H-Bridge Voltage Drop Problem

[metaphysics1221]

Ok, first a little backround.

I'm not even going to try and describe exactly how my H-bridge is set up. However, it is the same circuit that is show on the main SOR website here: http://www.societyofrobots.com/schematics_h-bridgedes.shtml

This is the MOSFET I'm using, I purchased them from allelectronics: http://pdf1.alldatasheet.com/datasheet-pdf/view/22130/STMICROELECTRONICS/BUZ11A.html

Unfortunatly, when I connect a 12V battery (with more than enough Amp/hr and discharge rate) and use a small wall wart transformer to simulate logic from my microcontroller, the circuit is only outputting 6.2V! Does anyone have any ideas as to what could be causing this??

The H-briges are otherwise working great, I have a 1k ohm resistor between the signal in and the signal pins and a 1M ohm resistor going from each signal pin to ground. The only real variations I made were to apply power to only one of the top MOSFETS and then run a wire from that power pin to the power pin on the other top MOSFET.  I reasoned this shouldn't be a problem. In addition, I've connected the signal pins of the MOSFETs that share a signal wire together instead of running two separate wires. Again, seems harmless to me. Everything is wired with jumper wires from my breadboard (addictive convenient) and 22AWG hookup wire, all soldered onto a perf board. They do of course heat up a great deal under operation, but the heat sinks keep them operational and, at 6.2V, they run fine for extended periods.

It's not heat damage or anything, they all tested this same way brand new in my breadboard.

Does anyone have any thoughts?

[ArcMan]

It's hard to diagnose without a circuit diagram...

Are you using P-channel MOSFETs on the high side and N-channel MOSFETs on the low side?  You should be for your simple circuit.
Are the minus terminals of your battery and your wall wart connected together?  The must be so you have a common ground.
Then drive your high side MOSFET by connecting the gate to 0V, and drive the low side MOSFET on the opposite motor lead by applying at least 8V to the gate.  You'll get the full 12V to the motor and your MOSFETs will not get hot because they are fully ON.

[metaphysics1221]

1. Nope, I'm not. I'll go check into that, but is it something that could half the voltage?

2. Yep, common ground.

3. I'm not sure I follow, which are you referring to as the HIGH side?

4. 8V? Hm, maybe this is the problem. I've been applying 5V. What term should I be looking for on the datasheet for the "full on" voltage? "Gate Source Voltage" ? Because I was looking at something else, but it says the drain source should be -+20V. Or is it the "Gate Threshold Voltage" which is max. 4v?

[Coleberbot]

Just a question about the 5V vs 8V.   Is the output of the Axon 5v?  If so do we need to use an amplifier transistor to bring the PWM output signal up as high as 10V?

Thanks...... I might be having the same problem

[SmAsH]

yes, the output from most standard microcontrollers on their pins is 5V.

[metaphysics1221]

I'm pretty sure, however, that my MOSFET's are fine with the 5V in. I've done some research, and using the PNP MOSFETs for the high side of the circuit can indeed provide the full voltage.

However, I DO have one more question. On the tutorial, it says to make the top to "PNP transistors"

Not MOSFETS. But in my other sources, MOSFETs are used there too. Does it make a difference?

[Soeren]

Hi,

Look for Logic Level MOSFETs and DO use either a circuit with P-ch. at the top (and the proper drive through a NPN BjT), or use a dedicated driver which generates the voltages needed.

Depending on how you need it set up, the high side could be driven by the opposite low side MOSFETs.

Your problem is, that you never even get the low side fully on and with N-ch devices at the high side, the gate voltage needs to be 12V + a bit more than what takes it  beyond the Miller Plateau for effective switching (i.e. a bit more than 6V for the BUZ11A), which mean that you need to supply the gate with a bit over 18V.
Your low side devices needs their gate at a bit more than 6V relative to the source pin (i.e. 0V).


MOSFET means Metal Oxide Semiconductor Field Effect Transistor.
BjT means Bipolar Junction Transistor.
PNP in a BjT is P-ch. in a MOSFET.

And the mantra that MOSFETs are the only useable devices nowadays only stems from lack of knowledge. Both types have good sides as well as bad sides depending on what's important in the circuit at hand - The clever designers use what's best in a given circuit, based on effectiveness, price, availability etc. (not necessarily in that order, everything is compromises).

[metaphysics1221]

OK, I think I understood most of that

I've found some new MOSFETs on Digikey, PNP and NPN, is it important that they be MADE to work together?

[Soeren]

No, as long as the weakest of them can handle the job (again, with an overhead).

If you want to get the switching effective, you'd need to look at their capacity (mainly [Ciss]) and the Total Gate Charge [Qg]. All parameters in a datasheet is important if you wanna maximice a design (they don't print them, if you don't nee them), but the above mentioned parameters goes a long way in the selection and use for your purpose.

Ciss should be as low as possible for the power handling you need, but the higher the current, the higher the Ciss, so don't just go for a 200A device if your motor has a 20A stall current, that's putting a bullet into your own foot.
Qg tells you how much energy you need to pout into Ciss to pass the Miller Plateau and turn the device fully on.


For easy selection, Fairchild Semiconductors web site is the best I've found so far. International Rectifier makes just as good devices and have done for longer, but they're far beyond Fairchild in making it easy to select what you need.

[metaphysics1221]

Thanks for that info! I did reassess the max current needs I expected, and I decided to go from 20A to ~10A, seeing as my motors are advised to be fused at 3-5A each.

Another question. I've been looking at the circuit diagrams, and the ones that use the PNP MOSFETs on top seem to connect the gates on each the left and right sides, instead of crossing and connecting the top left and bottom right, etc. Why would you do that?? It doesn't make sense to me?

[Soeren]

Hi,

I've been looking at the circuit diagrams, and the ones that use the PNP MOSFETs on top seem to connect the gates on each the left and right sides, instead of crossing and connecting the top left and bottom right, etc. Why would you do that?? It doesn't make sense to me?

Err, the "P-ch. MOSFETs" I'm sure you meant 
No, it doesn't sound right, as that would create a short whenever they were activated (assuming at least one of the N-ch devices was turned on as well).

If you've still got the link, post it and I'll take a look at it.

[metaphysics1221]

Yes, I did mean p-ch, thanks lol.

It's not a website, it's in "The Robot Builders Bonanza 2nd Edition" on page 357.

[Soeren]

Hi,


Is this the schematic?


If it is, I misunderstood your description.
The gates of Q1 and Q2 can (although it's bad design) be connected, since Q1 is open when its gate is "low" and Q2 is open when its gate is "high".

Does that address your concerns?

[metaphysics1221]

Not entirely. So the P-channel transistors are open normally and CLOSE when voltage is applied? Doesn't one usually connect the gates of Q1 and Q3 to each other???

[Soeren]

Hi,

No, the P-ch. device, which is mounted sort of upside-down in the schematic (source and drain are reversed) works opposite of the N-ch. device.

N-ch.
When the gate is positive by say 6V or more, relative to its source, it will open. When the gate-source voltage potential gets under the threshold voltage (say 2.5V, it differs from device to device) is will close.

P-ch.
When the gate is negative by the same 6+V, relative to its source, it will open.  When the source-gate voltage potential gets under the threshold voltage is will close.

The buffered control signal will switch between ground level and +V potential. When at +V, Q1 is off lacking any gate drive and Q2 is on, as it gets the +V potential at its gate.
When the buffer output goes to ground, the reverse situation happens.

Connecting the gates of Q1 and Q3 will toggle between braking and coasting the motor (assuming Q2 and Q4 is off, otherwise... wel, we probably all know the smell and taste of toasted silicium.

Some other problems with the p.357 schematic is the total lack of gate resistors. 10 Ohm to 100 Ohm should be in series with the gates, or it will probably oscillate.
A CMOS buffer is a bad choice for a MOSFET driver, since you need ample inrush current to drive it quickly through the region where both transistors conduct,. as this is in effect a short circuit.