Hello there

I am new to this forum, so if this thread is posted incorrectly please say so.

Currently i am running a project where we are supposted to control a BLDC motor (to run a skateboard).
Anyway. Our bridge and control is running (supposedly) correct, since we can run the motor in open loop.

Now to control the commutation we are looking at our BEMF signal, which unfortunately is looking incorrect.
I do not have any chance to upload the actual oscilloscope pictures, but I can draw it quite decently.
The picture below shows how each of the A, B or C BEMF looks like. This signal is Not very easy to calculate a zero-cross from.

I can get Scope-pictures tomorrow of everything you might want, to be able to help me.

Hopefully you have an idea of what could be wrong.

 

I assume you've built a circuit to generate this signal, and you want help in making it work. We cannot guess the architecture of your circuit without a schematic, thus we would not be able to help debug it.

 

Sry Bout' that, i thought that maybe someone would have encountered the same problem. Our design is here:
I can see that the N-Channelmosfet didnt catch my pen right, so here are the components:
P-Channel MOSFET: FQP27P06
N-Channel MOSFET: IRLZ44N
Gatedriver: MCP14E7
Nand: 74hc00

 

Appears to be connected/controlled as a 3ph motor not BLDC?
Max.

 

The Brushless DC motor is a 3 phase motor indeed
The problem is getting the correct back-emf signal to be able to control the motor properly. For some reason our bemf-signal is quite different to the exspected signal.

 

But there is a subtle difference, to be precise a 3ph motor has three wingdings simultaneously powered, a BLDC or brushless DC is so named because only two windings are powered at any one time, simulating a DC brushed motor turned inside out.
Both are constructed basically the same.
Max.

 

Ok, well this is indeed a BLDC, and is as you say powered on 2 phases while the last phase is floating (we look at the BEMF on this phase ofcourse). The program that PSoC 4 is running is turning on/off the Fets in this order: (The hall signal is supposed to be constructed through the BEMF if just the signal was as expected)

 

Here is how I align the Halls (or encoder equivalent tracks).
Back feed the motor around 200rpm to read the sine wave generated on a double beam 'scope.
Max.

 

Concerning the drawings of you scope traces in post #1. Why would you think the top trace (bemf) should be like the bottom one, which looks like the trapezoidal wave form that drives the motor phases? After being exposed to the inductance and capacitance of the motor coils, do you really think the 'scope traces will be the same?

 

We have been looking on some different application notes to get some ideas of how to meassure where to commutate. From bemf it should be done from detecting the zero-crossing. In all the notes i have been looking on so far, the bemf looked likehttp://ww1.microchip.com/downloads/en/appnotes/00857a.pdf, page 14.
I have attached the Bemf from the application note.

For some reason ours dont look like it. It might just be us not understanding the curcuits i dont know :/

 

Hi All, I didn't notice "snubbers" around the High/Low switching FETS? The BEMF must have a current path (reversing) for the off-Switching times for built up EMF to drain before the next energized cycle? Reverse-biased silicon diodes across Drain-Source.......Google 3-phase high/low switching design....just a thought ;o

 

Hi! well i thought the Snubbers inside the Mosfets would be enough, but i'm not surtain, i will try this.
I wonder if the problem could be controlling the Motor?

Q1: Am i right to say the dutycycle of the pwm should control the the speed of the motor? So lets say we
increase the dutycycle the electrical fields would force the magnet to rotate faster right? This would mean,
that a zerocross is detected sooner and commutation is done as calculated?. Would this be the general way to control?

We defined the period of the commutation with 100% duty cycle. Would this force the motor to commutate too late?
If commutated too late is the signal in post #1 what would be expected?

Q2: How do you start up the brushless motor best? With BEMF you would have to run in an open loop for a little right?

 

just to point it out incase you used it to build your circuit.

The timing waveforms do not match the hall devices placement in the drawing you posted. specifically H2 and H3 are reversed physically OR by label in the waveform. timing of these two do not match drawing with dir of rotation given.

the trapazoidal waveform you refer to is a current wave and your scope is showing a voltage trace. I think that is right, based on my faulty re-memories of learnings I have had.

 

Hi! well i thought the Snubbers inside the Mosfets would be enough, but i'm not surtain, i will try this.
I wonder if the problem could be controlling the Motor?

Q1: Am i right to say the dutycycle of the pwm should control the the speed of the motor? So lets say we
increase the dutycycle the electrical fields would force the magnet to rotate faster right? This would mean,
that a zerocross is detected sooner and commutation is done as calculated?. Would this be the general way to control?

We defined the period of the commutation with 100% duty cycle. Would this force the motor to commutate too late?
If commutated too late is the signal in post #1 what would be expected?

Q2: How do you start up the brushless motor best? With BEMF you would have to run in an open loop for a little right?

Hi There, the BEMF will equal the drive current thru the phase, and if I remember brushless DC correctly, the rotor shaft has diagonally placed magnets for the start rotation situations, and I think driving "A" first, then C then B, but don't quote me on that, you folks are quite aware of motor workings.... As for speed commands, the brushless DC should react similar to the 3-phase AC drive, where "frequency" variations control the speed commands.....