FOC identification

Electric Motors and Controllers
Post Reply
User avatar
johnrobholmes   1.21 GW

1.21 GW
Posts: 3955
Joined: Jul 07 2008 2:19pm
Location: Missouri
Contact:

FOC identification

Post by johnrobholmes » May 31 2018 3:02pm

I've been having some lively discussions over a new controller in the RC realm that is claiming closed loop FOC. By all accounts, I can't make it fit what I know to be FOC. Mainly, when looking at the voltage from the controller, it is quite trapezoidal in shape. The area under the curve has a linear rise and fall, and at half throttle spends 1/3 of the time at full duty and 1/3 of the time at 0 duty. There is a 14khz AC waveform on top of the typical commutation, in order to lock in the motor speed to the drive frequency. The motor is a nearly perfect sine bEMF, so the controller output should be close to a sine wave with the typical dips in the middle where reluctance is at min/max. So far, I have quite a few engineers in agreement that this is either a poor attempt at FOC, or not FOC at all. They are using an encoder on the motor, but encoder doesn't = FOC.

What say you? Any more data I can gather short of destroying the controller?
Attachments
switching grab.jpg
real life
switching grab.jpg (138.89 KiB) Viewed 1044 times
FOC voltage profile.png
the typical FOC voltage waveform
FOC voltage profile.png (121.91 KiB) Viewed 1044 times

User avatar
johnrobholmes   1.21 GW

1.21 GW
Posts: 3955
Joined: Jul 07 2008 2:19pm
Location: Missouri
Contact:

Re: FOC identification

Post by johnrobholmes » May 31 2018 3:03pm

a zoomed in look at the switching at 50% throttle. Very linear rise and fall to duty cycle, with very large portions spent at 0 and full duty. Curious.
Attachments
grabbed frame 2.jpg
grabbed frame 2.jpg (138.18 KiB) Viewed 1043 times

User avatar
Lebowski   1 GW

1 GW
Posts: 3307
Joined: Jun 28 2011 1:38am
Location: beautiful Zurich, Switzerland

Re: FOC identification

Post by Lebowski » May 31 2018 3:16pm

Its the typical moving midpoint sinusoidal output... this does not mean its foc just that it is sinus out.

The clou is that each if the waves looks kind of trapezoidal, but the DIFFERENCE between two voltages is sinusoidal (and with a max amplitude larger than what fits in the supply, which is the reason for moving midpoint). The motor only cares about the voltage difference between its terminals.

I do the same in my controller IC.

With your scope, between controller out and scope put a 1 kHz RC lowpass filter. Then use the subtract function to see the sine.

User avatar
johnrobholmes   1.21 GW

1.21 GW
Posts: 3955
Joined: Jul 07 2008 2:19pm
Location: Missouri
Contact:

Re: FOC identification

Post by johnrobholmes » May 31 2018 4:04pm

They are running a 14khz AC waveform on all three phases, like a stepper driver. Seems to cause currents on the RC filter, smoked the first set :lol: . The bits I have to build a 1k LPF are 10ohm resistors and 10uf caps, I'll need some different values to tamp down the currents I suppose.

User avatar
johnrobholmes   1.21 GW

1.21 GW
Posts: 3955
Joined: Jul 07 2008 2:19pm
Location: Missouri
Contact:

Re: FOC identification

Post by johnrobholmes » May 31 2018 4:35pm

I differenced two phases without the filter and it’s a square wave. Ground on the negative terminal, probes on the phases.

lizardmech   10 W

10 W
Posts: 67
Joined: May 19 2017 4:16am

Re: FOC identification

Post by lizardmech » May 31 2018 10:09pm

Which controller is it?

User avatar
johnrobholmes   1.21 GW

1.21 GW
Posts: 3955
Joined: Jul 07 2008 2:19pm
Location: Missouri
Contact:

Re: FOC identification

Post by johnrobholmes » May 31 2018 11:09pm

A hobbywing unit. It is only compatible with their motors and offers no tuning of motor parameter, so it's quite limited in this regard.

lizardmech   10 W

10 W
Posts: 67
Joined: May 19 2017 4:16am

Re: FOC identification

Post by lizardmech » Jun 01 2018 3:27am

johnrobholmes wrote:
May 31 2018 11:09pm
A hobbywing unit. It is only compatible with their motors and offers no tuning of motor parameter, so it's quite limited in this regard.
Looking at it, it probably is FOC, it won't run other motors as FOC needs to know the motor resistance, inductance and flux linkage. FOC controllers have to either be matched to motors or include motor detection functions.

User avatar
Lebowski   1 GW

1 GW
Posts: 3307
Joined: Jun 28 2011 1:38am
Location: beautiful Zurich, Switzerland

Re: FOC identification

Post by Lebowski » Jun 01 2018 10:15am

So this is from my controller:
DSC02113-800x800.jpg
raw output signal
DSC02113-800x800.jpg (139.88 KiB) Viewed 963 times
DSC02115-800x800.jpg
low pass filtered (here done with RC, normally done by motor inductance)
DSC02115-800x800.jpg (74.29 KiB) Viewed 963 times
DSC02116-800x800.jpg
blue is the difference between yellow and magenta, a sine wave.
DSC02116-800x800.jpg (77.08 KiB) Viewed 963 times

User avatar
fechter   100 GW

100 GW
Posts: 14384
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: FOC identification

Post by fechter » Jun 01 2018 1:09pm

I think it would be hard to tell if it has true FOC just by looking at the output waveform. My cheap sine wave controller is not FOC, but the output looks very similar and it does field weakening.

Can you tell if the controller has phase current measurement (shunt resistors or hall sensors)?
"One test is worth a thousand opinions"

User avatar
johnrobholmes   1.21 GW

1.21 GW
Posts: 3955
Joined: Jul 07 2008 2:19pm
Location: Missouri
Contact:

Re: FOC identification

Post by johnrobholmes » Jun 01 2018 8:17pm

It is potted so I can't really get into it without destruction. I have asked the company to reference the type and position of current sensors if they would like to give me some information to go on.

Post Reply