage rating of the FET). They are designed for repetitive breakdown voltage spikes. This is a protective feature of the FET. Whether or not the FET was "designed" to be exposed to regular high voltage spikes is debatable. I would use the external diode when dealing with any high inductive motor. Since you're drawing < 50w I think you will be okay without one unless the motor is some kind of super high torque low current motor that would have an unusually high inductance.


FYI, a reverse biased diode across and inductor is also known as a 'free-wheeling' diode because the current is allowed to cycle back into the inductor until the magnetic field collapses. When using PWM this makes the circuit more efficient because it maintains the magnetic field in the coils longer as power cycles on and off. If you put in fast switching diodes rated between 45v and the breakdown voltage of the FET then they will serve as both the zener and the free-wheeling diode. The benefit here, is that the diode heats up during the pwm off cycle instead of letting the FET heat up in both the off and on cycles.

 

Can i use C4558C instead of TL082?

Just asking

 

The circuit has low resistance values at the inputs of the opamps so TL082 FET input opamps are not required. Any opamps will work.

__________________
Uncle $crooge

 

Jeff, U5A pin 6 is missing a pulldown resistor, at least in the schematic.

__________________
Ron

 

Hi,
just one question about the design - why are you using OpAmps as comparators instead of real comparators ? This is in general not the best idea.

There was an article from TI explaining why somewhere but I cannot find it at the moment.

Why not try a simple comparator like LM393, LM339 or LM311 ? Should be enough for these purposes. (Actually I am using myself these comparators for H-bridge overcurrent detection).

Petr