Cycle Analyst V3 preview and first beta release

izeman said:

worked perfectly fine.

Click to expand...

Good news! Thanks for posting back w/results.

izeman said:

though i already adjusted my throttle handling to my mid drive (always applying a little throttle to put strain on the chain and drive train), i must say it would be fine to have this applied all the time.
i can't remember how far we discussed this feature suggestion,...

Click to expand...

Yep - you may remember you did some successful experiments for us about this a ways back. As a result, it's on the list as:

4638 - Add mid-drive/gear motor pre-tensioner for torque-throttle controllers

So, your idea is formally written up, it just needs to float to the top among other tasks competing for CA resources and developer time...

I know. I just wanted to share my idea of a possible solution on how to acchieve it.

I recently received a shiny new ca3-dps but struggling with it....

it worked ok for one ride. then I tried connecting to pc using my usb cable from my phaserunner. the connection between the pc and ca was pretty unstable. Since then, the ca now reverts to throttle output = rc pulse instead of voltage, which means my phaserunner goes into fault as soon as it gets power.

I changed back to throttle out = voltage and do a power cycle and its changed again. often it comes up on screen "setup changed!" without me pressing anything. I then tried to change the firmware but a couple of % in to the update and the connection was lost. Now I have a CA with no firmware and nothing on screen when turned on.

I have tried to sort the firmware by ticking the box to change all settings to default but it never gets above 7% before it says there is no connection.

I am using a linx windows 10 tablet, which I also successfully use with the phaserunner. any ideas please? dying to go for a ride.

Maybe something to do with the baud rate on the COM port? currently set to 9600 on windows. [edit- the setting are exactly whats stated in the cycle analyst manual)

any help greatly appreciated!

handbaked said:

...
Since then, the ca now reverts to throttle output = rc pulse instead of voltage, ...
...
often it comes up on screen "setup changed!" without me pressing anything.

Click to expand...

Yes, this funky behavior is a problem with 3.1 that is addressed in a new 3.11 release coming very soon. This new firmware is available as a beta release as described in these posts (read both):

(3.11b2) https://endless-sphere.com/forums/viewtopic.php?f=4&t=37964&start=4050#p1385644
(3.11b3) https://endless-sphere.com/forums/viewtopic.php?f=4&t=37964&start=4100#p1389389

I recommend that you re-flash with 3.11b3 selecting the first checkbox to 'Reset to defaults' then reconfigure from scratch.
All should be well.

I have already tried that. It never gets beyond 7%, sometimes not even above 0% before it says it has no connection and the whole data log window goes red....

• 

izeman said:

next questions: while replacing the CA i wanted to use the "new" digital aux switch feature as well, which is not covered yet in the unofficial CA guide afaik.
...
...i haven't found a schematic. read something about diodes and stuff, but am unsure how to wire it exactly.

Click to expand...

teklektik said:

I pulled down the DigiAux HowTo document when we switched to Schottky diodes and haven't had the time to update the doc. But since this is the second request in a week or so, I'll get on this.

Click to expand...

A revised PDF that describes how to build your own Digital Aux controls is available at the end of the [strike]Un[/strike]official Guide download post.

izeman-
Hope this gets your project underway again!

handbaked said:

I have already tried that. It never gets beyond 7%, sometimes not even above 0% before it says it has no connection and the whole data log window goes red....

Click to expand...

I see you've also posted this same issue another thread.
One thread is enough, please.
Just sayin'...

Try another flash selecting the first checkbox to reset defaults, but start it with the CA turned OFF. Power up the CA after the download is initiated (the Setup Utility will prompt you to power-cycle the CA because it can't find it). This will bypass all installed firmware and go directly to the bootloader.

If this doesn't work you have a driver/comms issue with your Linux or the Linux version of Setup Utility and will need to email Grin for assistance.

Sorry dude, I thought it was better to keep the question here. Should have deleted the other post.

I had already tried that too re: power off. It says it finds bootloader, then stalls either at 0% or something less than 10%. Red screen communications error.

Guessing this is one for grin. Cheers for the super quick replies.

teklektik said:

Hope this gets your project underway again!

Click to expand...

sure. great document, that even i can follow that's a really impressive, mighty, well thought out approach i must say.
i just will use the on/off/on momentary buttons, so this is my setup:

methods said:

gronph said:

are there any news with the Bluetooth/Android thig?

Click to expand...

I have no idea who is working on it or what it is.

I have an implementation of BlueTooth that I developed for a Google team (around an Arduino) that provides a framework for an Android Phone. It is implemented using MIT APP Inventor 2. Data is passed bi-directionally in ASCII between the Android Phone and the Target. There is a heartbeat. You can send commands from the Android and collect state data with verification (CRC).

Key is keeping the BlueTooth alive.... at least in my implementation... as I specified the $10 jank dongles you can buy on Amazon.

Happy to share it if I can - MIT App Inventor 2 is a purely graphical language like LabView and the compiler is free to use. Anything that runs on Android can be ported over to the CA by a professional in half a day.

Once the data is moving back and forth... users (the community) can move legos around to make the front end look however they like. There are tons of tools for buttons and displays.

It requires little to no programming experience... but then... without understanding of programming paradigms most folks wont get too far with it.

-methods

Click to expand...

Justin announced that some month ago on the Grin Homepage http://www.ebikes.ca/news/CA31-release/:

Next up on the CA3 front: Android app support, updated documentation, and more video tutorials!

izeman said:

teklektik said:

Hope this gets your project underway again!

Click to expand...

sure. great document, that even i can follow that's a really impressive, mighty, well thought out approach i must say.

Click to expand...

hmmm. seems i just found out that there IS a difference between a shottky diode and a 1N4148. the later ain't work so it seems. i will need to get some shottkys.
switching on one direction works, but the other doesn't. the one direction (middle pin to diode end pin works), the other one (middle pin to GND doesn't work), no matter of diode orientation.

izeman said:

...there IS a difference between a shottky diode and a 1N4148.

Click to expand...

Yep. The original design used plain old silicon diodes like the 1N4148 which have a Vfwd of about 0.7V. But to expand the available analog range, we went with Schottkys that are a little finickier in several respects but have a Vfwd of about 0.4V. There was an accompanying firmware change way back when for the altered voltage thresholds.

That said, for that simple circuit you are building, you can plug in a 470 ohm resistor instead of the diode and it should work fine. You might have to fiddle a bit with the values depending on the precision of your parts but it's a perfectly good solution in that digital-only case where the circuit doesn't have to accommodate several different plug-in analog controls as does the standard Grin button assembly. You just want to get a voltage solidly between 0.22 and 0.65V -- we shoot for 0.43V. The R1 value isn't critical either - it's just there to bias the diode ON while drawing a reasonably small amount of current - we used a value that draws a bit under 1ma but that's certainly not a magic number. So - hack away with what you've got on the bench...

Apologies to you earlier downloaders, but I screwed up a diagram.
Just fixed and re-posted the DIY Aux doc.
Oops.

teklektik said:

That said, for that simple circuit you are building, you can plug in a 470 ohm resistor instead of the diode and it should work fine.

Click to expand...

i can confirm that this works perfectly fine. soldered in the 470Ohm resistor and all is good.
i like the feature to start at "level 1" and set it to zero % power. so you have an extra safety feature. turn on the bike, but nothing happens when you turn the throttle, you need to press the button to set power level 2 by pressing right or highest power setting by pressing left. 8)

I've been seeing some strange CA3.1 behavior with the CA showing Setting Changed! on restarting and also Slow Power Up with the CA not showing correct voltage. I've run the CA with the Phaserunner at both 36 and 72volts but while trying to configure the set up today the CA started acting up as if there might be a short. I'm plan to update the firmware when I have a chance, the first couple of tries I wasn't able to do it. I haven't seen Slow Power Up in the documentation. The battery voltage on my LiGo(s) is fine.

teklektik said:

handbaked said:

...
Since then, the ca now reverts to throttle output = rc pulse instead of voltage, ...
...
often it comes up on screen "setup changed!" without me pressing anything.

Click to expand...

Yes, this funky behavior is a problem with 3.1 that is addressed in a new 3.11 release coming very soon. This new firmware is available as a beta release as described in these posts (read both):

(3.11b2) https://endless-sphere.com/forums/viewtopic.php?f=4&t=37964&start=4050#p1385644
(3.11b3) https://endless-sphere.com/forums/viewtopic.php?f=4&t=37964&start=4100#p1389389

I recommend that you re-flash with 3.11b3 selecting the first checkbox to 'Reset to defaults' then reconfigure from scratch.
All should be well.

Click to expand...

v3.11 Released

Title says it all. 3.11b3 has been rebranded as 3.11 and is now the official production release that goes out with new CA3s. See Grin News: http://www.ebikes.ca/news/CA3-11-Crystalyte/

If you are running 3.1 it is strongly recommended that you upgrade to 3.11 to get some significant bug fixes and much improved PAS performance.

If you are on b3, then an upgrade will buy a few tweaks to the Release Notes, but it's really the same FW - so no compelling reason to jump on it if your plate is full just now...

See the Release Notes on the Setup Utility Help pop-up or check out these posts for the beta release history:
(3.11b2) https://endless-sphere.com/forums/viewtopic.php?f=4&t=37964&start=4050#p1385644
(3.11b3) https://endless-sphere.com/forums/viewtopic.php?f=4&t=37964&start=4100#p1389389

As always, many thanks to all our loyal beta testers and especially those who sleuthed out issues and posted so we could tinker up fixes. Much appreciated!

Just in time for the weekend

kevinscargobike said:

I've also noticed that the relationship between my human wattage and motor wattage being mismatched in the other direction under more normal usage, like, when I'm set for 7x and producing 100 Hwt, I might only see 500 watts as the power reading. Does that mean I have something configured incorrectly, or is it not necessarily intended to be a perfectly 1-1 kind of thing?

Click to expand...

I just upgraded to 3.11 and did a quick test ride down my hill and back up, and this seems to be fixed. I'd gotten used to seeing the wattage always being about half of what the multiplier should do with my Hwt, and it seemed to match much much better now. Send my thanks to whoever did nice things to the PAS code!

kevinscargobike said:

kevinscargobike said:

I've also noticed that the relationship between my human wattage and motor wattage being mismatched...

Click to expand...

I just upgraded to 3.11 and ... this seems to be fixed.

Click to expand...

Sorry your original post didn't get a response here in the thread, but your heads-up about the issue helped catch the cause during the 3.11 PAS upgrade. Thanks for posting about this!

(...and glad to hear the PAS changes are working out )

Unfortunately, on my way home up the hill yesterday it seemed to be back to giving me half of the multiplier:
https://www.dropbox.com/s/kmv2ucfa1afszul/IMG_0506.m4v?dl=0 (sorry, wrong orientation again)

The "soft-pedal surging" seemed to be a little worse since the update, so I was trying to eliminate other factors. I had set my pole count to 18, which I realized was a dumb value, so I changed it back to the default of 24. I also upped Assist Start Level to 65 watts from 35 watts. The second one seemed to make it a little better, but I'm still finding that I move the pedals just enough that the gears almost engage and then the motor spins for what feels like a full second. (Today I changed my down rate from 4 v/s to 6 v/s, and my PAS up rate from 5 v/s to 3v/s, which did seem to mitigate it, even though they're not necessarily solving anything)

For the surging, it feels like the high pole count of the Sempu might be a factor - like it's seeing that the crank has gone past 2-4 poles and even though the torque value is small, it's assuming that because there's cadence without torque, it should just assume a minimum torque value. On a cadence sensor with half the poles, maybe I wouldn't be seeing this?

kevinscargobike said:

Unfortunately, on my way home up the hill yesterday it seemed to be back to giving me half of the multiplier:

Click to expand...

Hmmm - that's interesting. Tracking down the exact cause of this last behavior may take a bit of digging. If possible, please PM me with a setup file and a couple of words to describe your motor/controller/net vehicle weight. I guess you are using a Sempu torque sensor.

kevinscargobike said:

I'm still finding that I move the pedals just enough that the gears almost engage and then the motor spins for what feels like a full second. (Today I changed my down rate from 4 v/s to 6 v/s, and my PAS up rate from 5 v/s to 3v/s, which did seem to mitigate it, even though they're not necessarily solving anything)

Click to expand...

A down rate of 6v/sec means it will still take 1/2 second for the power to drop from full to zero. Seeing the setup file and all the settings that are in play may shed some light on this. It's presently necessary for the cranks to make an initial full revolution before the CA switches to the STOP threshold so if you have widely separated values and a very low cadence startup threshold, then it is going to take a bit for the CA to remove assist if you just briefly and gently engage PAS. Ideally, these thresholds are set for a rapid transition across this cadence range when starting/stopping. You should have adjusted the Start/Stop values downward as per the Release Notes if you carried your old 3.1 settings forward when upgrading:

3.11 Release Notes said:

An accompanying change has been made to PAS->StrtThrsh and PAS->StopThrsh default values to reduce sensitivity to slow pedal motion (e.g. PAS starts too soon, stops too late). This issue may present after upgrading the firmware and inheriting settings from the previous version (i.e. upgrading without loading defaults). If so, reset the PAS Start/Stop thresholds to the new defaults (Start, Stop) = (0.30, 0.19) or reduce custom setting values by about one third.

Click to expand...

kevinscargobike said:

... like it's seeing that the crank has gone past 2-4 poles and even though the torque value is small, it's assuming that because there's cadence without torque, it should just assume a minimum torque value. On a cadence sensor with half the poles, maybe I wouldn't be seeing this?

Click to expand...

There is a good chance that the high pole count is having an effect on computations, but not for the reason you call out. There is no 'minimum torque' - if you experience this, then my first guess would be to check that the torque is zeroed. Remember, too, that the minimum 'Assist' rpm is 50 to help getaways, so if you have this set to start at a very slow cadence, there will be an apparent RPM jump as soon as PAS becomes active. This is different than the displayed RPM which is the actual cadence. Also - when you are doing time studies based on screen indicators, be sure to crank down the display averaging otherwise everything will seem very lethargic purely as an artifact of display smoothing.

Anyhow - will pursue this - thanks for the details.

I tried to get some numbers from your video but things are varying quite a bit and I don't know your Assist Start value so this is hard to evaluate. That said, it doesn't look very far off - not a factor of two. You indicated you were using Start Assist values in the 35HW - 65HW range. If you subtract the Start Assist value from the indicated HW and then apply the W/HW multipler you should get the PAS motor watts (assuming that your cadence is above 50RPM).

As an experiment, I encourage you to set the Assist Start to zero and see if the displayed results look more as you expect.

Yeah, sorry, the video isn’t a great way to see. The multiplier was at 10x (I waggled the knob a bit to show the value), my Hwt was 100ish and the motor wattage was 5-600ish. My start is at 65w now.

I’ll try to send a PM with all the info later today to work on the surginess, if I can sneak away from party prep, but here’s some quick stuff:

I was on a Boda Boda (65lb Bike?) with a Mac/Phaserunner/Grin 52v 16ah battery. I think my max was 1300 watts. The grade in that spot is probably 6-7%, with a throttle it will happily go 15+ mph there.

When I posted before, I had it set to 10x and the motor was doing the full 1300 watts when I pedaled hard up the short bit of 15% I have at the end of my climb, and it was awesome.

——
Edit to clarify: I have two different issues, the multiplier seeming to produce half of the expected output during continuous steady pedaling, and the issue where a small low torque pedal movement causes the motor to surge with a value way above what the multiplier should produce, usually following high power usage. Like I go fast up the hill, then gently change which pedal is down as I roll down the other side of the hill to my mailbox and the motor spins up.

kevinscargobike said:

The multiplier was at 10x ... my Hwt was 100ish and the motor wattage was 5-600ish. My start is at 65w now.

Click to expand...

Yep those were the 'ish' numbers I saw but now knowing the Start Assist for sure: we expect to see a motor power of
motorPower = (100HW-65HW) * 10W/HW = 350W
You are getting more than that - over 500W -- not 50% of expected.

This larger value would occur if your cadence was less than 50rpm causing you to get the larger power based on you true torque but at the 50rpm minimum. Here an improper number of poles might be giving a low rpm reading. You need to configure RPM as a custom view as well as HW to ensure the detected cadence is reasonable. (Frankly, the Electrical Stats screen might benefit from replacing the V and A fields with Watts...)

So - something may be broken in the FW, but the RPM thing is looking a little suspect at this point. Unfortunately, I cannot duplicate your problem either on the bench or on the road - both situations look spot on - we need more info about your configuration and a little snooping for the displayed RPM.

kevinscargobike said:

...the issue where a small low torque pedal movement causes the motor to surge with a value way above what the multiplier should produce, usually following high power usage. Like I go fast up the hill, then gently change which pedal is down as I roll down the other side of the hill to my mailbox and the motor spins up.

Click to expand...

Yes - that is a separate matter from a month ago that remains unresolved at this time. If this is the 'soft pedal surging' to which you referred, I misunderstood - sorry, but a link or something to a previously reported issue will help give context - it's hard to stitch separated posts together to understand what is being said...

teklektik said:

Yep those were the 'ish' numbers I saw but now knowing the Start Assist for sure: we expect to see a motor power of
motorPower = (100HW-65HW) * 10W/HW = 350W
You are getting more than that - over 500W -- not 50% of expected.

Click to expand...

Well, half of what *I* expected (it might be helpful to update the Assist Scale Factor tooltip to reflect that Assist Start Level is subtracted, I would never have guessed that. )

I only set the Assist Start Level higher to get rid of the ghost surging, I'll try setting it to zero.

I just grabbed my CA config, and I'll send it along. I'm sure I've set all sorts of things strangely.

kevinscargobike said:

(it might be helpful to update the Assist Scale Factor tooltip to reflect that Assist Start Level is subtracted, I would never have guessed that. )

Click to expand...

Yep - I've argued about the sign as well, but from a certain Human Power perspective it makes sense. Unfortunately, my outlook just doesn't go that way...

PAS->StrtLevel ToolTip said:

TorqPAS: Sets the threshold human power before proportional torque assist begins to be applied. For instance, if set to 100 watts then there will be no assist when pedaling lightly but proportional assist will begin when rider effort exceeds 100W. This value can be set negative, which has the effect of providing motor assist by turning the cranks even with no effort; additional proportion power increases from this baseline with additional effort.

Click to expand...