I'm having a similar issue as described here, but in my case, the problem does not persist between power cycles of the Cycle Analyst.
It's a bit of a long story, but I'll try to be quick. :-D
The bike is a Yuba Supermarche' that I've electrified with a Phaserunner V6, RH212, Cycle Analyst (CA), and an E-Rider T17 torque sensor. Actually, this is the bike: Project View.
When I first put together the bike, I don't recall there being any issues.
Then, after a few months, I started having issues with the pedal assist. When the bike is already moving, there are no problems. But, when starting from a stop, the pedal assist will frequently cut out as the bike first starts moving. I then have to power cycle the CA for the pedal assist to return. Only the CA needs to be power cycled; the Phaserunner stays powered. I have also tested power cycling just the torque sensor by unplugging it from the CA (without powering down the CA) and this also restores pedal assist.
The problem is correlated with my use of the throttle. I've gotten into the habit of giving a touch of throttle when I first start up just to get everything moving. However, I think there was at least one incident where the pedal assist cut out even when I didn't use the throttle. In any case, when I start from a full stop, I'll apply a little bit of throttle, and about 20% of the time, the pedal assist will spontaneously die just as I do that.
It gets weirder, though. After the pedal assist cuts out (and before I reboot it), the CA will continue to show both torque and cadence (rotation) from the torque sensor! However, the values are... odd? The voltage displayed in the CA from the torque sensor appears to be a little higher in this error state (by about 0.5 V?) than in the normal state, but nevertheless, the little debug fire graph at the bottom of the CA main screen shows overall much less power input from the torque sensor than in the normal state. I'll have to check the cadence reported by the torque sensor next time it gets into this state and report back on whether or not that looks normal.
There is one additional bit of strangeness, which is that, sometimes, as the pedal assist cuts out, it will do so kind of gradually, over about half a second. That is, the pedal assist will at first drop to about 50% of its typical value, and then, as I try to pedal, go down to 0. It's almost as if some kind of illegal value or zero starts to get inserted into the pedal assist averaging algorithm and it ends up driving the average to zero.
Also, sometimes the problem will go away if I use the up/down buttons on the aux controller to adjust the pedal assist multiplier.
So, here's what I've tried to fix this:
1. I've grounded the battery negative terminal to both the front and rear subframes. This didn't change anything.
2. I've put RF chokes on the torque sensors wiring at both the Cycle Analyst and the torque sensor end. No change.
3. I flashed the Cycle Analyst to v3.15 so that all settings were wiped and carefully re-entered the settings by hand using the Cycle Analyst Linux software. No change.
4. I flashed the Cycle Analyst to v3.14 and re-entered the settings by hand using the Linux software. No change.
5. I flashed the Cycle Analyst back to v3.15 and re-entered the settings by hand using the Linux software. No change.
6. I rerouted the motor power cable a bit further away from the ERider signal cable and I routed the ERider signal cable away from the cable between the CA and the Phaserunner. This seems to have decreased the rate at which the issue comes up.
I would love to hear any ideas or suggestions. I can think of three possibilities:
1. EMI in the ERider cable corrupts its firmware and causes it to send bad readings to the CA.
2. The CA's torque sensor input circuity was damaged by ESD at some point and it will just randomly start to give corrupt readings sometimes. Maybe there is charge build-up somewhere where there shouldn't be?
3. Running code in the CA is corrupted by something (EMI?), which causes pedal assist to stop working.
#2 seems the least likely, because I don't see why a CA reboot should fix any kind of charge build-up. However, rebooting the CA also reboots the torque sensor, so it could be that's what's actually fixing the issue.
#3 seems possible, but has mixed support. The fact that disconnecting and then reconnecting the torque sensor will fix the issue does not support this hypothesis, but that adjusting the PAS level will sometimes fix it does support it. (Unless adjusting the PAS level briefly cuts power to the torque sensor, and that's why the fix works.)
#1 seems most likely to me, because power cycling just the torque sensor resolves the issue. I'm still confused about why the CA shows any input at all in the error state, or why adjusting the PAS level resolves the issue, though. I'll check to see what cadence is reported by the torque sensor the next time the problem comes up and report back. I'll also try grounding the Cycle Analyst through the aux power plug, putting RF chokes on the motor power cable, and winding the torque sensor cable through the RF choke one more time.
It's a bit of a long story, but I'll try to be quick. :-D
The bike is a Yuba Supermarche' that I've electrified with a Phaserunner V6, RH212, Cycle Analyst (CA), and an E-Rider T17 torque sensor. Actually, this is the bike: Project View.
When I first put together the bike, I don't recall there being any issues.
Then, after a few months, I started having issues with the pedal assist. When the bike is already moving, there are no problems. But, when starting from a stop, the pedal assist will frequently cut out as the bike first starts moving. I then have to power cycle the CA for the pedal assist to return. Only the CA needs to be power cycled; the Phaserunner stays powered. I have also tested power cycling just the torque sensor by unplugging it from the CA (without powering down the CA) and this also restores pedal assist.
The problem is correlated with my use of the throttle. I've gotten into the habit of giving a touch of throttle when I first start up just to get everything moving. However, I think there was at least one incident where the pedal assist cut out even when I didn't use the throttle. In any case, when I start from a full stop, I'll apply a little bit of throttle, and about 20% of the time, the pedal assist will spontaneously die just as I do that.
It gets weirder, though. After the pedal assist cuts out (and before I reboot it), the CA will continue to show both torque and cadence (rotation) from the torque sensor! However, the values are... odd? The voltage displayed in the CA from the torque sensor appears to be a little higher in this error state (by about 0.5 V?) than in the normal state, but nevertheless, the little debug fire graph at the bottom of the CA main screen shows overall much less power input from the torque sensor than in the normal state. I'll have to check the cadence reported by the torque sensor next time it gets into this state and report back on whether or not that looks normal.
There is one additional bit of strangeness, which is that, sometimes, as the pedal assist cuts out, it will do so kind of gradually, over about half a second. That is, the pedal assist will at first drop to about 50% of its typical value, and then, as I try to pedal, go down to 0. It's almost as if some kind of illegal value or zero starts to get inserted into the pedal assist averaging algorithm and it ends up driving the average to zero.
Also, sometimes the problem will go away if I use the up/down buttons on the aux controller to adjust the pedal assist multiplier.
So, here's what I've tried to fix this:
1. I've grounded the battery negative terminal to both the front and rear subframes. This didn't change anything.
2. I've put RF chokes on the torque sensors wiring at both the Cycle Analyst and the torque sensor end. No change.
3. I flashed the Cycle Analyst to v3.15 so that all settings were wiped and carefully re-entered the settings by hand using the Cycle Analyst Linux software. No change.
4. I flashed the Cycle Analyst to v3.14 and re-entered the settings by hand using the Linux software. No change.
5. I flashed the Cycle Analyst back to v3.15 and re-entered the settings by hand using the Linux software. No change.
6. I rerouted the motor power cable a bit further away from the ERider signal cable and I routed the ERider signal cable away from the cable between the CA and the Phaserunner. This seems to have decreased the rate at which the issue comes up.
I would love to hear any ideas or suggestions. I can think of three possibilities:
1. EMI in the ERider cable corrupts its firmware and causes it to send bad readings to the CA.
2. The CA's torque sensor input circuity was damaged by ESD at some point and it will just randomly start to give corrupt readings sometimes. Maybe there is charge build-up somewhere where there shouldn't be?
3. Running code in the CA is corrupted by something (EMI?), which causes pedal assist to stop working.
#2 seems the least likely, because I don't see why a CA reboot should fix any kind of charge build-up. However, rebooting the CA also reboots the torque sensor, so it could be that's what's actually fixing the issue.
#3 seems possible, but has mixed support. The fact that disconnecting and then reconnecting the torque sensor will fix the issue does not support this hypothesis, but that adjusting the PAS level will sometimes fix it does support it. (Unless adjusting the PAS level briefly cuts power to the torque sensor, and that's why the fix works.)
#1 seems most likely to me, because power cycling just the torque sensor resolves the issue. I'm still confused about why the CA shows any input at all in the error state, or why adjusting the PAS level resolves the issue, though. I'll check to see what cadence is reported by the torque sensor the next time the problem comes up and report back. I'll also try grounding the Cycle Analyst through the aux power plug, putting RF chokes on the motor power cable, and winding the torque sensor cable through the RF choke one more time.
