Simplifying H-bridge design

[cadena]

Hi everyone,

I am currently trying to build a PWM H-bridge, it has a power line of 12 V and the arduino uno provides the PWM. The MOSFETS I have are: IRLZ44S (N-channel) and FQD11P06TM (P-channel). Basically, the p-type MOSFETS go in the upper side of the circuit and the n-channel go at the bottom. I would be building 6 of these circuits to run 6 motors in an ROV. At the moment I am using this circuit (see image below) but I have been told that it might have problems due to problems of straight through currents which could burn the MOSFETS so I would have to add some capacitors. I honestly don't know which resistor values I should use and where to place the capacitors (or their values). Any help will be appreciated. I am more than welcome to accept a different H-bridge circuit which is simpler than the other one as it uses BJTs and, ideally, I would only like to use a simple circuit with only 4 transistors (the MOSFETS). Thank you for your support.


 

[JagV12]

If you want a cheap and simple circuit, search the web for "arduino PWM H-bridge" and you'll find plenty of ready made circuits (dirt cheap L298N dual channel board is rated for 35V and 2A max, not that bad !)

But if you want to learn electronics, have a look at the (IRLZ44S and FQD11P06TM) datasheet and look what happens when Gate voltage ramps from 0V to 12V on the left (or right) branch of the bridge. If both transistors are conductive at the same time  , you're in trouble...

[cadena]

Hi, thanks for your reply. I am trying to build the H-bridges using MOSFETS rather than ICs where I can't see what is going on and also trying to not spendore since I am nearly over budget. I am trying either to solve this circuit or find another one which uses N and P channel MOSFETS.

[cadena]

I was thinking, could this be solved by writing down a delay in the arduino code ? If this could be solved in this manner, how long should the delay be ( I am using arduino uno PWM) ?

[JagV12]

You can't use software delays here because the same optocoupler drives both top and bottom switches. If you just do a forward-reverse H bridge, you can use RC delays (+ diode) to create a Break before Make switching. But the slow Gate voltage change creates a slow Off-On transition => efficiency loss. Adding Schmitt Triggers is a solution.

But you're right, software delay is OK if you use 4 output pins instead of 2 and control each transistor individually (or 2 if you drive Forward (TopLeft-BottomRight) and Reverse (TopRight-BottomLeft) switches as a pair with the appropriate logic gates). Once again, search the web, there should be many nice solutions out there...

[T3sl4co1l]

More capacitance isn't the answer... that thing would barely survive two transitions, if your supply (and bypass) were beefy enough.  With average conditions, probably it will work, but it will get really hot or explode if it does much switching at all.

Simplifying isn't the answer, in this case.  Or, simplify the complication by using an integrated bootstrap gate driver like HIP2101.  Bonus, better performance because you can use all N-channel MOSFETs, no need for P.

Tim

[jonroger]

As others have said, either use separate opto-couplers or gate drivers.  And software that is careful about shoot-through.

[max_torque]

If you need to "save money" buy an off-the-shelf H bridge!  Sure you might think that it is "expensive" but once you've blown up 12 sets of mofsets, and wasted 3 months of your time, the cost will suddenly look very reasonable indeed!


If you want to learn about H bridges, then yes, make one, but what it won't do is "save" you money....

And of course, if you buy a working H bridge, then the quickest possible learning route is to understand how that working device actually works.  Probe its signals, reverse engineer its circuit, and you'll learn a lot more, a lot faster than starting from scratch yourself ime   

[Benta]

cadena, that design will burn. Back to the drawing board. You need separate drive for high side and low side. No way around it, sorry.

[soldar]

You need to avoid shoot-through, either by software or hardware.

You can do a "pseudo H bridge" with only two FETs on one side of the H and the other half replaced by two capacitors (voltage divider) and the whole thing fed at twice the voltage. I have used that configuration and it makes things much easier.

[cadena]

Hi, thank you all for your replies but I still do not know how to fix this circuit. Please if someone knows a better (and simpler) H-bridge than this one would you posting it, please ? Or at least tell me exactly how to fix this one ? I am still learning how to design circuits and I can't use an IC H-bridge. Please any help would very much appreciated.

[ajb]

The only bridge that will be both better and simpler (for you anyway) is an integrated one, as everyone else has already told you.  In order to avoid shoot-through in the bridge you need to be able to drive the top and bottom transistors separately.  You could do that by using separate optocouplers for each transistor, and in turn driving them with four separate PWM signals so that you can insert the required dead time between turning one switch off and the other on.  This requires careful configuration and the performance will still suck badly enough that it might still burn up because using a pull-up or pull-down resistor and an optocoupler makes for a horribly slow gate driver.  This means that switching takes a long time so at best it limits your switching frequency and at worst the MOSFETs will wind up working outside their safe operating area and die.

Sorry, but this is a situation where if you want it to work you really need to go back to the drawing board.

[sourcecharge]

Hi, thank you all for your replies but I still do not know how to fix this circuit. Please if someone knows a better (and simpler) H-bridge than this one would you posting it, please ? Or at least tell me exactly how to fix this one ? I am still learning how to design circuits and I can't use an IC H-bridge. Please any help would very much appreciated.

You could use the logic from a DFF (CD4013) and 4 NOR (CD4001) gates with a PWM input signal.  The PWM input signal would be 2x as fast as the actual frequency of the H bridge, and if you want to use it up to 20V the PWM input would probably need to be stepped up to the logic voltage as function gnerators only usually go up to a 10Vp.  The NOR gate outputs can directly gate the H-bridge at any duty cycle your PWM input is at, so if you use 1% duty cycle pulse, there would be a 0.5% duty cycle time between the ON times of the N and P channel mosfets.

The clock of the DFF is tied to the PWM, and 2 NOR gate legs.  The DFF has Q not tied to the D input and 1 of the NOR gates that are tied with the PWM on it's other input.  The other NOR gate that has the PWM on one of it's inputs, has Q tied to it's other input.  The outputs of these NOR gates have their signals inverted by the last two NOR gates.  The outputs of all four NOR gates can be used directly to their coresponding mosfet of the H bridge.  You should be able to see for yourself how this 2 chip 20V circuit works without resistors, optocouplers, boot straps, or caps.

It's a simplistic circuit, it's my own creation.  It's sort of cheating telling you about it, as it took me alot of thought into making it.  You can use any mosfets with it depending on how good you want them to be, because you can decide how much shoot through, if any, you want there to be by manually changing the frequency and duty cycle of the PWM input signal.  Old CMOS logic is good up to about 10Mhz

Hope that helps..

[Mechatrommer]

Hi, thank you all for your replies but I still do not know how to fix this circuit. Please if someone knows a better (and simpler) H-bridge than this one would you posting it, please ? Or at least tell me exactly how to fix this one ? I am still learning how to design circuits and I can't use an IC H-bridge. Please any help would very much appreciated.

you cant fix that youtube circuit in OP easily in SW, you can try fancy hardware RC delay fix etc but you can end up much complexity or poor efficiency... i'm with you for not using ready made H-bridge IC as they are limited to below 10A at affordable price, more than that, its unaffordable or difficult to find i think. once a good member guided me for more efficient minimal bjt count discrete driver....


from https://www.eevblog.com/forum/beginners/is-bjt-power-h-bridge-motor-driver-obsoleted-by-fet-h-bridge/msg759287/#msg759287

now i upgraded my "obsolesenced" bjt h-bridge to fet version (pcb ready to be fabricated). i dont use his version as although 2 bjt per side, it uses 3 diodes, and my smd diodes stock are big foot print i want minimal footprint, so i device my self driver 3 (sot-232 small) bjt, no diode needed... albeit more efficient, his version cannot deal with |Vgs| < |Vds| spec fets, mine will take |Vgs| as half the |Vds|. like my previous bjt h-bridge before, instead of trying to control each side (half of each bridge) from one input (like in youtube circuit in OP), each of my input will control each complete bridge, so deadtime (delay between input) can be programmed easily in mcu. i suspect turning on will be acceptable, since gate charges from turned-off upper p-mosfet will be used to charges its counterpart lower n-mosfet on the other side, but turning off will be a bit inefficient due to 1Kohm switch off resistance (havent build it in reality, only in sim). more efficient discrete h-bridge will requires more than 4 bjt on each side (push pull on each fet at least minimal).


from https://www.eevblog.com/forum/projects/h-bridge-motor-control-circut-burns-mcu-after-some-time/msg2320872/#msg2320872
the original circuit from where i got this idea i already forgot where its from, from internet i think, so may God bless the originator and show him the true way, ymmv.

[Doctorandus_P]

MOSfets in an H-bridge configuration only work well if you switch them properly.
Power MOSfets have a pretty high gate charge and this needs respectable current to charge / discharge quickly.

Also, P-channel MOSfets's are more expensive / or have worse specs than N-channel Mosfets.

Therefore a lot of different chips have been designed to solve this.
The IR210x gate drivers are an example of this.
They also have a 600V or so isolation barrier built in, and dead time generation to prevent shoot through, even if the software in your uC fails.

You can also have a look at the BTS7960.
This is an integrated Half bridge.
With 2 of these TO220 sized packages you have a complete H-bridge solution.

[jonroger]

I agree, say the BTN8962TA.  The protections and simplicity are likely to be worth it.