Hi all,
I'm a Mechanical Engineering underghraduate that got stuck with the job of creating a triple half bridge motor driver for some hobby BLDC (A2212) motors.

The motors have a max stall current of 10 amps at 12 volts and as such I am trying to make all of my components meet or exceed that rating. Though these motors are just used to launch ping pong balls so I expect they're just going to be driven at the no load current of 0.5A.

My problem is that I initially purchased some L6385E drivers for the fets so I wouldn't have to deal with P-channels and N-channel combinations. This hasn't worked out well for me as I've had a ton of trouble with these drivers, the nail in the coffin for them is that we've switched to a Teensy 3.1 for control and these drivers will not take 3.3v logic input.

My current situation is that I have a boost converter to supply high voltage from my main 12v power supply and I would like to still use my N-Channels (FQA32N20C) in both high and low positions. I think this is possible because I have the boost converter left over from trying to get the l6385's to work so I have the high voltage needed to turn on the high side N-Channels.

What I'm currently thinking is attached as test-driver.png. I've been looking online and it seems that people use a similiar circuit but with Q2 and Q3 reversed. I don't quite understand why that is. Wouldn't that invert the logic signal and never fully turn on the low side pnp? What am I missing?

Given that I already have 2 voltage sources, would my test-driver work? How could I make it better? Keep in mind that Vin is about 20v and VDD is 12v, I have no idea about naming conventions so I apologize if this is completely against the norm. The lines on the left are 3.3v login input and the single line on the right is the motor output.

I technically have no budget left but if anyone has a suggestion of a cheap component that would greatly simplify this, please feel free to suggest. I am pretty sick of IC's from st electronics though.

 

What am I missing?

Avoiding the shoot-through problem (read up about that). What drives Q5?

 

Q5 is just there right now to show that there is a low side, once I figure out the proper totem configuration I'll have to implement that there aswell. Though since these are logic level n-channels if I can persuade my group to go with a 5v mcu I could just leave it as is.

I'm not sure how a totem configuration prevents shoot through on a single fet other than slamming it on and off by using the source voltage. I do get that the pnp and npn switch at different voltages. I'm just confused why they're using a pnp on the low side and a npn on the high side? Why not arrange the bipolars as I have them?

To avoid shoot through we have a 1-3 us delay in the commutation sequence. Right now I'm mostly worried about the high side N-channel. After doing a bunch more reading I'm considering just going over budget and buying some specialized bldc predrivers while lazer etching a pcb. Is there a good way to implement this driver with discrete components while keeping in mind I have 2 voltage sources? All the other examples I've seen are about using a bootstrap circuit with one source and I'm not interested in implementing that again.

 

while lazer etching a pcb.

You seem to be a rich man. Are you?

 

No I just have access to an advanced robotics research facility at my university.

 

I was guessing something like that.

Laser burns the coating and then you drop it into etchant?

 

I've been looking online and it seems that people use a similiar circuit but with Q2 and Q3 reversed. I don't quite understand why that is.

It's because with your circuit as shown, when Q1 turns on so does Q2. There is shoot-through current via Q2 base and Q1 collector. Swap Q2 and Q3 and the problem disappears.

 

I see that now Alec_t, thank you.

The graduate student operating the lazer cutter for me claims the lazer can cut the copper by it self and we just need to program in the correct timing in order to not cut too deep. I'm a bit skeptical however as in my mind, the copper would distribute the heat and melt the entire pcb. As such, I was considering just using some coating and chemically etching afterwards if that didn't work out.

With Q2 and Q3 reversed this seems to just be a classic totem configuration and as such I'm guessing that it will work fine. I'll test it tomorrow with the components that are available to me. Thanks for your help and I'll update with the result. In case it doesn't work, are there any bldc predrivers or half bridge drivers that people have had good experiences with? I found the l6385e greatly differed from what their datasheets claimed and would be interested in any components people had good experiences with.