H-bridges again

Hi all,

I am still plugging away for the 100% perfect H-bridge design. While the driveing wave is 100% perfect, rock solid, down to just 10mV noise, totally fantasic, and yet. still H-bridge problems, and that should be the easy bit right ?

anyway, half bridge mode, can run either side seperate and everything works fine. but when you plug into full H-bridge mode one half of the bridge is having a interaction with the other half.

Now I have gone though piles of mosfets, you can have a totally perfect squarewave or have it totally mashed and the H-bridge output still works but theres some slight noise ripple there which I can't trace...

Tesiting was as before 50:50 duty cycle and its gets a tad hot, get down to about 45:45 and works fine(ish), works like any other H-bridge... Run at

50:30 and the noise problem goes away, run at 30:30 and noise comes back... So deadtime is not having any effect other than stopping cross conduction.

Load on the H-bridge does not matter one bit, you can have it O/C or loaded with several amps and waveforms etc are solid.

The only thing I can find is when you use a 1K gate resistor to charge the mosfet gate (read more before you yell), the mosfet gate works perfectly with no voltage on the mosfets, and as you increase voltage on the mosfet you start to get what appears to be a switch bounce, and the higher the voltage is the more bounce you get. While this is to be expected with a poor gate driver its the only thing I can see which is odd. As you decrease the gate resistor the bounce gets tighter until you no longer see it, though you do see dots on the rise so its still there.. but in anycase even though its a bit odd its not the problem...

If you run some PWM via the bridge, you end up with about 0.5V noise on the output, this only happens in full bridge, on half bridge everything runs perfectly... There seems to be some interaction between the 2 half bridges which is messing up. I've tried placing caps across the mosfet rails and this makes it worse, the more caps you put in the worse it gets. It appears to be down to the capacitance of the mosfet itself, the only one which seems to have effect is the Reverse Transfer Capacitance, I have tried running diodes in series with the mosfet to block the body diode to see if that had any effect... and nope...

Voltage on the bridge does not seem to have much effect, it does get (I think) a little worse on higher voltages though its hard to tell. I can run at 50:30 ratio and the problem goes away totally, the more you get to 50:50 the more of this noise you seem to get, reducing from 50:30 to 30:30 and the problem starts to come back again, running half brige works fine... so what am I over looking here ? some odd interaction between the 2 half bridges ?? Have read over and over about such things, also was using a lot of Texas stuff and actually been taking to one of thier app guys for months, (nice that a manufacture actually cares about its customers!!) I sent them all the results, all the waveforms, explained it all to them for weeks, and they are clueless, I mean you can't exactly build it wrong, it works but I want to know why it is not 100%.... and no im not going to be happy with 99% either ;-)

Chris

Reply to
exxos
Loading thread data ...

Hi When using one half only of an H-Bridge IC - the other half must have = it's inputs grounded i.e. at 0v. If they are not they will float at any = level, sometimes oscillate and this will cause the type of problems you = are describing. It will also get quite hot. Maybe the circuit you are using is allowing one side to float when the = other side is in use.

I know this applies to IC's but it may help in your case.

Alan

the=20

totally=20

easy bit=20

works=20

is=20

perfect=20

works but=20

down to=20

at=20

back...=20

conduction.

loaded=20

the=20

perfectly=20

mosfet=20

the=20

a poor=20

the=20

though you=20

though its=20

on the=20

runs=20

bridges=20

and=20

appears=20

seems=20

running=20

that had=20

(I=20

can run=20

50:50=20

and the=20

what=20

bridges ??=20

Texas=20

(nice=20

all the=20

they are=20

to=20

either=20

Reply to
Z80

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.