Cycle Analyst & Erider T17 torque sensor assist stopped working

[TTSProd]

Hi,

I have a Cycle Analyst with an Erider T-17 torque sensor and a Phaserunner controller. System was working great until a week ago when the Cycle Analyst stopped responding to the torque sensor input. I can see on the diagnostic screen on the CA that the torque sensor itself is still working with proper cadence and voltage that increases with torque pressure but the CA is no longer sending a corresponding throttle signal to the controller. The throttle, which is also connected to the CA still works fine. I did not make any CA configuration changes prior to this. It worked one day, then wasn’t working the next. If anyone has an idea what might be going on here I would really appreciate your insights.

 

[amberwolf]

What throttle mode is it in? IIRC Bypass mode disables all the PAS/etc but feeds the throttle directly out--but it doesn't prevent the other inputs from correctly displaying on the diag screen.

 

[TTSProd]

What throttle mode is it in? IIRC Bypass mode disables all the PAS/etc but feeds the throttle directly out--but it doesn't prevent the other inputs from correctly displaying on the diag screen.

Thanks for the reply, I just went and checked and it’s in Pass-Thru Throttle Mode. Does it need to be in current or power mode for the torque sensor to work? I could be wrong but I believe I was running in Pass-Through mode before it stopped working.

 

[amberwolf]

No, pass-thru should be fine; that's the default mode, where the CA's mapping and other limiting affects the throttle but still lets you use it, and also creates a throttle from the PAS / etc if those are connected and setup. (see below)

Cycle Analyst V3 - Grin Products - Product Info

The Cycle Analyst display is our first flagship product released in 2006 and now in its 3rd generation. While the hardware design may be showing its age, it still has capabilities and functionalities that are unmatched by modern ebike systems with their pretty colour screens. This device also...

ebikes.ca

Throttle Input Setup Menu​

This section configures how the Cycle Analyst responds to the input throttle signal of your ebike. The preview line provides a live readout of the currently measured voltage on the pin, as well as an indication of how this is scaled into either a pass-thru voltage, or a current or a watts limit. A typical hall effect ebike throttle will sit at about 0.9V when off and increase to ~4.1V when twisted all the way. When not thottle is plugged in the voltage sits at 0.0V


Click for Explanation of Throttle Input Settings

These settings refer to the operator throttle plugged into the black 3-pin connector of the Cycle Analyst. They determine how the throttle input signal is scaled over a min to max voltage range to obtain a 0-100% throttle interpretation over the entire operator throttle rotation. The default values (min 1.0V, max 4.0V) are broadly suitable for hall effect throttles, though you can tweak them to vary the deadbands at either end or to accommodate potentiometer and other throttle types.
[ ThrI->Cntrl Mode ]*
Chooser determining how the user throttle affects controller operation

• Pass-Thru: User throttle is linearly mapped from the input throttle range to the output throttle range, with throttle rate limiting applied if necessary. Throttle behavior is similar to having the throttle directly connected to the motor controller. [Default]
• Current (A): User throttle controls the battery current from zero up to the value in MaxCurrent. This mode provides for better modulation of motor power over the entire speed range. When using this mode, MaxCurrent should be set equal to or lower than the controller current limit.
• Speed (kph): User throttle directly controls the speed of the bike from 0 to the value in MaxSpeed. This mode can be useful for special projects but is not a recommended for most ebike scenarios as small throttle motions can result in large pulses of power as the bike attempts to track the changes.
• Power (W): User throttle controls the power to the motor controller from 0 to the value in MaxPower. The effect is very similar to Current Throttle mode, but provides a response that is insensitive to battery voltage sag. When using this mode MaxPower should be set equal to or lower than the battery voltage times the controller current limit.
• Disabled (ZERO): This mode ignores the input throttle and sets the output throttle signal low. The output can then only go high if you have one of the PAS modes enabled and are pedaling. You can achieve the same thing by simply unplugging the operator throttle, but this option can be useful if you want swap between different presets with and without active throttle modes.
• Disabled (WOT): This mode ignores the input throttle and sets the output throttle signal to MaxThrottle unless one of the limits is exceeded. This mode provides legacy support when using a V3 CA on an old controller designed for the original CA V2 connector standard. Controllers with this old standard can only limit an external throttle signal from the CA plug, they cannot use the CA plug as the primary throttle input.
  
  IMPORTANT: Do not select this mode if the CA V3 output is connected to a CA3 compatible controller since the CA will immediately apply full throttle on power-up causing the bike to run away.
• Bypass: This mode is the equivalent of connecting the throttle directly to the controller to completely eliminate the CA from motor/controller behavior. This differs from PassThru mode in that ThrIN is sent directly to ThrOUT with no processing whatsoever: no min/max range adjustments, no ramping, no limiting, and no autocruise or PAS support.
  
  IMPORTANT: This is only a diagnostic mode and is not intended for normal ebike use. All normal CA protections are inoperative (e.g. low voltage, max current).

 

[amberwolf]

Do you use presets, or have any programmed in? If so, you might make sure you're set to the right preset that uses the PAS. Holding the left CA button and pressing the right button should switch to the next preset, and you can cycle htru all three this way, the screen will show you whihc one you're in.



If it's not a preset thing, then the next thing to check is the PAS device setup menu. I don't recall if you'll still see torque sensor/etc readouts on the diag screen if the Sensr Type is seto to Disabled, but if so, that setting would turn off your PAS completely.

Pedal Sensor Device Setup Menu​

The hardware configuration of both basic Cadence PAS sensors and torque sensors is done in the PAS Device setup menu. Here the preview lines shows critical information about the state of the pedal sensor. The P and D arrows show the live state of the pedal sensor input signals, and one or both of these should toggle as the pedals are rotated for the CA3 to detect the rider's pedal cadence. This preview also shows a live voltage reading of the torque signal. Most torque sensors sit at either 1.5 or 2.5V with no force on the cranks and then increase or decrease in voltage as pedal force is applied. Understanding this behavior is critical to checking that a torque sensor is wired up and configured properly.


Click for Explanation of the PAS Device Settings

These settings tell the Cycle Analyst the physical characteristics of the installed pedal assist device. The CA needs to know the number of pulses per crank revolution and the forward/reverse pedal sense. If there is a torque sensor, it must also know how the torque signal is scaled into meaningful units of Newton-meters. These are one-time settings made at installation time; tuning the way the PAS device operates is achieved by the PAS Configuration Settings in the next section.
[ PASD->Sensr Type ]
Chooser to select the type of PAS sensor installed.

• Disabled: There is no PAS device installed. This deactivates all PAS functionality and makes SLim->MxNoPdlSpd inoperative.
• Basic PAS: The sensor is a simple cadence sensor, which provides pulses when the cranks are pedaled. Both 1 wire (only pulses with fwd pedaling) and 2 wire (quadrature signals) are supported.
• Row Bike: The sensor is a simple cadence wheel for a rowing bike, with signals to distinguish forward and reverse motion. Rather than computing the RPM based on rotation speed, the CA instead displays the strokes per minute based on how often there is a direction reversal of the cadence wheel.
• Custm TRQ: The sensor is either a custom device type with unique settings or one of the preconfigured types below.

• Thun BB:
• TDCM BB:
• NCTE BB:
• Sempu BB:
• ERider BB:
• CyclStokr:
  These are sensors for known device types. These selections cause preconfigured defaults to be copied into place for other settings in this category. The sensor type remains as selected if those other defaults remain unmodified, however, changing a default setting will change this sensor type to 'Custm TRQ'. Downloading one of these preconfigured sensor types from the Setup Utility will similarly set other settings to the associated preconfigured defaults exactly as if the preconfigured type had just been selected by CA Console Setup, overriding the other settings that may have been present in the Setup Utility.

Note: The Sempu BB is manufactured with two different interface styles. The version selectable here has the '2-wire' interface distributed by Grin. Units from other vendors may require subsequently customizing PASD->SignlType to '1-wire' in CA Console Setup. The ERider sensor was originally supplied with an 18 pole speed signal, and was updated in 2021 to 36 poles.
[ PASD->PAS Poles ]
Number of pulses generated in one full rotation of the pedal sensor. This is equal to the number of magnets on simple magnetic ring cadence wheels. For internal sensors it can be measured by counting the number of times the P arrow on the SETUP PAS DEVICE preview screen flips up and down during one pedal rotation. This setting is populated automatically when a known sensor type is selected, and must be set correctly for accurate display of human pedal RPM.
[ PASD->Signl Type ]
Both Basic PAS and Torque sensors generate a cadence signal pulse as the cranks are rotated. The signal may pulse only on forward pedal rotation with just a single wire, or it may also indicate direction of rotation which requires two wires.
This setting determines the number of input wires that carry cadence pulses so the CA can best use the available information. The type can be determined by examining the arrows next to the 'PD' on the SETUP PAS DEVICE preview screen as the crank is slowly turned. On a single wire sensor only the P arrow will changes, while on a 2 wire sensor both the P and D arrows will flip UP/DOWN.

• 1 Wire: Cadence pulses appear on the RPM input with no simultaneous change to the DIR input.
• 2 Wire: Quadrature encoded cadence pulses appear on both RPM and DIR inputs.

[ PASD->Dir Plrty ]
Controls whether 5V on the Dir pin is considered forward or reverse pedaling. If the Dir pin is not connected, then it should be set to 5V = Fwd. Trial and error tests may be required to find the proper FWD or REV setting if the device is quadrature encoded. This setting is populated automatically when a known sensor type is selected.
[ PASD->Trq Scale ]
Sets the scaling factor for converting torque sensor output voltage to newton meters. For devices that sense torque on only one side of the crank, the value should be doubled to simulate the net left and right pedal torques. The value can be set either positive or negative and is populated automatically when a known sensor type is selected. For TDCM Sensors, additional tuning is required and may be initially guesstimated as the number of teeth on the front chainring. So a 44T chainring would be about 44 Nm/V. This option is present only for torque sensor types.
[ PASD->Zero Torq ]
The configured zero-torque offset voltage and the present live torque voltage are displayed. The right button should be pressed with the pedals unloaded (zero torque) which configures a new zero-torque voltage from the present live torque voltage. This voltage is displayed when the button is released. Note that magnetostrictive torque sensors (like THUN and NCTE) don't return to the same zero point very well after high torque excursions. This option is present only for torque sensor types.

 

[TTSProd]

Do you use presets, or have any programmed in? If so, you might make sure you're set to the right preset that uses the PAS. Holding the left CA button and pressing the right button should switch to the next preset, and you can cycle htru all three this way, the screen will show you whihc one you're in.



If it's not a preset thing, then the next thing to check is the PAS device setup menu. I don't recall if you'll still see torque sensor/etc readouts on the diag screen if the Sensr Type is seto to Disabled, but if so, that setting would turn off your PAS completely.

Pedal Sensor Device Setup Menu​

The hardware configuration of both basic Cadence PAS sensors and torque sensors is done in the PAS Device setup menu. Here the preview lines shows critical information about the state of the pedal sensor. The P and D arrows show the live state of the pedal sensor input signals, and one or both of these should toggle as the pedals are rotated for the CA3 to detect the rider's pedal cadence. This preview also shows a live voltage reading of the torque signal. Most torque sensors sit at either 1.5 or 2.5V with no force on the cranks and then increase or decrease in voltage as pedal force is applied. Understanding this behavior is critical to checking that a torque sensor is wired up and configured properly.


Click for Explanation of the PAS Device Settings

These settings tell the Cycle Analyst the physical characteristics of the installed pedal assist device. The CA needs to know the number of pulses per crank revolution and the forward/reverse pedal sense. If there is a torque sensor, it must also know how the torque signal is scaled into meaningful units of Newton-meters. These are one-time settings made at installation time; tuning the way the PAS device operates is achieved by the PAS Configuration Settings in the next section.
[ PASD->Sensr Type ]
Chooser to select the type of PAS sensor installed.

• Disabled: There is no PAS device installed. This deactivates all PAS functionality and makes SLim->MxNoPdlSpd inoperative.
• Basic PAS: The sensor is a simple cadence sensor, which provides pulses when the cranks are pedaled. Both 1 wire (only pulses with fwd pedaling) and 2 wire (quadrature signals) are supported.
• Row Bike: The sensor is a simple cadence wheel for a rowing bike, with signals to distinguish forward and reverse motion. Rather than computing the RPM based on rotation speed, the CA instead displays the strokes per minute based on how often there is a direction reversal of the cadence wheel.
• Custm TRQ: The sensor is either a custom device type with unique settings or one of the preconfigured types below.

• Thun BB:
• TDCM BB:
• NCTE BB:
• Sempu BB:
• ERider BB:
• CyclStokr:
  These are sensors for known device types. These selections cause preconfigured defaults to be copied into place for other settings in this category. The sensor type remains as selected if those other defaults remain unmodified, however, changing a default setting will change this sensor type to 'Custm TRQ'. Downloading one of these preconfigured sensor types from the Setup Utility will similarly set other settings to the associated preconfigured defaults exactly as if the preconfigured type had just been selected by CA Console Setup, overriding the other settings that may have been present in the Setup Utility.

Note: The Sempu BB is manufactured with two different interface styles. The version selectable here has the '2-wire' interface distributed by Grin. Units from other vendors may require subsequently customizing PASD->SignlType to '1-wire' in CA Console Setup. The ERider sensor was originally supplied with an 18 pole speed signal, and was updated in 2021 to 36 poles.
[ PASD->PAS Poles ]
Number of pulses generated in one full rotation of the pedal sensor. This is equal to the number of magnets on simple magnetic ring cadence wheels. For internal sensors it can be measured by counting the number of times the P arrow on the SETUP PAS DEVICE preview screen flips up and down during one pedal rotation. This setting is populated automatically when a known sensor type is selected, and must be set correctly for accurate display of human pedal RPM.
[ PASD->Signl Type ]
Both Basic PAS and Torque sensors generate a cadence signal pulse as the cranks are rotated. The signal may pulse only on forward pedal rotation with just a single wire, or it may also indicate direction of rotation which requires two wires.
This setting determines the number of input wires that carry cadence pulses so the CA can best use the available information. The type can be determined by examining the arrows next to the 'PD' on the SETUP PAS DEVICE preview screen as the crank is slowly turned. On a single wire sensor only the P arrow will changes, while on a 2 wire sensor both the P and D arrows will flip UP/DOWN.

• 1 Wire: Cadence pulses appear on the RPM input with no simultaneous change to the DIR input.
• 2 Wire: Quadrature encoded cadence pulses appear on both RPM and DIR inputs.

[ PASD->Dir Plrty ]
Controls whether 5V on the Dir pin is considered forward or reverse pedaling. If the Dir pin is not connected, then it should be set to 5V = Fwd. Trial and error tests may be required to find the proper FWD or REV setting if the device is quadrature encoded. This setting is populated automatically when a known sensor type is selected.
[ PASD->Trq Scale ]
Sets the scaling factor for converting torque sensor output voltage to newton meters. For devices that sense torque on only one side of the crank, the value should be doubled to simulate the net left and right pedal torques. The value can be set either positive or negative and is populated automatically when a known sensor type is selected. For TDCM Sensors, additional tuning is required and may be initially guesstimated as the number of teeth on the front chainring. So a 44T chainring would be about 44 Nm/V. This option is present only for torque sensor types.
[ PASD->Zero Torq ]
The configured zero-torque offset voltage and the present live torque voltage are displayed. The right button should be pressed with the pedals unloaded (zero torque) which configures a new zero-torque voltage from the present live torque voltage. This voltage is displayed when the button is released. Note that magnetostrictive torque sensors (like THUN and NCTE) don't return to the same zero point very well after high torque excursions. This option is present only for torque sensor types.

I don’t use presets so I have none programmed in.

I used the Erider (not Erider T-9) preset to configure the torque sensor. I noticed on the product page it is listed at a 2 wire quadrature, but the preset selects 1 wire and with that setting I can see the P and D arrows toggling as I rotate the pedals and torque voltage when riding. I just recalled this preset again a few minutes ago and still no power is output from the torque sensor signal. I’m going to connect the CA to my laptop when I get home and recall the settings I was using before it stopped working. I find the overview of the software easier to see than the on screen CA interface. Thanks so much for trying to help!!!

 

[amberwolf]

Hopefully that corrects the problem. If not, there might be other ways to fix it using the same method (overwriting settings).

Sometimes it's just that a setting or even a piece of code has gotten corrupted....but somehow still displays correctly onscreen. When reading the settings using the setup program, it may also still show them all correct. Re-uploading the same settings may also not work. Yet...resetting the entire CA to factory (except for the hardware-specific amplifier calibrations) and then re-uploading the settings to it can make it work.

The same thing happens to many electronic devices; I expect it has to do with particle radiation, mostly that from energetic cosmic rays and other charged particles that hit the atmosphere and send showers of further particles down; sometimes one of those hits something, and memory locations/etc are so small nowadays that they can be directly affected by these events.

But what really causes it...who knows?

 

[TTSProd]

All fixed! I reloaded my previous settings last night and this morning it was working. Before I reloaded my settings I read the current setting from the CA and took a screen shot. You were correct, there was no evident change in the settings that I can see after reloading, the only difference is it works now.

I, like you, have encountered issues like this with other devices that inexplicably stop working or just stop working properly. I’m not ruling anything out because I can never REALLY know, and it could very well be cosmic rays. But while trying to figure this out I was starting to suspect that maybe static was the culprit here. It’s winter here and I commute to work and that cold dry air is bad for static electricity build up so maybe I accidentally zapped the CA. Luckily, whatever it was didn’t cause permanent damage. I sure am happy to have torque sensing back! Thanks so much for all your help!

 

[amberwolf]

I'm glad it worked.

If it was static, with most electronics that usually causes permanent damage, which is usually on the inputs (buttons, wires, etc) or occasionally damages a display (causes lines or dead spots, wierd coloration, etc). It's not always complete failure--more often it is "wierdness" that just never goes away, unpredictable behavior or simply incorrect behavior, etc.

If the MCU board in the CA was directly accessible without opening the insulative plastic case, static from a touch could damage things, but even the most extreme ESD that wouldn't injure you won't make it thru the case; probably not even thru the thin rubber over the buttons, so unless you had a charge on you and touched an actual electrical connection, I doubt any ESD could reach anything in there.

It can also be just defective or damaged memory locations in the MCU itself, from the factory or subsequent handling (where ESD *could* reach it and damage things).

It could even be a bug in the software, either the built in stuff in the MCU from the factory, or the Grin firmware. Software can be very complicated. I doubt they do it this way in the CA due to memory limitations, but there are softwares that for instance take the settings you see and change, then copy those to "working memory" and run from those.

If they use specific memory locations (which the CA probably does, coded at the machine-language / assembly code level, agian due to memory limitations) for specific variables, then a failure of that location could cause a permanent problem with that function even though all the settings appear correct, and rewriting those settings wouldn't fix it.

If the location isn't damaged but the data in it got corrupted somehow, and that data is only placed into it at the moment a setting changes (while the setting itself is still in a different location), the situation you ran into could happen--and in these cases, changing the setting to something else and then back could fix it, but it might not, depending on how the data is read and written and what exactly is wrong.



High-energy charged particles from the atmosphere or space, on the other hand, could penetrate most device casings (possibly even metal controller casings, but certainly plastic ones). So...without knowing what really happened (which in almost every incident is the case), my bet is for this.

 

[TTSProd]

I'm glad it worked.

If it was static, with most electronics that usually causes permanent damage, which is usually on the inputs (buttons, wires, etc) or occasionally damages a display (causes lines or dead spots, wierd coloration, etc). It's not always complete failure--more often it is "wierdness" that just never goes away, unpredictable behavior or simply incorrect behavior, etc.

If the MCU board in the CA was directly accessible without opening the insulative plastic case, static from a touch could damage things, but even the most extreme ESD that wouldn't injure you won't make it thru the case; probably not even thru the thin rubber over the buttons, so unless you had a charge on you and touched an actual electrical connection, I doubt any ESD could reach anything in there.

It can also be just defective or damaged memory locations in the MCU itself, from the factory or subsequent handling (where ESD *could* reach it and damage things).

It could even be a bug in the software, either the built in stuff in the MCU from the factory, or the Grin firmware. Software can be very complicated. I doubt they do it this way in the CA due to memory limitations, but there are softwares that for instance take the settings you see and change, then copy those to "working memory" and run from those.

If they use specific memory locations (which the CA probably does, coded at the machine-language / assembly code level, agian due to memory limitations) for specific variables, then a failure of that location could cause a permanent problem with that function even though all the settings appear correct, and rewriting those settings wouldn't fix it.

If the location isn't damaged but the data in it got corrupted somehow, and that data is only placed into it at the moment a setting changes (while the setting itself is still in a different location), the situation you ran into could happen--and in these cases, changing the setting to something else and then back could fix it, but it might not, depending on how the data is read and written and what exactly is wrong.



High-energy charged particles from the atmosphere or space, on the other hand, could penetrate most device casings (possibly even metal controller casings, but certainly plastic ones). So...without knowing what really happened (which in almost every incident is the case), my bet is for this.

I would say your assessment does make the most sense. After riding all week there doesn’t seem to be any lasting effects that would indicate anything was permanently damaged. Just a bit of corrupted data I guess. If I have inexplicable stuff like this happen in the future my first step will be to load the last known working setup from the configuration app. Thanks so much for your, fulsome responses.

 

[amberwolf]

You're welcome...I'm good at "fulsome responses".

(also known as overexplaining far beyond what anyone really wants to read )