Go Down

Topic: MOSFET H-bridge Voltage too low. (Read 2914 times) previous topic - next topic

Jonathanese

First of all, I have my circuit attached as an image. The text file can be imported into this java applet.

http://www.falstad.com/circuit/


The H-Bridge performs exactly as expected logically. However, the voltage drops across the MOSFETs are enormous. When I'm working with 12v, an ~5.5v drop across 2 transistors results in ~1v for the motor. The intent was to get at least 9v to the motor.

Which means there's something fundamental about MOSFETs that I'm forgetting. Can you help me troubleshoot this?

MarkT

Firstly you've used the wrong symbols for power MOSFETs, which have an
integrated body-diode as part of the symbol.  No need for other diodes as a
result.  BTW you must distinquish source from drain in a power MOSFET.  You
need source to the power rail, drain to the motor in this kind of bridge.
Top-side switches p-channel, low-side n-channel logic-level.

The circuit is wrong in various ways:

1) Shoot-through - you turn on MOSFET on as the other is turning off.  This is really bad.
You must turn off one completely before the other starts conducting otherwise very large
current pulses flow through both MOSFETs, potentially destroying them and wasting a
lot of heat in them.   Thus you must separately control the high and low-side switches on
each leg of the bridge and allow some dead-time between switch-off and switch-on.

2) gate resistor.  This should be low in value, perhaps 0 ohms even...  MOSFET
gates are highly capacitive, trying to drive them from a 33k resistor will result in
ultra-slow turn on and turn off times, dumping huge amounts of power wastefully
into them if PWM is used.  If driving a MOSFET direct from the Arduino it must
be logic level and its wise to use a 150 ohm resistor to protect the Arduino pin
from excessive current (the MOSFET gate would happily and usefully take more than
the Arduino can supply while charging/discharging).  Something like 1k or 470 ohms
for the pull-up resistor on the p-channel gate is usually good (limited by heat dissipation
in the resistor perhaps).

[BTW which actual devices are you using?  How much current does the motor
take under load and at stall, what's the rating of the power supply?]
[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

Jonathanese

-Got it. So diodes are normally built-in. I'll have to verify that with the part numbers.

-These were J-FET symbols built into AutoCAD. I later modified them to look more like MOSFETs.

-I just realized I didn't actually need those resistors. I had them on there in a previous version where the gates came directly from an arduino, and wanted to counter the capacitive effects of MOSFET gates.

"[BTW which actual devices are you using?  How much current does the motor
take under load and at stall, what's the rating of the power supply?]"

I haven't even fabricated any part of this circuit yet, so this is really all irrelevant to why a simulation would be saying the motor is only getting 1 volt. But I guess I'll be getting 12 volts from a 300W Power Supply, and I'll be powering probably a 9v LEGO motor or something. Those things draw something like 300mA stalling. I don't have the transistors on me right now, but I remember they were rated at something like 60Amps peak and 20Amps sustained, though I'd imagine the transition times and gate capacitance are the specs you're looking for.

Elecrow

i think you should firstly learn some basic usage of MOSFET to create H-bridge. the whold schematic is wrong. :smiley-sad-blue:
i suggest you take this schematic for reference: http://www.elecrow.com/wiki/index.php?title=Dual_Channel_H-Bridge_Motor_Shield
Hope it helps for you!

Jonathanese

Dang, that bad, huh?

It's okay, it was just something I threw together based on what I know about a switch-based H-bridge and my engineering courses on electronics, where we covered BJT's and MOSFETs. And then I connected the gates together with the intent of making it impossible to generate a short. (Oddly enough, creating one during the switching transition.)

I'll have to give that article a read, but I'm at the presumption now that a solid-state H-Bridge isn't something you can do as a hobbyist with discrete parts.

MarkT


-Got it. So diodes are normally built-in. I'll have to verify that with the part numbers.

Power MOSFETs(*) without diodes don't exist, the diode is fundamental to the device
geometry, the symbol for a power mosfet has the diode as part of the symbol.  This
doesn't apply to IGBTs though.
Quote


-These were J-FET symbols built into AutoCAD. I later modified them to look more like MOSFETs.

Just draw it and take a photo, unless you have the right tools!
Quote


-I just realized I didn't actually need those resistors. I had them on there in a previous version where the gates came directly from an arduino, and wanted to counter the capacitive effects of MOSFET gates.

"[BTW which actual devices are you using?  How much current does the motor
take under load and at stall, what's the rating of the power supply?]"

I haven't even fabricated any part of this circuit yet, so this is really all irrelevant to why a simulation would be saying the motor is only getting 1 volt. But I guess I'll be getting 12 volts from a 300W Power Supply, and I'll be powering probably a 9v LEGO motor or something. Those things draw something like 300mA stalling. I don't have the transistors on me right now, but I remember they were rated at something like 60Amps peak and 20Amps sustained, though I'd imagine the transition times and gate capacitance are the specs you're looking for.

300mA??  You can get an H-bridge on a chip for that.

(*) vertical current flow
[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

Jonathanese

I retracted my previous post. I was in a bad mood and felt you guys were essentially doing everything except helping me create a MOSFET H-Bridge. I lost it at "Just draw it and take a photo, unless you have the right tools!" and figured you  had no interest beyond finding things to point and laugh at.

Nevertheless, my circuit was based on this: http://www.bristolwatch.com/ele/h_bridge.htm which shares a lot of the same properties as my circuit. I didn't originally change the symbols because I ran a simulation in the circuit simulator before cleaning it all up, and noticed a huge drop, thus posting it here. I figured you guys would get the point, or a least see past it, but I didn't expect that the DS voltage drop would be completely ignored behind things like using the wrong symbol.

I think I will just get an h-bridge chip instead of actually learning  about solid-state h-bridges by making my own.

MarkT

Alas that site's circuit is a bad circuit, and its very hard to know which circuits are
good and bad out there on the internet without getting advice first.

To be fair with the devices they quote (as opposed to modern devices) the shoot-through
currents would be limited to dozens of amps, not the hundreds that would flow if
using low-on-resistance devices available today.  Imagine 12V lead-acid battery driving
two 10 milliohm loads in series - 600A would flow.  With that 10k pullup resistor the
shoot-through condition would last for perhaps 10 to 20 microseconds, meaning 100mJ
dumped in the devices on each switching.  For 1kHz PWM that's 200W dissipated in
the devices (although the bonding wires will vaporize first).

These are the sorts of considerations you need to consider when building an H-bridge -
dissipation due to switching transients and other losses - its worth persevering to
make your own, but this example circuit is just going to explode MOSFETs, and I
mean explode, because its happened to me.

Here's a much more plausible H-bridge design out there, I'd just change R3 and R4 from
10k to 560 ohms to make it switch faster. http://www.electro-tech-online.com/threads/planning-to-build-this-mosfet-h-bridge-motor-control.93196/

Note that so long as you activate only one of the inputs at once, leaving a few us after
turning one off before turning the other on, shoot-through cannot happen.  Its a clever
design where the high switches turn on the opposite low switch automatically.  Alas it
only works for about 12V supply and doesn't generalize.

Have a look at the HIP4081A datasheet, it shows you the generic approach
that's universal these days.
[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

Jonathanese

Hey, thanks for the good info!

I think I'll also look around for some H-Bridges in a chip as well.

Go Up