AWD BLDC driver project - driver question

[Nikan]

Good morning everyone!

Before I start, I am still a student, so please don't be too hard on me 

So I was thinking about this high speed AWD go-cart for a while...
and now that I have some free time I decidet to start it.

I have already Build a BLDC driver futuring a MP6532 running 6 CSD19535KCS NMOS at 40V from a linear supply capable of 10A for testing. I thought the MPS driver would be a perfect fit because it is prettey simple to set up, but it seems like that chip aint workin right. The motor is spinning and I can adjust the speed, but there is almost no torque behind it at low speeds, it getss better at higher speeds though. I can also only measure around 2A of current draw when I try to stop the motor. I have scoped one of the high and low side gates of the mosfets and I have to admit I havent seen such clean switching for a while! But I think there is something wrong with the driver. Or I am just to dumb to understand whats going on here!
measuring one of the mosfet gates while trying to stop the motor I can barely see a change in frequency.
I will attach a a picture of my board, routing is prettey clean as well as all signals well at least for me I'm not routing boards everyday...
DISCLAMER I have forgotten to put the silk screen on while ordering those pcbs

Anyways I dumbed that idea, but I would still love to find out why that driver is not pushig those mosfets!
Open for any ideas to find my mistake!

While that board is still not working as expected I came up with someting a little more complex.
I thought of using 3 high/low-side mosfet drivers to program the deadtime per software.
I was also thinking of using 3 AND gates and distribute the PWM signal from the PIC18 that way.
A rough schematic of what I mean will be attached as well so its clear what i mean.

I would like to hear what you guys think about driving a 3 phase motor that way and also know if anyone has a clue why the mp6532 is troublesome.

BTW as soon as I get the motor running on its full potential I will make 4 identical boards and connect them via ISP to a raspberry where the torque vectoring fun begins, well at least thats what I have in mind so far 

pcb size is 40x80mm and If I need to make new ones, they cant be bigger due to the casing I already have.

thanks in advance and best regards from Germany!

[H.O]

Regarding the MP6532 (which I have not uesd BTW).
As far as I can see the MP6532 itself is not a (closed loop) speed controller, all it does is run the 3-half bridges according to A) the PWM input and B) the hall sensor feedback from the motor.

When the dutycycle of the input PWM signal is low (as in trying to run the motor slow) the average motor current (and therefor motor torque) will be low.

Since you don't show your schematic it's "quite" difficult to know what you're doing but if you want to control the speed while maintaining torque you need to have some sort of closed loop system that monitors the speed of the motor and adjusts the dutycycle of the PWM signal going to the MP6532, increasing the dutycycle (and therefor current/torque) when the load on the motor increases.

You will need to setup the OCP feature of the MP6532 according to your specific motor or you'll risk damaging the motor and/or driver.

[Benta]

Does your BLDC motor have Hall sensors?
If not, forget about the approach. Sensorless BLDCs won't start reliably under load and have very little torque during starting.

Tell us about your motor, that's more telling than your driver design.

[Nikan]

Well all I am doing is using a buck converter + an LDO to get rid of some of the noise from the buck and power with that the µC. I am running the MP6532 as stated in the datasheet. I have just chosen the bootCap etc around that chip for the specs of my mosfets. the PWM signal is runnig at around 28KHz.
The OCP is set up to trigger on around 99A.
So if I understand you correctly I also have to check one of the hall sensors with the µC, trigger on each state change and count the poles with that to measure the RPM and change the PWM accordingly.
But how am I going to start the motor when its in stall position?
Should I go full throttle to spin it up first?
The weird thing is that the Ton/Toff still remains the same on the mosfet gates when I am on 100% dutycycle and try to stop the motor. But the pulse length should be changing accordingly to the resistance on the motor...

[Nikan]

That motor was indeed a sensorless BLDC, but I have taken it apart and installed the sensors my self.
the sensors are spaced exactly 120° from another.
I have checked each sensor against a pair of windings.
They are perfectly matching. There is either no phase shift or 120°/240° shifted accordingly.

[Benta]

That motor was indeed a sensorless BLDC, but I have taken it apart and installed the sensors my self.
the sensors are spaced exactly 120° from another.
I have checked each sensor against a pair of windings.
They are perfectly matching. There is either no phase shift or 120°/240° shifted accordingly.

That may be, but that doesn't mean they are placed correctly. A BLDC needs to be "timed".

What you need to understand is, that the Hall switches control the MOSFET drivers directly, they're a replacement for the mechanical commutator. PWM is just something "on top" for speed control.

[Nikan]

I finially got my hands on a cheap sensored bldc driver and tested the motor.
The motor is working fine, something is wrong with the MP board.
I thought that I might have hooked it up the wrong way to my board.
So I just tried every connection order possible and curiously had to witness, that the motor runs no matter how I hook it up.

It's runing weird, but it runs and is not really stalling. Something is really off 
There is just one order which works somewhat fine, but really not that much better then the rest.

[Nikan]

Finally found the issue.   
I had some spare time last weekend so I poked around my board with my scope trying to figure out what's really going on...
As it turns out I had really bad crosstalk on the signal lanes (PWM, sleep, brake) which obviously explains why the driver is behaving that weird.
Lesson Learned!