H-Bridge Design

[Falven]

Hello,
I am pretty new to electronics. I want to design an H-bridge so I may switch a Linear Actuator On/Off and reverse the polarity using a microcontroller. The Motor/Linear Actuator is rated at 12V, 5A on load. The Microcontroller can generate 5V and a Maximum of 40mA to switch the circuit. How many inputs would I require from the Microcontroller? What kind of transistors and with what power ratings should I use?

Thanks

-Falven

 

[MrAl]

Hi,


Since this is going to be a one piece job you should go with over rated transistors. At least 10 amps and say 50 volts. You may want to go higher with the current rating to get lower 'on' MOSFET resistance.

You dont need any INPUTS for the microcontroller unless you plan to monitor current. You'll need at least 2 OUTPUTS maybe 4 depending on your design.

 

[Falven]

Hi,


Since this is going to be a one piece job you should go with over rated transistors. At least 10 amps and say 50 volts. You may want to go higher with the current rating to get lower 'on' MOSFET resistance.

You dont need any INPUTS for the microcontroller unless you plan to monitor current. You'll need at least 2 OUTPUTS maybe 4 depending on your design.

How do I know if my microcontroller could switch such transistors?
So I can just buy 4 Mosfetts (What power specs?) and set them up in this configuration, no diodes, resistors or anything and it will work reliably? What would be the duty cycle? Keep in mind I have never designed or built an H Bridge or anything similar really before so detailed help would be really appreciated.
-Falven

 

[MrAl]

Hi,

Well normally you dont use the uC directly but use a driver in between the uC and the transistors. There are many kinds of drivers out there and a search for "H Bridge Driver Chip" should turn up a lot.

Have you done soldering and PC board making and stuff like that before? If so, name some of your past projects. This will tell me where you are "electronically"

 

[Falven]

Hi,

Well normally you dont use the uC directly but use a driver in between the uC and the transistors. There are many kinds of drivers out there and a search for "H Bridge Driver Chip" should turn up a lot.

Have you done soldering and PC board making and stuff like that before? If so, name some of your past projects. This will tell me where you are "electronically"

Yes, of course i've had many projects, just not dealing with transistors before.
I made a Persistence of Vision LED bar using some LED drivers to reduce the number of outputs required from my microcontroller.
Ive also used relays A LOT, hooking up different things.
Ive also hooked up an RFID reader circuit, servos...
Again, I just haven't done much dealing with transistors and larger amounts of power, etc.
What is the point of using a driver? I thought transistors amplified power or could be used as an on off switch 0 - 1, why would I need a driver? Also, wouldn't I need a diode between the power supply and H bridge to prevent EMF?
I'm trying to learn, but some more detailed feedback would really help
-Falven

 

[Sceadwian]

What is the point of using a driver? I thought transistors amplified power or could be used as an on off switch 0 - 1, why would I need a driver? Also, wouldn't I need a diode between the power supply and H bridge to prevent EMF?

Because the output from a micro controller is very limited in it's voltage and current. For a bipolar transistor they often can't supply enough current, for a Mosfet transistor they often can't supply enough voltage, although there are logic level mosfets it is almost aways better to use a proper driver as it will result in faster switching and lower on resistance, as long as the switching speed is not too high no EMF diodes are required for a mosfet based hbridge because all Mosfets have a parasitic diode in anti-parallel with the source drain.

 

[MrAl]

Hi,

Some good points Scead.

Falven:
If you use logic level MOSFETs you *might* get away without using a driver but you still need a transistor (or two) to drive the upper transistors of the H bridge, and you'll have to allow enough dead time for the opposite MOSFETs to turn off before turning on the other set. Switching will be relatively slow but for your app may work if it doesnt have to switch on and off too fast. For example 1kHz might be ok.
You at least need transistors to drive the upper MOSFETs because their gates have to be driven 5 volts more positive than the highest output voltage (which in your case is 12v, that means 17v). If you use P type for the uppers then you have to have at least 12v drive to turn them off.

For transistors arranged like this:

LU RU
LL RL

where LU=left upper, RU=right upper, LL=left lower, RL=right lower,
with all off you would turn on LU and RU for one direction, then turn off LU and RU, then wait at least 1ms (for example).
For the other direction after you wait at least 1ms you turn on RU and LL. Before you can turn LU and RL back on again you have to turn RU and LL off and wait 1ms again. So whenever you turn one set off you wait a short time, then turn the other set on. If there is a lot of time between switches that would work good.

You'll also have to check your particular microcontroller for it's startup sequence. Does it turn all pins high or low or to high impedance? You have to know this so you can design the right drive circuit because you can not have all four transistors turn on accidentally just because the uC is powered on.

Also, what manufacturer are you going to use for the uC chip?

 

[Falven]

Hi,

Some good points Scead.

Falven:
If you use logic level MOSFETs you *might* get away without using a driver but you still need a transistor (or two) to drive the upper transistors of the H bridge, and you'll have to allow enough dead time for the opposite MOSFETs to turn off before turning on the other set. Switching will be relatively slow but for your app may work if it doesnt have to switch on and off too fast. For example 1kHz might be ok.
You at least need transistors to drive the upper MOSFETs because their gates have to be driven 5 volts more positive than the highest output voltage (which in your case is 12v, that means 17v). If you use P type for the uppers then you have to have at least 12v drive to turn them off.

For transistors arranged like this:

LU RU
LL RL

where LU=left upper, RU=right upper, LL=left lower, RL=right lower,
with all off you would turn on LU and RU for one direction, then turn off LU and RU, then wait at least 1ms (for example).
For the other direction after you wait at least 1ms you turn on RU and LL. Before you can turn LU and RL back on again you have to turn RU and LL off and wait 1ms again. So whenever you turn one set off you wait a short time, then turn the other set on. If there is a lot of time between switches that would work good.

You'll also have to check your particular microcontroller for it's startup sequence. Does it turn all pins high or low or to high impedance? You have to know this so you can design the right drive circuit because you can not have all four transistors turn on accidentally just because the uC is powered on.

Also, what manufacturer are you going to use for the uC chip?

It seems like going with a driver would provide me with more flexibility in duty cycle, what about reliability/longevity?
My uC is an Arduino Duemilanove
From **broken link removed**

Pins configured as OUTPUT with pinMode() are said to be in a low-impedance state. This means that they can provide a substantial amount of current to other circuits. Atmega pins can source (provide positive current) or sink (provide negative current) up to 40 mA (milliamps) of current to other devices/circuits.

So from here, do you know what kind of "Driver" and transistors I should get, their specifications are very different from those of Relays, so I am confused. I was looking through mouser and digikey.
Thanks
-Falven

 

[Falven]

Bump
I feel like I am very close to an answer!

 

[ronv]

Here is a driver.

https://www.electro-tech-online.com/custompdfs/2012/04/ir2184.pdf


And a FET
https://www.electro-tech-online.com/custompdfs/2012/04/FDP6030BL.pdf

Post your circuit as you go along.

 

[Falven]

Here's the shematic so far. I am not sure how to hook up the H-Bridge drivers as I have never seen or played with one before. I did not include diodes on the H-Bridge. How do I hook up the driver?
View attachment 63981
Thanks,
-Falven

 

[ronv]

OK.
You need 2 IR21844 drivers - 1 for each side of the H Bridge.
Following the diagram in the data sheet for the IR21844:
Vcc to +12
Vcc capacitor .1 Ufd. ceramic
In is the input from your micro
Dt resistor 47K
R gate (from the outputs to the FET gates) 33 ohms
Diode 1N5817 or other schokky diode
Vss to your micro ground.
Cap from the diode to Com .1 Ufd. ceramic
Com to the source of the bottom FET (power ground)
Cap Vb to Vs .047 Ufd ceramic.
Keep the leads short from the driver to the gates and sources of the FETs.

 

[audioguru]

Your schematic correctly shows P-channel Mosfets on the top and N-channel Mosfetrs on the bottom. But they wrongly all have the N-channel part numbers.

 

[Falven]

Here is an updated Circuit Diagram.
I am not sure on the positioning of the two bottom Mosfetts, and shouldn't I connect the FET's across form each other to the same driver, instead of both on either side?
View attachment 64003

Here is the list of parts:
4 FDP6030BL Mosfets
2 IR21844 drivers.
Vcc - capacitor .1 Ufd. ceramic
IN - 40 mA PWM Microcontroller output.
Dt - resistor 47K
R gate - (from the outputs to the FET gates) 33 ohms
Diode - 1N5817Vss to your micro ground.
Cap from the diode to Com .1 Ufd. ceramic
Cap Vb to Vs .047 Ufd ceramic.

 

[ronv]

Just about. The FETs are upside down. The source on both the top and bottom ones should point down.

No each driver goes to the FETs on the same side of the H Bridge (see figure 4 in the data sheet for the timing), In that way the driver can make sure both FETs are not on at the same time creating a short.

For your parts: You can use 1% resistors and capacitors rated at 50 volts and save some $$$$.

Don't forget your clamp diodes. 3 amp schottky diodes (MRBS340) would be good.

 

[Falven]

Ok, I revised the schematic. I flipped the FETS (Is this because electrons, in reality, travel from ground, into the source of the FETS?). Oddly enough, the schematic in the Driver's data sheet lists the FETS the other way around, source up, drain bottom.

View attachment 64024

Don't forget your clamp diodes. 3 amp schottky diodes (MRBS340) would be good.

This is the diode between Vb and Vcc? Is there something wrong with the ones I listed? (#6)
I changed some parts, but not all:
Ceramic Cap at .1uf 50V 1% were a lot more expensive.
How did you find the parts/power ratings required by that schematic?
Lastly, I changed all the confusing part names to numbered items pertaining to the numbers in the schematic.

0. 6 x N-Channel Mosfetts
1. 3 x IR21844 drivers
2. 6 x Capacitor .1 Ufd. ceramic
3. 40 mA Digital Microcontroller output
4. 3 x 47K Resistor
5. 6 x 33 ohms Resistor
6. 3 x Schokky Diode 1N5817
7. To Microcontroller Ground
8. 3 x .047 Ufd Ceramic Capacitor

Thanks for all the help thus far!
-Falven

 

[ronv]

the Driver's data sheet[/url] lists the FETS the other way around, source up, drain bottom.

I think the source is down.

This is the diode between Vb and Vcc? Is there something wrong with the ones I listed? (#6)

No those are fine. You need some diodes to clamp the inductive kick from your actuator. See attachment.

I changed some parts, but not all:
Ceramic Cap at .1uf 50V 1% were a lot more expensive.

Sorrry, I was talking about 1% resistors. The caps can be 20%

How did you find the parts/power ratings required by that schematic?

The driver I like because it prevents the 2 fets from being on at the same time and allows you to use NFETs for both high and low side.

The .047 Ufd. This cap stores the voltage used to turn on the top fets. It just needs to be about 10X larger than the gate capacitance of the FET. I always make it a bit more than that.

The .1 Ufd. are just decoupling caps to supply the driver. Ceramic caps are good for this because they have low ESR (equivilent series resistance) and good high frequency response for the high speed driver.

The 33 ohm resistors are to help damp oscillations caused by the capacitance and inductance in the gate source path. The larger it is the slower the fet will turn on. 33 ohms was just consistant with the data sheet. The power is very low because once the gate capacitance of the FET is overcome there is no current flow.

The 47K sets the time that the FETs have to be off before the other can turn on. There is a chart in the datasheet showing the "dead time" vs the resistor value. I chose 47k to give plenty of time.
[/QUOTE]

 

[Falven]

No those are fine. You need some diodes to clamp the inductive kick from your actuator. See attachment.

no EMF diodes are required for a mosfet based hbridge because all Mosfets have a parasitic diode in anti-parallel with the source drain.

I will gladly leave these in though, better safe than sorry.

View attachment 64035

Can I cross the two uC grounds? Can I also cross SD to one digital output to turn off the whole system?
Also, I should be using a digital output on my uC, right?

I updated the Schematic, and parts. About to buy the parts, do these all seem right for the job?

0. 6 x N-Channel Mosfetts
1. 3 x IR21844 drivers
2. 6 x Capacitor .1 Ufd. ceramic
3. 40 mA Digital Microcontroller output
4. 3 x 47K Resistor
5. 6 x 33 ohms Resistor
6. 3 x Schokky Diode 1N5817
7. To Microcontroller Ground
8. 3 x .047 Ufd Ceramic Capacitor
9. I couldn't find MRBS340, so 6 x 3A 20V Schotky Diode?

If so, once the parts arrive I will post back with pictures of my progress, errors, and final results of the project.

 

[strantor]

I've never seen an h-bridge with the diodes forward biased in the bottom quadrants. Is that by design ronv? Looks to me like they would short out.

 

[johansen]

I've never seen an h-bridge with the diodes forward biased in the bottom quadrants. Is that by design ronv? Looks to me like they would short out.

neither have I

some learn by releasing the magic smoke for themselves i guess...