Ugly Motor waveforms

I built my first E-bike last week and it seems to run okay but I just poked an oscilloscope probe on the motor phases and Hall outputs and things do not look well. The image is of a scope trace with the one Hall output on top and one phase on the bottom trace. There are spikes of significant width that descend to zero volts. All three phases and all three Halls have the same pattern on the scope.

The spikes look the same at part throttle. That is they are in the same place and have the same amplitude as at full throttle waveforms more or less. I filtered the Hall signals with small capacitors thinking the Halls may be causing the phase spikes and this cleaned up the Hall signals but this did not help the spikes on the phases. The Controller is A Golden motor magic controller and a Golden motor 250 Watt, 36Volt geared hub motor. Battery is a 40 VDC Headway pack. All wiring is very short in length. As I am new as this I am not sure what to expect from the waveforms but this does not seem right to me. The scope was connected from the battery Negative power lead at the controller to the Hall and Phase controller pins. Any ideas?


Ken

I want to plug in a scope to check out my stuff too! How much was your usb scope?

As for ideas, could the glitches in the phase voltage be the pulse width modulation of throttle control?
Some PWM never actually goes 100%. I don't know enough about the BLDC controllers though.

TJ

Looks like the spikes are at 60deg spacing which is when one of the other phases is switching.

Actually, the waveform looks pretty clean compared to others I've seen. The spikes are from commutation, but those are pretty much normal, at least the ones at the ends of the flat sections. The ones in the middle do look out of place and may indicate either the main caps in the controller are wimpy or the scope was referenced to a noisy spot.

Here's a BMC motor:

Those look like clock pulses from your timing source hard to judge accurately from that screen but 'm guessing 13000 hz or so. But the spikes are so short and sharp they almost can't be present in the motor. The duration appears faster than anything in the controller could switch.

I would suspect it's noise in the probe, caused by the controller RFI.

I have two scopes, one is an older analog Tectronix, the other an DSO-2090 USB scope adapter that I connected to my Laptop and which produced the traces shown in the image. Both scopes have identical traces. The spikes are significant in width with respect to the width of the commutation cycle. The large spike that occurs at the top of the trace must be reducing the power to the motor by at least a percent or more. I have tried adding additional capacitance on the battery leads where they enter the controller without any improvement at all. The spikes are consistent at all speeds. As stated cleaning up the Hall signals with filter caps did not reduce the phase spikes. It may be that the PWM modulator program is doing this do to a program bug.

The DSO-2090 was bought on E bay for under $200 and works quite well. I am experienced in using scopes over several decades. Storage scopes used to cost half a years salary at one time. Be careful though as this USB scope can not handle high voltage. The probes that it comes with can be set to X10 which is necessary when working on motor phases where spikes can exceed the input voltage spec of the scope.

I will have to get deeper into the controller. This is a very small and light controller which is why I went with it as I want the lightest overall E bike possible.

cool picture looking at the yellow trace in the image you can see that the timing is off same goes for the picture in fechter's post

you are retarded while fechter is advanced (the timing I mean )

especially fechter's picture is nice, you can see the discharge of the motor inductors (the sharp spike
before the down/opward slope) and the conduction through the FET protection diodes (just before the middle
for instance, after the sharp downward spike the voltage is a tad higher than while the FET is on, this is the 0.7V
forward voltage of the diode)

Here's Samson's pic shrunk down so you don't have to open it separately:

Clearly the spikes are happening at the commutation points. It may be something in the software that's killing the high side output during every commutation step. You could try probing the FET gates.

Lebowski said:

cool picture looking at the yellow trace in the image you can see that the timing is off same goes for the picture in fechter's post

you are retarded while fechter is advanced (the timing I mean )

Click to expand...

Interesting observation. Hmmm... seems like that may be a way to automatically correct the timing. If a circuit can detect the timing error, it could be corrected automatically. A little advance is good at higher motor speeds.

The low side fet gate signals are clean, the high side shows the spikes. All three phases look the same on the scope.

Can anyone direct me to a schematic of this controller or failing that a controller schematic that shows a gate drive circuit so I can try to filter the spikes. I suspect is more of a software thing but would like to try filtering the gate drive some while I have it apart.

you're measuring the high side FET'ses gate drive with respect to the ground, which explains your very
messy signal in the bottom plot. The sources of the high side FETs are connected to the motor phases, so
on the gates you see the sum of the motor phase and gate-source signal. As far as turning the high side FET's on
your only interested in the gate-to-source voltage, not the gate-to-ground. Try to measure the high side
FET's gate to source voltage, they'll look much better

I was able to find a controller schematic for a different model and see how they do the gate drive. Now that I have a better understanding of how it works I probed the high side drive. This image is of the High side gate drive wrt to the motor buss. It is clear the spikes are coming in the gate drive. I tried filtering them with .04 uf but no improvement.

In addition I observed that the phase middle spike starts out 40 usec wide and drops to 20 usec half way down This represents significant width and a small loss of power.