pedal sensor error when cable tied to frame

[kylerlaird]

This problem has been vexing me for over a month. I think I finally narrowed it down. I need to talk through it...

I'm now building a 26" fat tire Mongoose. I a similar 20" Mongoose. It's a VESC controller, Sempu T9 pedal sensor, 48V battery, and eBikeling front hub.

The Sempu sends about 1.5V for torque at rest. I tweak the controls, everything is good, then I take it out for a test and it's buggy.

After much testing on the stand, I finally realized that the problem comes when I zip tie the Sempu cable to the frame (right before I take it out for a ride). The torque sensor output will go low (~0V) sometimes. I assumed it was a bad connector. I switched cables to the controller a few times. I went through three other T9s. The problem remains.

It doesn't feel like a typical bad connection. It's much more subtle. There's a short cable with a connector coming from the T9. It mates to another cable that goes up to the controller. If I press those connectors up against the frame and roll it just right, the torque voltage takes a dive. The cadence sensor still works, though, so I don't think it's a power cable issue.

This sensor works with "wireless" (inductive) power. I've been thinking that it could be an interference problem. Tonight I tried replicating the motions that were causing failure against other pieces of metal (like the bike stand). No effect. It's only when I hold the connectors against the frame.

But why the frame??? Ah ha! The motor! I finally disconnected the motor and the problem ceased. I'm guessing that motor noise is disrupting the torque sensor's wireless power when the cable is in some positions.

This analog/RF stuff is not my forte. Do I need to provide a better ground between the frame and fork? Throw a choke on the cable? Shield the cables? Offer a sacrifice to Telsa?

 

[amberwolf]

If there is interference from a hubmotor actually in the frame of the bike far from the motor itself, it might mean teh motor has a phase shorted to the stator, which then connects electrically to the axle and then the frame, so it's sending electrical garbage signals out like a big antenna.

IF you disconnect the hubmotor completely from the controller, then use your multimeter set on ohms, find a shiny metal spot on the fork or axle where you can hold the black meter lead against and then use the red one to touch each of the motor cable's phase wire pins. This should read OL on every one of them--if it doesn't, there's a slight connection somewhere. If it does read OL on all, then set the meter to continuity if it has this setting, and try again. Also retry on Diode test (looks like an arrow with a bar on the pointy end), and on the 2Ohms setting. (in each mode the meter behaves differently, so you may get a result in some modes but not others).

If it is OL in all modes, then it's probably just induced voltage from the phase wire cable from the motor to the controller into the bike frame. Wrap that in a layer of aluminum (or copper, but most people don't have that laying around) foil and see if it the problem goes away. (you can also wrap the Sempu cable in foil to test that too, but since the Sempu is *in* the frame it might not fix it, whereas shielding the phase wires from the frame might). Or even just make sure the phase wire cable is held away from the entire frame as far away as it will go, for the test.

 

[kylerlaird]

I checked conductivity between each motor leg and the fork. I didn't get any flow.

Then I checked for AC potential between the fork and the controller's ground. It was over 1.5V whenever a leg of the motor was connected to the controller. (I did get measurable potential between the fork and frame.) I'm planning to rig a ground strap from the controller to the fork. I might also throw a scope on there to see how that signal looks.

I suspected that the Sempu doesn't ground the case (into the frame) but I tested it. Confirmed.

For this test, I used a dedicated 9V battery to run the Sempu. I was trying to eliminate any problems due to the 5V->9V booster I usually use. It has not seemed to have any effect so far.

I also ordered some ferrite beads in case this gets more complicated.

Oh! And I took some foil out but didn't try it on the motor cable. I will...

 

[verdulleme]

Last february I installed a sempu T2 in my recumbent trike. Its running with a Nucular controller and MAC puma motor.
Problem was that the moment the motor started the sempu stopped. Since the controller was set to torque and not PAS this would immediately stop the motor. Result, 5 to 20 short 'hickups' from the motor before it kept going. But every time I stopped pedalling it would hickup again several times. Highly annoying.
Then after a lot of experimenting found that connecting the frame to the battery minus solved the issue. Seems like same as the OP, too much interference on the frame for the sempu to work properly.
Me happy!

But wait: thanks to a tip from Vasiliy (Nucular) I now have 2 safety capacitors between the frame and battery minus, not a hard wired connection. This way even if the frame is by accident (by motor or controller fault) connected to the power supply there won't be a massive short resulting in a sparkling light show from your battery.
These safety capacitors (X or Y capacitor) can be found in almost any scrap circuit board which has a grid input, in my case 230Vac. They are either connected between the grid lines or between grid and ground. Any work, the bigger the better.
So now me very happy.

One thing I haven't tried yet is putting the cable back through the inside of the frame, right now it is still tied on the outside.