Cycle Analyst V3 preview and first beta release

[TotalConfusion]

Hi All,

I have a CA problem i believe and tried to skip over most of the pages but couldn't find anything on it But Tek mentioning a few post ago about 4.5 Calibrate Current and Voltage which caught my eye.

My Problem is an incorrect voltage display on my CA .... In Short..... I did a battery test to see how much capacity i have lost over the 1st 12 months of use on my battery by running it down to the BMS cut off limit of 2.75 volts and when i recharged my battery i was seeing 88.7 Volts on the CA instead of 88.2.

I Have checked my battery once fully charged and it is only showing 88.1 volts after the charger stops which is normal but the CA is putting a reading out of 88.7 volts so i decided to put the Multimeter on the charger and its only putting out 88.2 volts like it should.

If i charge the battery to 90% by flicking a EM3EV charger over on the switch it charges to 86.7 volts instead of 86.2, There seems to be a 0.5 Volt discrepancy.

Has anyone any ideas how this has happened and if i can change the values at all to be back to normal.

 

[amberwolf]

Before fiddling with teh CA's settings:

What multimeter? Some are not very accurate. I have had a lot of meters over the years, and some were off by several tenths of a volt in certain range settings. The only one I have that I actually trust as an "absolute" reference is my old Fluke 77 III; the rest are for basic checks though several of them have proven reliable enough in general.

Did that same meter previously read identical to the CA? (if you don't know, it's tough to use as a comparison, unless you can compare them both to something else that you do know they both read the same before).

Did you check the multimeter's own battery? Sometimes a low one can cause improper readings (though those usually would be higher than actual).

 

[TotalConfusion]

Before fiddling with teh CA's settings:

What multimeter? Some are not very accurate. I have had a lot of meters over the years, and some were off by several tenths of a volt in certain range settings. The only one I have that I actually trust as an "absolute" reference is my old Fluke 77 III; the rest are for basic checks though several of them have proven reliable enough in general.

Did that same meter previously read identical to the CA? (if you don't know, it's tough to use as a comparison, unless you can compare them both to something else that you do know they both read the same before).

Did you check the multimeter's own battery? Sometimes a low one can cause improper readings (though those usually would be higher than actual).

Its not a Fluke but its German made and batteries are good in it.

when the CA is showing 88.7 volts on the display after charging to 100% if i stick the multimeter probes into the CA's Output Power Jack my multimeter reads 88.2Volts where as the Display is showing 88.7Volts, Check battery and get a reading of 88.2Volts and the same on the charger.

It seems the CA is giving a wrong reading of only 0.5 Volts but it had me worried about over charging at 1st but this doesnt seem the case.

 

[teklektik]

a short question about "battery internal resistance": how often is this display refreshed? i know how it's calculated (max voltage sag/max current), but it seems to be nailed to a value which doesn't change even a bit, even a thousand of a ohm.

does someone have an answer of that? or maybe i found a bug? this value just doesn't change a single bit. so i have no idea if this value is valid at all.

Hey izeman-
Here's a 'non-authoritative' view that I believe to be correct:

• The RBatt (IR) value is recomputed once a second. It's a weighted rolling average of previous RBatt results and the voltage and current information over the last 1 second interval - this sort of 'accumulates validity' as more samples get folded into the rolling average - old values become increasingly discounted. There's some other twists and turns in there, but that's the 30,000 ft view...

You shouldn't expect that the resistance of your battery should change so the fact that yours seems fixed is A Good Thing. I generally check it after a ride and find it more or less constant - it does vary with temperature (either ambient or from internal heating when flogging the bike). Once in a great while it jumps up a bit which is an indication of a power connector issue - a little Anderson re-seating and things go back to normal on the next ride. This is a handy diagnostic side-effect of a value that primarily is there to support the Gas Gauge.

• 
  FWIW: Since the open circuit (unloaded) voltage is obviously unavailable when underway it would normally be impossible to use measured voltage to determine state of charge for the Gas Gauge icon. RBatt is used to work backwards to an estimated open circuit voltage using the 'underway and loaded' V & A. This estimate can then be applied to the SoC tables according to chemistry to operate the Gas Gauge display.

 

[izeman]

thanks. so you say it does collect data and averages it. i may try and load a new firmware which deletes all data and start from scratch. or i could remove some of the packs. it's a 12s4p setup. so i can remove one or two 6s packs from each side and then the IR must change. it should double. i think i can remember that i already did that and it did NOT change. so maybe a bug. it sits at 0.199 and doesn't change a thousand of an ohm. i'l check and update the thread.

 

[Alan B]

I think it only calculates it under certain conditions, so it won't change until the conditions trigger a calculation. It needs the right data to do the calc.

 

[izeman]

I think it only calculates it under certain conditions, so it won't change until the conditions trigger a calculation. It needs the right data to do the calc.

maybe justin could add some comment here. he would be the only one to explain the algorithm w/o guessing

 

[stonezone]

So at the beginning of a ride yesterday my CA rebooted and when it came back was reporting -90 Amps and 40k watts (I forget the exact W but I think that's about right). I reset it to 0 on the trail from the menu but still no spin on the motor. Went home and swapped to a backup CA, and everything worked. Swapped throttles on the CA but still no motor spin (throttle in/out was reading fine. Flashed firmware and loaded configs from working CA but no response. No ebrakes or LV cutoff enabled and when I plug the throttle directly into the controller it works as long as the CA isn't plugged in (even with throttle set to off/0v or WOT).

This is on a Lyen 18f with 15s Lipo. Everything was find for a couple of weeks. I wasn't running hot or at much of a load.

any suggestions?

 

[teklektik]

So at the beginning of a ride yesterday my CA rebooted and when it came back was reporting -90 Amps and 40k watts...

You did the proper steps for repair, but your CA may have corrupted calibration values in EEPROM and wonky values can give the CA fits. Simply re-flashing the firmware portion will not cure this.

Please follow the steps in this post. This will replace your factory device-specific calibration values with some valid 'pretty damn close' generic ones.

It's a little unclear how messed up your other existing settings might be and saving them to file first may have questionable results if restored. It's best to follow the post and then enter a fresh configuration to ensure all is clean and tidy...

If this doesn't fix it, then you need to email Grin Support, reference your post here so they see what you did, and they will take it from there - probably repair/replace.

 

[stonezone]

thanks for the quick response... I'll follow those steps!

 

[r3volved]

Can you save your setup and compare files against the 'default'? Is the file encoded/encrypted?

 

[teklektik]

I think it only calculates it under certain conditions, so it won't change until the conditions trigger a calculation. It needs the right data to do the calc.

maybe justin could add some comment here. he would be the only one to explain the algorithm w/o guessing

Configurable setup data for the calculation:

• Filtered (rolling averaged) IR stored in EEPROM,
• 'Filter Rate' which scales the effect on the existing filter (rolling average) value of the 'new' IR value, and
• threshold criteria 'delta current' over the last second needed for the calculation to proceed.

 

[izeman]

I think it only calculates it under certain conditions, so it won't change until the conditions trigger a calculation. It needs the right data to do the calc.

maybe justin could add some comment here. he would be the only one to explain the algorithm w/o guessing

Configurable setup data for the calculation...

cool! thanks a lot. never realized that section of the setup program - which i don't use very often i must admit. doing it on the fly directly is all i need most of the time. i'll try that!

 

[stonezone]

After some guidance from Justin, it turns out my blue Sp had a broken connection that was causing the crazy negative amp reading. It also may have shorted with the red V+ where it was broken.

This is some good info for this thread:

"you'll get a huge amperage like that whenever there is a bad connection on either the S+ ,S-, or Ground wires between the CA and the controller (pins 2, 3, and 4 of the 6-pin plug). If it's the ground wire (pin 2), then your voltage will be wonky as well. You don't want to rezero the amperage while you have the bad contact or you'll just be pegging the saturated op-amps as your new zero point and won't see any figures.

Use a multimeter and check that you have effectively perfect continuity between these 3 pins on the controller plug, and do a visual inspection of the connector itself."

So now the CA works fine, and a throttle connected to the controller spins the motor whether the CA is plugged in or not (before i fixed Sp it wouldn't spin with the CA plugged in). I can see throttle in/out voltage functioning correctly in the CA but it's not feeding back to the controller? I swapped another CA, loaded my same last working configs, and the throttle works from the CA.

So... something in my CA isn't happy, and it's not sending the throttle output back to my controller like it should?

Tomorrow's another day.

 

[rowbiker]

Subject: No throttle output from CA-v3 when using TDCM BB

We need help resolving an issue in which the CA-v3 is not supplying a positive throttle signal to the controller when we believe it should be doing so, given both positive cadence (PAS) and torque (+ n-M values) signals provided to the CA by a TDCM bottom bracket.

We installed a TDCM PAS/Torque bottom bracket in a "Skybike" (see http://skyridetechnology.com/skyride-products/recumbent-skybike/). The pedals in the Skybike run in a custom made, but otherwise standard bottom bracket shell, now fitted with a Grin TDCM_114 (Torque Sensing Bottom Bracket, CA3 Compatible. 114mm Spindle; see http://www.ebikes.ca/shop/ebike-parts/torque-sensors/tdcm-114.html). The output from this TDCM BB is connected to the CA with the provided cabling and connectors (no extensions or adapters). The chainline from the BB runs vertically instead of horizontally as on a "normal" bike, but the TDCM was installed to compensate for this. We believe the BB is working correctly at the hardware level because the relevant analytical screens on the CA show the expected "live" activity and data. Specificially, both the RPM and Direction icons in the PAS screen section are active when the pedals are turned. Also, a varying torque readout in n-M units in the CA's Torque screen section is displayed when the pedals are touched. Even with simple hand-held pressure on the pedals, values upward of 20 n-M can be easily reached. Letting the pedals return to "idle" position drops the n-M readings to zero, or slightly below. The torque transitions are smooth and appear to be proportional to the amount of pressure applied to the pedals.

The "pod" is powered by a custom version of the Revolt 100 BLDC motor (made by Alexey in Israel -- used by others on the ES forum as well), which is located inside the monorail track from which the SkyBike is suspended. The motor is a very typical BLDC, with the standard 3 phase power wires, and five smaller wires for the three Hall sensors plus two for the excitation voltage.

We are controlling the motor drive with a very recently acquired CA-v3 (prelim9 firmware) -DP and GrinFineon (40 Amp) controller, running in sensorless mode. To do our testing, we're using the Grin 48V eZee 10.5AHr battery pack. When using a standard ebike (Hall sensor type) manual throttle, connected either to the CA or the controller, everything works fine, including the variable regen on the ebrake connected directly to the controller. Because of this, we believe the problem we are experiencing lies in the realm of software in the CA, either in a configuration setting we have incorrectly applied, or in something else that keeps the CA from issuing a positive throttle signal/voltage when the BB sends active PAS and torque data.

The PAS/Torque configuration settings we are using are those gleaned from the ES forum, including Justin's and teklektik's posts which are specifically geared to the TDCM, as opposed to the THUN, for instance. (If there is an "official" set of instructions for programming the CA-TDCM combination, please let us know!) These settings are as follows, corresponding to their display order in the "CA Setup Utility V1.20 for Windows", recently downloaded and being run on a laptop running Windows XP. Programming is done via the stock cable supplied by Grin, with the recommended FTDI drivers installed. Serial port for this installation is at COM5:

PAS section:
PAS Poles 12 Pulses/Rev
Quadrature on Dir Pad Enabled (displays as "2 Wire" on CA)
Direction Polarity 5V = Reverse (arrived at via "trial and error")
Start RPM Threshold 3 RPM
Stop RPM Threshold 11 RPM
AutoPAS Power Level 500 W (default, not used)
Max No-Pedal Throttle 99.00 mi/hr (default)
PAS Assist Mode Torq PAS


Torque Sensor Settings
Torque Sensor Type Custom
Torque Scale 66.0 Nm/V (depends on chainring diameter, per Justin. Began with value of "50")
Torque Offset 2.50 V
Torque Assist Factor 2.00 x HW
Trq Assist Start Point 100 Watts
Torque Assist Averaging 8 Poles

Notes about these settings:

Direction Polarity:
We initially used "5V = Forward" for "Direction Polarity", but that resulted in the motor making some jerking motions, but only when pedaling backwards. Changing this manually via the buttons on the CA resulted in the motor running smoothly while pedaling forward. Unfortunately, we've not been able to replicate this -- occasionally the motor still stutters momentarily when pedaling forward, but mostly there is no throttle signal at all. Another concern is that frequently this setting is not correctly written from the Setup Utility to the CA's EEPROM. Selecting "5V = Reverse" from the Window program's dropdown box, saving it to the relevant configuration *.hex file, and then writing it to the CA's memory results in the CA then displaying "5V = Fwd". Other values can be changed and written without issue.

Quadrature on Dir Pad:
An "Enabled" selection in the CA Utility program apparently translates into a "PAS -> Quadrtr 2 Wire" setting in the CA's memory.

Start/Stop RPM Threshold:
We used different values for these categories, following suggestions found in ES posts, but none changed the behavior described here.

While we understand that a number of these settings are for "tweaking" the performance obtained from the PAS and torque enabled bottom bracket, we are puzzled by the complete lack of throttle "out" signal when the CA is obviously receiving both PAS and torque input signals. We think it's fair to rule out settings like Low Voltage Cutoff, etc., since the system works entirely as expected when operated via a manual throttle. The system certainly isn't behaving as if the EEPROM is corrupted, since the shunt value is correct, as are other voltage and current measurements which depend on the correct shunt configuration.

Are we missing something in the PAS/Torque configuration realm that would explain what we are experiencing?

[This missive has been simultaneously posted in the ES "Cycle Analyst V3 preview and first beta release" thread, and sent via email to Justin at info@ebikes.ca. If there is a better place to have CA - TDCM BB discussions, let me know.]

 

[teklektik]

(If there is an "official" set of instructions for programming the CA-TDCM combination, please let us know!)

Defaults for the TDCM BB are on p 42 of the Guide.

Torque Offset 2.50 V

Suspiciously, this looks like the default value.
Please use the console, enter Setup, navigate to the Trq Offset screen and press-hold the right button with no pressure on the pedals.
This will measure and store the no-pedal torque offset baseline for subsequent torque measurements.

Leave/re-enter Setup and examine the SETUP TRQ SENSR section screen while you fiddle the pedals. See if the new zero offset voltage is there and the Nm are about 0 with no pedal force. Apply some torque and Nms and voltage should increase.

See if that helps.

 

[stonezone]

I must have missed something the other day. Was going to verify throttle out voltage today and it's working again. So it was just the broken connection on the blue wire, nothing shorted. thanks for the help, I'm glad I didn't have to wait for the slow boat to Canada from here...

SZ

 

[rowbiker]

Regarding the TDCM issue posted above.

Tek, thanks for the link to page 42. I had relied on my hard copy of your manual, which at 2 years old didn't yet have the updates I needed. Lesson learned.

Justin sent a very detailed email response with tons of debugging information. As soon as we finish working our way through that, I will post an update here summarizing what we learned about correctly setting up the TDCM bottom bracket.

 

[teklektik]

I will post an update here summarizing what we learned about correctly setting up the TDCM bottom bracket.

That would be good. I have no TDCM experience and the Guide is pretty thin in that regard....

 

[Erikjan]

Hi, i have an CA v3 on a 2WD bike with 2 Mac motor's.
It works good, only the speedlimit is a little problem. I can not find the right settings for the gain values.
Is there sombody with a bike @50V and 50A that can tell his settings so i can try if they work for me too.

i try the steps from the manual but i get osilating around the speed i set

sorry for the bad spell.

 

[mrbill]

Battery gas gauge never shows "three dots".

Anyone else observe that their battery gas gauge drops from 4 to 2 dots, but never shows 3 dots (bottom row filled)? I'm using a large capacity battery made up of Samsung INR18650-29E cells with the CAV3 (Prelim9) set to "LiPo". Whether I wire them as 7s or 14s on two different units, I have never seen the display show only 3 dots.

For a while I thought I never happened to glance at the display when it showed 3 dots, but yesterday I made a point of watching it as I discharged from 4 dots directly to 2 dots, then back up to 4 dots a few minutes after I removed the load.

Is there an explanation for this behavior?

Thanks.

 

[teklektik]

Battery gas gauge never shows "three dots".
...
Is there an explanation for this behavior?

That's the design.
There are (16 values 0-15) but (16 'charge' pixels + 'No Charge' = 17 states) so a little sleight of hand happens and one combination goes away...

(There must an award for noticing this.... )

 

[mrbill]

Battery gas gauge never shows "three dots".
...
Is there an explanation for this behavior?

That's the design.
There are (16 values 0-15) but (16 'charge' pixels + 'No Charge' = 17 states) so a little sleight of hand happens and one combination goes away...

(There must an award for noticing this.... )

Although the reasons are completely unrelated, this reminds me of the occasions I have stayed in a high-rise hotel (esp. in places like Las Vegas) and discovered that there was no 13th floor as I watched the numbers tick by on the elevator display.

 

[TotalConfusion]

Before fiddling with teh CA's settings:

What multimeter? Some are not very accurate. I have had a lot of meters over the years, and some were off by several tenths of a volt in certain range settings. The only one I have that I actually trust as an "absolute" reference is my old Fluke 77 III; the rest are for basic checks though several of them have proven reliable enough in general.

Did that same meter previously read identical to the CA? (if you don't know, it's tough to use as a comparison, unless you can compare them both to something else that you do know they both read the same before).

Did you check the multimeter's own battery? Sometimes a low one can cause improper readings (though those usually would be higher than actual).

Its not a Fluke but its German made and batteries are good in it.

when the CA is showing 88.7 volts on the display after charging to 100% if i stick the multimeter probes into the CA's Output Power Jack my multimeter reads 88.2Volts where as the Display is showing 88.7Volts, Check battery and get a reading of 88.2Volts and the same on the charger.

It seems the CA is giving a wrong reading of only 0.5 Volts but it had me worried about over charging at 1st but this doesnt seem the case.

Anyone have any ideas on what could be causing this please as my CAV3 is still showing 0.5 Volts extra on the display than what the battery is actually at.

 

[marcn]

Love this thread.

Thanks to Keplers post here https://www.endless-sphere.com/forums/viewtopic.php?f=4&t=37964&start=1575#p752388 for some insight into some issues with PAS we're trying to figure out ourselves. We didn't want the PAS kick some have experienced, and slowing down the fastrate which affects throttle as well wouldn't be ideal.

We're using an Aux Pot, and would like to have it control the PAS level when we're using AutoTorqPAS. Is this possible, and if so how would it work in relation to the tables found on page 52 and 53 of the guide?

For example, at <=55RPM we would like 500w, and upto 90RPM, 1000w.

Can we use the 5k potentiometer to adjust these values on the fly from 0,0 to 500,1000w, once we've set the max limits (by inputting trqscale, asstfactr and asststart)?

Edit: Just read that the potentiometer input would only change the AsstFactr, and using AutoTorqPAS would have no effect. Is this correct?