Noisy halls on sinus controller

Hi. I have a Golden Motor HPM05KL (5kW BLDC) and a Kelly KLS7250D Controller that I have connected together. I seemed to work nicely except some slight clicking noise at high speed that I didn't think too much about. But when I applied load to the motor it started to make very hard knocks which sounds like hall effect sensor problems. So I took the motor and controller out of the boat and on to the bench to do some troubleshooting. I started by putting a scope on the hall effects to try to capture any irregularities, but what I found surprised me. There seems to be an abundance of high frequency noise present.
It looks like the noise is 50kHz. Since the controller is Sinusoidal Wave I'm guessing that the noise I see is the PWM frequency or similar.
What I'm wondering now is if this nose might be the source of my problem or if I'm just measuring at too high impedance or something silly like that.

I think you should look at the output of the motor phases. Kelly doesn't play well with blocky sine output motors. I had some serious issues with the Denzel motors and had to go with a sensorless squarewave controller because the incorrectly timed halls caused the motor to overheat within minutes.

The PWM is 10kHz (measured on the output) and I can see some voltage drops right when my clicks appear, but there is no change in the pulse width. I'm thinking the the 50kHz might be from the power supply I'm using to test the motor (a battery charger not meant for this usage).

Anyway, I contacted Kelly Controllers and got a new firmware version. With this new firmware the clicking disappeared. I'm hesitant to call it fixed yet since I haven't tried it under load, but it looks promising. If it isn't fixed, I'll try the scope again with a battery supplying the power rather than a switching power supply.

I do not think that the power supply could be the problem, there is a lot of capacitors at the controller entrance that will probably flatten all/most of it.

I think that could be that the hall sensors are not properly mapped, so I suggest that you run a phase identification angle test throw the controller. If you have done it before do it again using a higher current for it.

As well you can test with higher PWM frecuency in order to reduce noise, but noise produced by low PWM is usually more like a "huummmm" than clicking, so not sure about this make any effect.