Intermittent Map Change to Green

[emonk]

I’m having an intermittent problem with my 2022 Epure Race. A couple of days ago, I started riding in blue mode with tick over on.  Very quickly, it switched to green.  Pushing the selector button again to blue, it again switched back to green when riding a short distance (it stayed in the same TKO mode).  I rode back to the van, took off the covers, wiggling wires and harnesses did not trigger it. What did trigger it was hitting the right side of the motor with the palm of my hand. It was surprisingly easy to trigger it this way. Also, tilting the bike to the extreme left triggered it.

Unfortunately, the problem stopped after about 15 minutes and I haven't been able to trigger it so easily. I rode for over an hour with it happening once or twice.  Once I did get the map button flashing red, but that went away when pulling off the tether.

Suspecting the issue was related to the motor encoder, I disassembled to inspect. The photos show what it looks like.  The JST connector is for a thermistor. I put on a test plug and measured the resistance while warming the engine case. It appears to be working properly (no spikes or drops when warming with a heat gun or hitting the engine case with my hand). It possibly could be the encoder itself.  

I've fixed other intermittent problems such as faulty progressive regen harness and faulty TKO switch.  These other problems caused issues consistent with the component. For example, the power would cut out when I had a faulty regen harness. With a faulty TKO button when hitting a bump, the mode would switch to green and change the TKO mode. I hardwired the TKO mode to be on (a few months ago) since I never ride with TKO off.

For the current problem I'm looking for some pointers. For example, would a flakey encoder cause a sudden switch from blue to green?

The bike was in my van which was about 10 degrees warmer than outside. The problem started immediately when starting to ride, but stopped after about 15 minutes. It wasn't a hot day, but the van was a little warmer. I’m speculating that there may be some correlation with the temperature change.

[konrad]

I just looked at the 2022 switchgear (P/N TL02R-60407-00-00) in case it was different from my 2021.  The 2022 also has a single button to scroll through the maps.

So transitioning directly from Blue to Green is not possible via the normal map-change firmware sequence.  

That makes it seem like a controller error condition triggering the change.

The siliXcon controller can be configured to run "sensorless", and may be doing that as a fallback for a glitchy position encoder.  In which case, it may be changing to the softest map as well.

What happens if you unplug the encoder and attempt to ride the bike?  Do you have access to an oscilloscope?

[emonk]

An error condition triggering the mode change makes sense.

Based on your suggestion of unplugging the encoder, I powered on the bike with the encoder unplugged (and removed from the side of the motor).  The modes were able to cycle through normally, but without the buzz that occurs on mode change. Also, there was no movement from the motor due to tickover being on, or throttle input.

 I've ordered some connectors to help bench test the encoder. I wanted to see if I could catch an erratic change in the sine or cos outputs while jigging things (only using a multimeter).  An oscilloscope would be better. I've been considering getting one for a while now, and this problem may prompt me to pull the trigger.  Ideally I'd like to collect the signals while riding and see if the data shows a glitch when the mode changes unexpectedly.

[emonk]

@konradI ended up getting an oscilloscope since the multimeter wasn't helping me.  I found out that I had to connect the encoder to the controller and tap into the sin/cos/ground wires while the bike was powered up.  Bench testing the controller off the bike with a power supply didn't work, apparently the sin/cos lines need some sort of termination circuit. 

But with the encoder hooked up in the regular way, turning the wheel by hand gives reasonable sin/cos curves.  I didn't see anything odd taping on the encoder while gathering data.  Attached is a screenshot of the capture.  I'm going to think about setting up gathering data while riding. But any input would be appreciated.

[konrad]

Congratulations on getting a 'scope.  That's what the waveforms should look like. 

The siliXcon controller input circuit shows a pull-up resistor to +5VDC: https://docs.silixcon.com/docs/hw/esc/ax1/IO/#sin-cos-sensor-connection

I don't think general-purpose data logging would be practical.  Notice that your 'scope is sampling at 100kS/s.  You could sample slower than that and still make out waveshapes, but the sample rate needs to scale proportional to rpm.  Riding speed and duration will result in a massive amount of data.  Even if you could store it, you would have to sort through it visually.

Try heating the sensor with something like a hair dryer while testing.  But honestly, I doubt that will reveal anything.  

I think the most likely problem is an intermittent connection somewhere.  But I can't think of a failure mode that would produce your symptoms.  But EM keeps changing the firmware, and all I really know is my 2021.

P.S. Try running the bike on a stand and monitor the sin/cos signals at normal running (and varying) RPMs.   When I do things like that at high speed, I like to strap the bike to my trailer so it can't get away!

P.P.S. Another thing I have done to promote safer stationary testing is to remove the front sprocket. 

Edited 8 hours ago by konrad

[emonk]

6 hours ago, konrad said:

Congratulations on getting a 'scope.  That's what the waveforms should look like. 

The siliXcon controller input circuit shows a pull-up resistor to +5VDC: https://docs.silixcon.com/docs/hw/esc/ax1/IO/#sin-cos-sensor-connection

I don't think general-purpose data logging would be practical.  Notice that your 'scope is sampling at 100kS/s.  You could sample slower than that and still make out waveshapes, but the sample rate needs to scale proportional to rpm.  Riding speed and duration will result in a massive amount of data.  Even if you could store it, you would have to sort through it visually.

Try heating the sensor with something like a hair dryer while testing.  But honestly, I doubt that will reveal anything.  

I think the most likely problem is an intermittent connection somewhere.  But I can't think of a failure mode that would produce your symptoms.  But EM keeps changing the firmware, and all I really know is my 2021.

P.S. Try running the bike on a stand and monitor the sin/cos signals at normal running (and varying) RPMs.   When I do things like that at high speed, I like to strap the bike to my trailer so it can't get away!

P.P.S. Another thing I have done to promote safer stationary testing is to remove the front sprocket. 

 

Thanks, that circuit link is very helpful, there is a lot of other good information in those pages as well.  Also, your https://www.electricmotiontech.com/ has been a valuable resource.

The logging would indeed generate a lot of data.  I could try to pare it down with triggers. I could also set up a GPIO button on one of the scope's inputs so that I could manually signal when something odd occurs.

The best is to catch it happening while not riding.  Great suggestion of taking off the front sprocket. 

Yes, I keep thinking there is a connection or wire break somewhere, but the tapping on the right side of the motor seemed to indicate the encoder.  But it could be something else.