Cycle Analyst V3 preview and first beta release

[Doctorbass]

Justin, I have an idea that might interest some of us ( i dont know if it already exist for the C-A but it might be interesting! )

FOR THESE HYPERMILING GUYS: WE COULD ALSO JUST PRESET THE DESIRED WH/MILE AND HAVE THE THROTTLE AUTOMATICLY ADJUSTED TO FOLLOW THAT TARGET!




I made a thread about that here: http://www.endless-sphere.com/forums/viewtopic.php?f=3&t=51282

Doc

 

[MattyCiii]

A cautionary tale:

I've had an intermittent problem where one day the bike works fine, the next... Intermittent power drops. This can be quite scary because I would get very powerful bursts of power - a jerky and dangerous ride.

I'll give the bottom line up front: The magnet for my speedometer was just a little too far away from the pickup to give a clear signal. It would somehow induce ridiculous speeds - like 675mph, bouncing to 322mph, to whatever. And since these are all above the 99mph default speed limit, my throttle would cut out. As a fix I installed a higher quality magnet - the one I had kept rotating around the spoke. Testing to begin tomorrow.

The long story:
At first I thought I had a loose phase wire. The CAv3 showed consistent throttle input and output through the diagnostic screen - so this was at least consistent. Then as I slowly ruled that out, I thought maybe the interface between the CA shunt and the RC controller was loose. I mean, it really does feel loose. It bothered me that at a standstill, I could not replicate the problem (even while wiggling the signal and phase wires, to mimic vibrations from riding). But who would
But after some careful riding and observation, I saw the speedo bounce around in lock-step with the problem.

So... A problem all my fault, and in the grand scheme, the CA is working exactly as designed... but the 99mph cutoff combined with crazy ridiculous phantom speeds made for a scary ride.

 

[teklektik]

Has anyone else tried dual controllers with the CA3 throttle by wire?

Yep. I built my bike on the CA V2 and upgraded it to V3 last year as soon as the V3 pre-production units were available. I am feeding two EB312 controllers through a common external shunt for the CA and using closed loop 'power throttle' mode. Works nice.

It didn't work well for me with EB2xx Infineon type controllers. It worked but the controllers seemed to fight over the throttle signal and not share equally.

We have slightly different situations - I have 2WD with gear motors (only 3.3kW) and you have 2-controller 1WD (22kW) which probably makes my situation a bit more tolerant than yours. I spent some time on a test incline adjusting the controllers so they they yielded fairly close performance climbing from a dead start. This resulted in slightly different controller configurations - probably due to shunt and other component variations.

The situation is puzzling since you are not reporting controller interaction issues when running the throttle with direct connect to the the two controllers (only with the CA). How are you determining that one controller is detecting more throttle than the other? How is you CA obtaining power and shunt? There may be GND reference differences between the controllers when under load that might skew the apparent throttle for one controller if the CA draws it's ground from the other. Fairly serious GND reference issues with dual controllers, a single throttle, and high-amp motors have been reported in other threads - your issue might be similar but tied to the CA Gnd.

I ran it for a while and I seem to have messed something up in the CA3's throttle circuitry in the process. I was able to get home, but the throttle became like an on/off switch,...

So - did (or does) the Diagnostic Screen show the proper OUT voltage? Does this coincide with the actual voltage output as measured with a DMM or is the output voltage wonky?

 

[teklektik]

I've done some investigations and I have a clue:
The CA output is OK in Open loop but not in close loop (bike still, speedometer still, back wheel lifted): the CA output on the debug display (4.2V) does not fit with the measured one (saturation at around 2.2V).
...
Should not the debug display output always fit with the measurement?

I'm thinking that a 'scope on the output might reveal that the output is not a DC level but is pulsing instead - your meter is reading the average.

Frankly, it would be intellectually comforting to better understand operation on the stand, but with the motor unloaded and unable to draw the expected current and with no weight to accelerate to provide damping, operation on the stand clearly throws closed loop operation outside its defined operating range. I wouldn't lose any sleep over it - as long as you can achieve full speed when actually riding...

 

[Luge_a_foin]

I wouldn't lose any sleep over it - as long as you can achieve full speed when actually riding...

Well, actually the thing is I cannot reach full speed when I use the CA as throttle input while riding. This is a big limitation with regard to CA functions (only observer in this case).

 

[teklektik]

...May be it is because I've deactivated the limits with some 0 values as indicated in the UM.

Does the bike actually achieve full speed under load (not on the stand)?

I've done some investigations and I have a clue:
The CA output is OK in Open loop but not in close loop (bike still, speedometer still, back wheel lifted)...

I wouldn't lose any sleep over it - as long as you can achieve full speed when actually riding...

Well, actually the thing is I cannot reach full speed when I use the CA as throttle input while riding.

It is unclear what the bike is doing in PassThru mode when riding - let's forget about operation on the stand.

I think what you need to do is re-flash the firmware, follow the setup instructions for basic PassThru mode without making extraneous settings, verify that you can achieve full speed in PassThru mode, and verify that ramping is adjusted and works properly. Verify also that the current and voltages look okay under load. If the current looks strange, then calibrate the shunt. This will get the bike to a known state with a working throttle and proper defaults in place.

Only when this is successful, try Current Throttle mode. If that shows problems, please show your settings.

I know this may sound like a step backward, but the re-flash and setup go very quickly. The best way to get the bike working is to get it to a known state - there are so many interacting settings that it is difficult to remotely diagnose problems without some solid understanding of the configuration and verification that basic operation is okay.

 

[teklektik]

I've had an intermittent problem where one day the bike works fine, the next... Intermittent power drops. This can be quite scary because I would get very powerful bursts of power - a jerky and dangerous ride.

I'll give the bottom line up front: The magnet for my speedometer was just a little too far away from the pickup to give a clear signal. It would somehow induce ridiculous speeds - like 675mph, bouncing to 322mph, to whatever. And since these are all above the 99mph default speed limit, my throttle would cut out. As a fix I installed a higher quality magnet - the one I had kept rotating around the spoke.

Contact bounce with the reed switch in the pickup seems to be problematic. I'm thinking that over-150mph indications might be filtered out by firmware (allowances made for Luke and Doc Bass ). It seems the bogus contact bounce indications have been reported in the 300-600mph range.

I had a cheapie PAS wheel that I was considering mounting to the brake rotor by drilling and inserting some short 1/8" aluminum standoffs into the PAS wheel against which the rotor bolts could be tightened. I was just looking for a higher frequency speed detector to make Speed features work smoothly at low speeds - no real issue with my speedo pickup - but the hall detector would eliminate any bounce as a side effect. But - I used the PAS unit as originally intended instead...

If you wanted a permanent contact-bounce fix, you might affix a hall sensor to a scrap of plastic with a bit of epoxy putty (Home Depot) and tie-wrap the new sensor into place in lieu of your present pickup (or use a cheap PAS wheel pickup). This would draw an extra 4-5ma if you have the current available.

Just a thought...

 

[justin_le]

Hey everyone, back in the swing and will have a lot of catching up to do with some of the suggestions and discussions on this thread. In the meantime though I have uploaded a Prelim6 version of the CA3 firmware that has the following changes:


1) The timing error debug screen has been eliminated so you shouldn't see those occasional instances where it looks like the CA has "crashed" during otherwise normal operation when it's doing too much processing. There is still a watchdog timer running so if some circumstances ever do cause the CA to get locked up in a loop, the throttle output would turn off you would instead see the message "WatchDog Error" with some numbers that could be useful for me.

2) The AutoCruise throttle features has been substantially changed for better average tracking and should be much easier to have latch even with a twitchy thumb throttle. The allowable range for the cruise sensitivity is increased to +-0.5V, but you should find something in the 0.15-0.2V range is good for most cases.

3) There is an additional option to set the averaging of the torque sensor when running in torque assist mode. So if you have say a THUN sensor with 8 PAS poles, you could set the averaging to 8, 16, or 24, depending on if you want to average over the last 1, 2, or 3 full pedal rotations. Averaging over 3 rotations will naturally be the smoothest, but then also a bit slower to respond to changes in your pedal force. If you are just starting to pedal from a stop, then it will simply average the torque up to that point, so the behaviour when starting off is consistently responsive and unaffected by the longer average.

4) The bug that could make the datalogging output cease after certain mode changes or parameter changes in the setup menu is fixed. So it should always be sending data whenever the screen is in normal operation, and not sending data while in the setup menu.

5) Some minor tweaks in the setup menu, like the the PAS_Power setting not showing up if in the 2nd or 3rd profile..

Earlier there had been reports of people using PAS mode having periodic short cutout glitches while riding. I've tried extensively to replicate this both with the THUN sensors and also just the magnet ring PAS sensors without success (and that includes the whole 1800km ride to SF). But I realized that one thing which could cause this is if people have their Min Throttle Input set too close to the actual off throttle voltage. In that case, there could be problems from noise glitches on the throttle line that would momentarily bring the throttle higher than the min throttle value (But still effectively zero throttle). When the CA sees this it would assume that the user is engaging the throttle and hence would shut off the PAS mode and follow the throttle instead. Let me know if that might ring true as a possibility. I would say that at least 0.15V of buffer zone between your actual throttle off voltage and the Min Throttle should be recommended.

-Justin

 

[justin_le]

Is there a way to display voltage by dividing my total voltage by how many cells I have in a series. For example , I would rather look at 3.85 then 69.3 for an 18s setup on the main cycle analyst screen.
I know its the same thing but for some reason I prefer looking at what each individual cell is to better gauge my battery life.

This is a decent suggestion, and a fairly easy one to implement as an option to show a normalized per-cell level voltage rather than the total pack voltage on the main screen for those who want it. I tend to ride these days with a 10S pack so the math is easy, but at 48V / 13S I'm often doing mental arithmetic to figure out just what the 43V display means on a per-cell basis. The CA is already doing everything at the cell level when doing the SOC calculations so it's just a matter of displaying this rather than keeping it all internal.

So we'll put this on the consideration bin for the V3.1 code. The plan for the next round of upgrades after the official release is to add a lot more display customizations (like choosing a graphical display of temperature instead of human watts, selecting which items to toggle in the bottom right for the Ah/km/temp display etc) so this would be well inline with that.

-Justin

 

[justin_le]

as the bike is higher powered than the legal 250-500w limit of most countries, i had to set throttle voltage ramp up to a very slow setting to have a nice engagement of the pas sensor. this effects the normal throttle as well, and this is very sluggish now, but a fast response can be needed in those situations where you have to accelerate fast.

This is indeed a bit of an oversight. At the moment, since the auto PAS feature is using the power feedback loop of the CA, you can make it engage more slowly when pedaling by setting the WGain value really low (like 5 or 10), and then leave your output ramp rates fast for the good throttle response. So then what limits the ramping when you pedal is not the overall throttle ramp rate but the power limit feedback loop. But this won't work if you have configured your throttle as a power throttle as well, it would have to be a current or pass-thru throttle.

There is room to add one more item to the PAS menu to set the autoPAS ramp rate in addition to the power. Uh oh time for a prelim7?

.) a more realistic battery gauge for lifepo4 cells. even with the new values, a 24s pack with 80v is NOT empty. that's 3.33v per cell.

What happens here is that at 3.33V/cell, the CA has no idea if you are at 20% charged or 80% charged, so it's not updating the SOC info from your voltage but is instead using just the incremental Ah. However, if at some point you turned off your ebike and it has a slow capacitor ramp down in voltage, or it had a lot of sag under load, then the CA would have assumed that your pack was indeed flat (<3V / cell) and adjusted the SOC accordingly, and then when you turned it back on again it restores the last SOC it was at when it fully shut off. However it can't really increase with confidence the SOC in a LiFePO4 battery until it sees over 3.4 V/cell so it just sits there looking mostly empty. me

Is it possible that this explains your situation? There is definitely some tweaking still needed but can you try starting off with a fresh charge on the LiFePO4 pack so that the CA's SOC is reset to 100% and then see if it shows a consistent decline the capacity graphic?

.) a freely configurable screen. me for example wants to know motor temp all the time, and trip km is of no use. and looking at the display while doing a climb and trying to get the temperature can be dangerous. you look at the display, trip distance is shown, you wait for the temp to show, wait, wait, and should better watch the street

See, all you need to do is synchronize your glancing to the known frequency at which the display toggles. Then you are guaranteed that each time you look down it is a new piece of data that and no need fix your eyes on the screen waiting :wink:

Tentative plan (V3.1) is to have the lower right corner be configurable much like you can select which display screens show up on button presses, and give the option of Km, Ah, Temperature, Wh/Km, AvgS, MaxS, etc. It would then toggle though each of those that you had selected. If you only selected one item, it wouldn't toggle at all and would only show that.

-Justin

 

[Architectonic]

I use the speedometer sensor, I tend to get unrealistic peaks on the 'max speed', but I never actually see the speed being displayed incorrectly.

I'm guessing this is due to some sort of noise/vibration caused by my front disc brake? (BB7/8" rotor)

 

[Luge_a_foin]

It is unclear what the bike is doing in PassThru mode when riding - let's forget about operation on the stand.
I think what you need to do is re-flash the firmware, follow the setup instructions for basic PassThru mode without making extraneous settings, verify that you can achieve full speed in PassThru mode, and verify that ramping is adjusted and works properly. Verify also that the current and voltages look okay under load. If the current looks strange, then calibrate the shunt. This will get the bike to a known state with a working throttle and proper defaults in place.

Only when this is successful, try Current Throttle mode. If that shows problems, please show your settings.

I know this may sound like a step backward, but the re-flash and setup go very quickly. The best way to get the bike working is to get it to a known state - there are so many interacting settings that it is difficult to remotely diagnose problems without some solid understanding of the configuration and verification that basic operation is okay.

You're right, I will start again like this.
Nevertheless just to be clear I didn't use the same definition as yours in my posts for Open Loop and Close Loop (I had again a look to your doc). What I did call Open loop was CA throttle output not plugged to controller. And for Close loop the CA throttle plugged to the controller and in my case the CA configured to pass-thru. Actually I've never tried any Current modes or others.

 

[teklektik]

... I will start again like this.
Nevertheless just to be clear I didn't use the same definition as yours in my posts for Open Loop and Close Loop ... Actually I've never tried any Current modes or others.

Thanks for the update - that makes the situation much clearer. Please post back if problems persist for the basic setup in 'Round Two'.

 

[izeman]

What happens here is that at 3.33V/cell, the CA has no idea if you are at 20% charged or 80% charged, so it's not updating the SOC info from your voltage but is instead using just the incremental Ah. However, if at some point you turned off your ebike and it has a slow capacitor ramp down in voltage, or it had a lot of sag under load, then the CA would have assumed that your pack was indeed flat (<3V / cell) and adjusted the SOC accordingly, and then when you turned it back on again it restores the last SOC it was at when it fully shut off. However it can't really increase with confidence the SOC in a LiFePO4 battery until it sees over 3.4 V/cell so it just sits there looking mostly empty. me

Is it possible that this explains your situation? There is definitely some tweaking still needed but can you try starting off with a fresh charge on the LiFePO4 pack so that the CA's SOC is reset to 100% and then see if it shows a consistent decline the capacity graphic?

hmmmm. not sure. normally i fully charge, reset data, then ride and watch the gauge. even after 10Ah used, which means 10Ah left in the pack the gauge shows an almost empty battery instead of a half full one. i don't turn on/off the ca during a ride normally.
but now that i know that i takes used Ah into the equation i will closely watch that.
the problem with lifepo4 is that it sits at 3.3v most of it's time - as you already mentioned. as you asked for voltage sag: no. not really a lot. ca shows internal resistance of 0.06mohm. so quite good.
can't wait for 3.1 looks that it brings all i still want

 

[Merlin]

as the bike is higher powered than the legal 250-500w limit of most countries, i had to set throttle voltage ramp up to a very slow setting to have a nice engagement of the pas sensor. this effects the normal throttle as well, and this is very sluggish now, but a fast response can be needed in those situations where you have to accelerate fast.

This is indeed a bit of an oversight. At the moment, since the auto PAS feature is using the power feedback loop of the CA, you can make it engage more slowly when pedaling by setting the WGain value really low (like 5 or 10), and then leave your output ramp rates fast for the good throttle response. So then what limits the ramping when you pedal is not the overall throttle ramp rate but the power limit feedback loop. But this won't work if you have configured your throttle as a power throttle as well, it would have to be a current or pass-thru throttle.

There is room to add one more item to the PAS menu to set the autoPAS ramp rate in addition to the power. Uh oh time for a prelim7?

Hello Justin,
I have allready Wgain to 5.
with 5KW on your throttle you have to set ramp up to minimum 1 sec.

with a nice responsive throttle (0,1sec) u cant pedal gentle.
sure, while pedaling theres no problem...but when you stop and start again you will get a kick (till the pas max limit comes in play)
the kick is about 0,1-0,3 seconds and its not funny
beside that a ramp up time for PAS under "1 sec" is not nice to use, because the bike accelerate to fast to pedal "human Power" with it.
with 1 sec uptime on PAS it is really perfekt. no need for a Torque Sensor
....but the throttle is a pain waiting for power.....

"we" just need that the throttle override starts where the PAS (voltage) IS....throttle override AutoPAS without starting the ramp Up time again from 0,0v

 

[Luge_a_foin]

... I will start again like this.
Nevertheless just to be clear I didn't use the same definition as yours in my posts for Open Loop and Close Loop ... Actually I've never tried any Current modes or others.

Thanks for the update - that makes the situation much clearer. Please post back if problems persist for the basic setup in 'Round Two'.

So I've went for it and it still doesn't work. Here is what I've done :
- Flash the latest CA firmware (v6),
- I've not modified any settings (excepted to disable the torque sensor which was by default at Thun BB in this new firmware),
Then test the throttle on the stand:
-The CA throttle I/O are correct on the debug display but I cannot reach full speed,
-so I've used a meter to check the CA Outputs. When throttling the meter fits with CA display until around 2.2V and after the meter is stuck to this value even if the throttle outputs displayed on the CA goes until 3.65V. Even if I do not have an oscilloscope I would be surprised it is due to a mean value as this mean value does not change as the displayed voltage on the CA increases (nevertheless it's still possible).
-Then I've added a reistor of 3.3 k (the only one I had) between the controller GND & +5V. Just in case the controller would measure some current draw (this would be weird as the throttle would not be completely inhibited). But it hasn't change anything.
As a reminder my controller is a Lyen Sensorless 12 mosfets. But it works with the throttle directly plug in.
I think I'm running out of ideas for this time being. Could this come from a too important current draw the CA is not able to supply? Unfortunately I do not have any meter to measure it.

 

[justin_le]

I think I'm running out of ideas for this time being. Could this come from a too important current draw the CA is not able to supply? Unfortunately I do not have any meter to measure it.

Hi Louis, could you do me a favour and use the multimeter to measure the resistance between your controller throttle signal and ground? Most controllers have about 10-20kOhm of impedance here, but it sounds like yours might be more like 1kOhm or less. The CA's output voltage is reasonably stiff but it does go through a 270ohm resistor (R18) that can ultimately limit things if you draw too much current:

 

[teklektik]

So I've went for it and it still doesn't work. Here is what I've done :
- Flash the latest CA firmware (v6),
- I've not modified any settings (excepted to disable the torque sensor which was by default at Thun BB in this new firmware),

You do not mention it specifically, so: Did you set Thrl->MinInput, Thrl->MaxInput, and Thrl->FaultVolt according to the Guide? Apologies if I was unclear on this point, but we want to follow the setup procedure...

Then test the throttle on the stand:
-The CA throttle I/O are correct on the debug display but I cannot reach full speed,
-so I've used a meter to check the CA Outputs. When throttling the meter fits with CA display until around 2.2V and after the meter is stuck to this value even if the throttle outputs displayed on the CA goes until 3.65V.

The 2.2v is wrong, but I don't inderstand why the displayed output only goes to 3.65v instead of 4.99v as in the setup procedure of the Guide. Did you adjust (ThrO->MinOut and ThrO->MaxOut) = (0.0v, 4.99v)?

What should be happening is that when you set Wide Open Throttle, the CA should display OUT=4.99v. The ThrO signal should follow this. I would recommend testing the CA ThrO level open-circuit (disconnected from the controller) and verify that it tracks the displayed OUT voltage 0-4.99v (step 1/5/2 on page 17). Then reconnect OUT to the controller and see if the voltage is unchanged or pulled down by a low controller input impedence.

Ahh - almost simultaneous post by Justin above looking at the same possible problem...

 

[Luge_a_foin]

I think I'm running out of ideas for this time being. Could this come from a too important current draw the CA is not able to supply? Unfortunately I do not have any meter to measure it.

Hi Louis, could you do me a favour and use the multimeter to measure the resistance between your controller throttle signal and ground? Most controllers have about 10-20kOhm of impedance here, but it sounds like yours might be more like 1kOhm or less. The CA's output voltage is reasonably stiff but it does go through a 270ohm resistor (R18) that can ultimately limit things if you draw too much current:

Hi,
5.45 kOhms. What do you think of it? It's a bit between your two ranges.
Tekle, here it's the night so I may have to wait until tomorrow to run my bike again...

 

[MattyCiii]

Suggestions please - firmware upgrade problem!

I tried to upgrade my CAv3 from Prelim5 to Prelim6. I hooked up the wire, fired up the FW Update tool v1.1, and everything went as expected. The firmware uploaded and tested satisfactory - the FW tool turned green. Then I noticed I uploaded Beta20 instead of Prelim6!. No problem, I thought... but yes, problem. With Beta20 loaded in this particular CAv3, I can't connect to the FW updater to change the firmware to any FW version! I tried using the 1.0 tool and 1.1.

Some more details...
1. When the FW tool says "please cycle power on the device" and I do just that, the CA screen lights up, and nothing happens no matter how long I wait. I can power cycle a couple times, same result. The moment the FW update activity on the FW tool times out (30 seconds), I can cycle power on the CA (still tethered to the minijack to USB cable) and it will boot up. I get the "version" screen, then I get the standard functions of the CA.
2. I hooked my analogger to the device, it's putting out data over serial just fine/as expected.
3. I have several CAv3's. I was actually upgrading two of them when this happened. I actually loaded FW version Beta20 onto both of them before realizing my mistake. One of them took to the upgrade from Beta20 --> Prelim6 just fine, the other has a problem. Identical gear, different results.

Any thoughts or suggestions?

 

[teklektik]

... the CA screen lights up, and nothing happens no matter how long I wait.

Sent PM with older Uploader that I use (v2012.10.16).
Operation needs a couple of extra button pushes, but it seems to have more relaxed timing requirements.

 

[MattyCiii]

... the CA screen lights up, and nothing happens no matter how long I wait.

Sent PM with older Uploader that I use (v2012.10.16).
Operation needs a couple of extra button pushes, but it seems to have more relaxed timing requirements.

THANKS Teklektik!
It worked great... I'm now on version Prelim6 on both CAv3's.

I'll be keeping that older FW update utility in my toolkit for certain.

 

[izeman]

... I'll be keeping that older FW update utility in my toolkit for certain.

teklektik's version is the only one working for me. ever. maybe ppl in charge of the update tool may look into it, and redesign timing as it was implemented in this very first version.

 

[izeman]

... At the moment, since the auto PAS feature is using the power feedback loop of the CA, you can make it engage more slowly when pedaling by setting the WGain value really low (like 5 or 10), and then leave your output ramp rates fast for the good throttle response. So then what limits the ramping when you pedal is not the overall throttle ramp rate but the power limit feedback loop. But this won't work if you have configured your throttle as a power throttle as well, it would have to be a current or pass-thru throttle.

my throttle is setup as pass-through and a tested WGain of 5 and 10 (was 50 before). there was absolutely NO difference how PAS engaged.

 

[Luge_a_foin]

You do not mention it specifically, so: Did you set Thrl->MinInput, Thrl->MaxInput, and Thrl->FaultVolt according to the Guide? Apologies if I was unclear on this point, but we want to follow the setup procedure...
The 2.2v is wrong, but I don't inderstand why the displayed output only goes to 3.65v instead of 4.99v as in the setup procedure of the Guide. Did you adjust (ThrO->MinOut and ThrO->MaxOut) = (0.0v, 4.99v)?

What should be happening is that when you set Wide Open Throttle, the CA should display OUT=4.99v. The ThrO signal should follow this. I would recommend testing the CA ThrO level open-circuit (disconnected from the controller) and verify that it tracks the displayed OUT voltage 0-4.99v (step 1/5/2 on page 17). Then reconnect OUT to the controller and see if the voltage is unchanged or pulled down by a low controller input impedence.

Hi,
I followed the process strictly. In full open loop (disconnected from the controller) the CA output is correct (range 0 -> 4.99V). But when I plug the controller I cannot go to more than 2.1V. The measured resistance of the controller between Gnd & Thr Input is 5.3kOhm. Could this fit with the CA limit to 2.1V? What could be the additional tests to confirm that? And what would be the possible solution? To replace a resistor?
Thanks for your help.