Cycle Analyst V3 preview and first beta release

Get real world experience and user feedback on the electric bicycle products.
User avatar
NeilP   10 GW

10 GW
Posts: 4983
Joined: Nov 27 2010 3:18pm
Location: 49.17303, -2.05354 Jersey, Europe
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by NeilP » Aug 07 2014 2:35am

Annoying behavior when entering setup and then leaving setup...the trip statistics are automatically reset.

I suppose for a fully set up bike, not an issue.. but bloody annoying when trying to setup the unit and during riding around and tuning settings. Go in to setup to change a parameter, and exit, all trip stats are back to zero

At the end of the ride you have no idea of miles covered/ah used etc etc.

EDIT Possibly solved with firmware re flash.

Also found 12v 5 amp psu so will swap the 15 volt one out for this one
:) Only Irish coffee provides in a glass all four essential food groups: alcohol, caffeine, sugar, and fat :)

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Aug 07 2014 7:49am

Neil-
It looks like the higher Vshutdown was not successful for you. This should work and you should see the 'LOW V' message after the save. I looked at the spec for your MW and see that its rated for 120v DC input. You are running it low - which should be fine - this is the standard wall-wart trick. But - I'm thinking that the 40v I suggested above is not high enough and is still below the dropout voltage.

Try temporarily running the converter off your balance leads and keep jacking up the voltage until it runs. This is probably a bit different than the dropout voltage but good enough. Set your Pref->Vshutdown between that and your Batt->VltCutoff (higher is better). This should ensure the converter will be running when when Vex crosses the threshold instead of my earlier wild ass guess of 40v.

Got to run now - good luck!
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
NeilP   10 GW

10 GW
Posts: 4983
Joined: Nov 27 2010 3:18pm
Location: 49.17303, -2.05354 Jersey, Europe
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by NeilP » Aug 07 2014 2:53pm

Running a12volt 5a laptop supply now.
Set low voltage shut down to 55 and all appears good
Will do a few test runs tomorrow
:) Only Irish coffee provides in a glass all four essential food groups: alcohol, caffeine, sugar, and fat :)

User avatar
NeilP   10 GW

10 GW
Posts: 4983
Joined: Nov 27 2010 3:18pm
Location: 49.17303, -2.05354 Jersey, Europe
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by NeilP » Aug 08 2014 6:06am

regarding setting of PS gain etc as per the UUG
D2 Speed gain adjustments
Follow this procedure:
1. Set IntSGain to 1 and DSGain to zero
2. Calculate initial PSGain = 6 x ( ThrO->MaxOut - ThrO->MinOut ) / ( SLim->MaxSpeed ).
3. Increase PSGain to the point where a single modest overshoot occurs before the speed settles.
If no setting produces overshoot, this procedure is not appropriate (bike power too low). However, in such
cases some success has been reported by setting PSGain to the initial value from step (2) and skipping to
step (4) - this is a largely untested and experimental strategy for low power tuning.
4. Reduce PSGain to 1/2 of its value.
5. Increase IntSGain until oscillations die out with no more than 1 or 2 additional upswings after the initial
overshoot.
6. Increase DSGain until settled operation is just jittery (power feels rough or ratty), then reduce 30%.
Are we expecting the speed to settle to the SLim->MaxSpeed setting. i used as suggested 15mph.

I see an overshoot and a settling, but always at a higher speed that set by SLim->MaxSpeed.
the higher I set PSGain the closer it gets to SLim->MaxSpeed, but never reaches it.

is this normal for this procedure?
:) Only Irish coffee provides in a glass all four essential food groups: alcohol, caffeine, sugar, and fat :)

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Aug 08 2014 7:05am

NeilP wrote:Running a12volt 5a laptop supply now.
Set low voltage shut down to 55 and all appears good
Very good news!

The Guide clearly will need some enhancement or even restructuring to better explain shutdown considerations/settings when using a DC/DC converter for CA power. As always - thanks for the posts!

BTW - I did a little digging and found a schematic for an EB206 controller which shows a 100uf cap on the input of the 5v regulator which is powered from the controller 'ignition' wire. This is not unusual practice for LM317 regulators and it's reasonable to assume the same or similar is found on your particular controller (presumably an EB 3xx). Since your MW supply is actually a power supply, not a converter, there is a bridge on the front end that would prevent the input cap voltage from appearing on the input. So - based on the reasonable strategy of powering the converter and 'ignition' wire in parallel from the bike main switch, I think the Pref->Vshutdown issue derived from the controller regulator cap - not the MW cap - holding up the Vex voltage.
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Aug 08 2014 8:21am

NeilP wrote:Are we expecting the speed to settle to the SLim->MaxSpeed setting. i used as suggested 15mph.

I see an overshoot and a settling, but always at a higher speed that set by SLim->MaxSpeed.
the higher I set PSGain the closer it gets to SLim->MaxSpeed, but never reaches it.

is this normal for this procedure?
Ya - it can be.

In a nutshell, there IS no standard procedure for tuning PID controllers that always works - it's sort of a Black Art. Frankly, Justin is pretty good at this - I'm more at the 'apprentice' skill level. The Guide procedure is an attempt to help get you in the neighborhood of a 'Good' setting, so anything you get that works is 'correct' :D.

Here's what's happening:
You can look at PSGain, IntSGain, and DSGain as addressing the Present, Past, and Future speed errors respectively.
  • PSGain is addressing the error between Actual and Desired MaxSpeed NOW and tries to fix things immediately in one shot - but if it tries too hard, it will overshoot (eg tune for running on the flat with a specific load, PSGain will work just fine, but change the terrain and load and the setting starts to work poorly).
  • IntSGain adds up all past errors and keeps growing until the correction has the desired effect - sooner or later it should sneak up on a big enough adjustment to fix the issue but it might take awhile and the fix may be slow in coming (eg you are climbing a hill and need more gain than on the flat, so IntSGain can keep increasing the correction when the initial try isn't enough).
  • DSGain looks ahead and tries to avoid correcting too much so that there will not be future error (overshoot) - if it tries too hard the performance is rough and ratty as the PID controller makes lots of nervous adjustments trying not to make a future mistake. (eg big power bikes can overshoot on hard acceleration, so the throttle may need to be eased - where a wimpy bike could tolerate WOT with no ill effects)
Wikipedia has a discussion of PID controllers that you may find helpful. Here's an image from that article that shows how the various Gain settings amplify the correction for the three different error types and how that discussion relates to the V3:
CAV3_SpeedPidControllerOview.png
The 3 Gain settings control amplification of the error corrections before they are summed to the final throttle 'fix'
CAV3_SpeedPidControllerOview.png (22.53 KiB) Viewed 2961 times
So - you may be seeing things a little high because either:
  1. there is very little IntSGain and you are relying entirely on PSGain for correction (this is the case in step 2). Increasing PSGain increases the correction for speed error by reducing the throttle more, bringing it more closely to the desired target value. But relying on this single adjustment without the Past or Future components leaves you open for over/undershoot because of too much or too little gain for different situations. So we back off on the Present a bit and throw in some past and future view (IntSGain, DSGain) - ('crap, this adjustment isn't working - MORE CORRECTION!", or "Not too much now! YOU'RE GOING TO GO TOO FAST!")
  2. you are relying on IntSGain which is accumulating errors in the past so by the time it gets things corrected the speed is a bit high (this might be in later tweaking steps). Increasing PSGain is bringing you closer to the target value because you are adding more 'NOW' into the equation and getting the speed fixed sooner than if you rely more on IntSGain to build up sufficient 'Whoops - fix a little more, whoops - fix even more...'.
Hope that helps a bit - it's a slippery matter with more than one 'correct setting' although some might be better than others. For instance, in a recent post about his new web trip analyzer, Justin pointed out that looking at the Analogger data from his Maker Faire trip last year, it was clear the PID controller was oscillating power quite a bit - even though the bike rode smoothly. This was probably wasting a bit of power, but the setting was acceptable from a performance and rideability perspective. So - if Justin doesn't always nail these adjustments, don't feel bad if you find them a bit elusive as well.... :D
Last edited by teklektik on Aug 08 2014 9:10am, edited 2 times in total.
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
NeilP   10 GW

10 GW
Posts: 4983
Joined: Nov 27 2010 3:18pm
Location: 49.17303, -2.05354 Jersey, Europe
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by NeilP » Aug 08 2014 8:55am

OK, thanks that is all good to hear,

I have now been called in to work, so don't know if I'll be back at home today or tomorrow or the next day. Will do more again when I am back at home
:) Only Irish coffee provides in a glass all four essential food groups: alcohol, caffeine, sugar, and fat :)

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Aug 08 2014 11:20am

AN IMPORTANT CORRECTION to UUG 3-0i Shunt Calibration Procedure

NeilP has discovered a significant error in the Guide procedure for instrumented shunt calibration. There was a transcription error that reversed the iCharger polarity in steps A.4.4 and A.4.5.

This is the corrected text:
CaV3_UUG_3-0p6-i-pdf_CalibrationProcedurCorrection.png
CaV3_UUG_3-0p6-i-pdf_CalibrationProcedurCorrection.png (54.51 KiB) Viewed 2956 times
This revision will appear in the next Guide release (guess it better be soon... :D)

Posting Guide errors gets things fixed up and directly assists future V3 users. Very much appreciated.
Thanks Neil!
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
NeilP   10 GW

10 GW
Posts: 4983
Joined: Nov 27 2010 3:18pm
Location: 49.17303, -2.05354 Jersey, Europe
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by NeilP » Aug 10 2014 11:34am

Back to the subject of 10k NTC thermistors.

Yes, I have seen the previous links...but...I am already compiling a big order from my favourite supplier...RS Components... so wanted to tag on some thermistors too.

So could someone check my reading of the spec of this part EC95F103WS427

or any of these others any better ?

http://export.rsdelivers.com/product/ep ... 62780.aspx
http://docs-europe.electrocomponents.co ... ef3fe6.pdf

http://export.rsdelivers.com/product/ep ... 88536.aspx
http://export.rsdelivers.com/product/ep ... 91892.aspx


Thanks
:) Only Irish coffee provides in a glass all four essential food groups: alcohol, caffeine, sugar, and fat :)

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Aug 10 2014 7:31pm

NeilP wrote:Back to the subject of 10k NTC thermistors.
...
So could someone check...
  1. this part EC95F103WS427
    • EC95F103WS427
      Beta 25/100 = 3981K
      thermal time constant = 10s
      tolerance +/-0.9%
  2. http://export.rsdelivers.com/product/ep ... 62780.aspx
    http://docs-europe.electrocomponents.co ... ef3fe6.pdf
    • NTC Epoxy Bead Thermistor, Leaded, 10K
      B57863S103F40
      Beta 25/100 = 3988K
      thermal time constant = 15s
  3. http://export.rsdelivers.com/product/ep ... 88536.aspx
    http://export.rsdelivers.com/product/ep ... 91892.aspx
    • Miniature NTC S861 thermistor,10k
      B57861S0103J40
      Beta 25/100 = 3988K
      thermal time constant = 15s
      tolerance = 5%
All appear workable and are within reasonable range of the target Beta of 3900K (that is - between 3800K and 4000K is okay).

That said, your first choice (A) has a teensy better Beta (25/100 degC) and appears to have the best tolerance as well as the fastest response time.
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
NeilP   10 GW

10 GW
Posts: 4983
Joined: Nov 27 2010 3:18pm
Location: 49.17303, -2.05354 Jersey, Europe
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by NeilP » Aug 11 2014 12:35am

Thanks.

I wanted to make sure I was reading correctly before purchase.
:) Only Irish coffee provides in a glass all four essential food groups: alcohol, caffeine, sugar, and fat :)

sacko   10 kW

10 kW
Posts: 826
Joined: Jul 04 2012 2:15am
Location: Kent, United Kingdom

Re: Cycle Analyst V3 preview and first beta release

Post by sacko » Aug 11 2014 3:52am


User avatar
NeilP   10 GW

10 GW
Posts: 4983
Joined: Nov 27 2010 3:18pm
Location: 49.17303, -2.05354 Jersey, Europe
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by NeilP » Aug 11 2014 4:32am

Thanks, but since I already have an order going in to RS Export, and they do free shipping to the Channel islands..was just going to tag these on to the order.
:) Only Irish coffee provides in a glass all four essential food groups: alcohol, caffeine, sugar, and fat :)

User avatar
NeilP   10 GW

10 GW
Posts: 4983
Joined: Nov 27 2010 3:18pm
Location: 49.17303, -2.05354 Jersey, Europe
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by NeilP » Aug 12 2014 2:05am

Is it too late / impossible to have the Aux Min and Aux Max voltages as Preset variables rather than Global.

I had been trying to set up my presets and I just can't get it as I want with global aux limits.
:) Only Irish coffee provides in a glass all four essential food groups: alcohol, caffeine, sugar, and fat :)

User avatar
Merlin   1 MW

1 MW
Posts: 1847
Joined: Jan 29 2013 10:00am
Location: europe

Re: Cycle Analyst V3 preview and first beta release

Post by Merlin » Aug 12 2014 8:42am

found another "interessting" thing.

AutoPAS - 1000w
Throttle- Current (2500w)

PAS works 100% as expected

start with a bit throttle, start Pedaling, release throttle and still pedal.
Bike acclerates with alot more than 1000w for 1-2 Seconds + AFTER stop to pedal.

i could make alot of surprising things in my combination of Motor,Controller & CA.
did we have someone here who has a CA Datalogger? would we see what happens there?

User avatar
mrbill   10 kW

10 kW
Posts: 512
Joined: Jun 10 2008 6:47pm
Location: Silicon Valley, California
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by mrbill » Aug 16 2014 3:02pm

Hi Folks:

I'm looking for a solution to a speed limit control tuning problem when my CAV3 is in street-legal mode (750 watts, 20mph).

When I initially set up my CAV3 I found the speed gain parameters that resulted in good behavior as speed approached the limit, then was pegged at the limit. Rapid increase to the limit, followed by minimal overshoot with minimal pumping or oscillation while staying within 0.1 mph of the limit (flat road, no wind).

SLim->MaxSpeed = 20.0 mph
SLim->IntSGain = 200
SLim->PSGain = 0.85 v/mph
SLim->DSGain = 200

ThrO->UpRate = 3.5 V/s
ThrO->DownRate = 40.0 V/s
ThrO->FastRate = 5.0 V/s
ThrO->FastThrsh = 0.0 Amps

PLim->MaxCurrent = 45 Amps
PLim->AGain = 50
PLim-> MaxPower = 750 Watts
PLim->WGain = 50

The system is a crank-drive limited by the CAV3 to 750 watts input (but capable of about 1300 watts input). Speed is limited by the CAV3 to 20 mph. Cruising at 20mph on my bike draws from the battery something in the neighborhood of 200-250 watts.

The remaining problem occurs when my speed exceeds the limit while the motor system is active but idle due to speed limit being exceeded, either from pedaling or from coasting down a hill. As my speed crosses the limit from above to below the speed limit, the overshoot can be as much as 4mph or more (e.g. downhill followed by uphill), the bike slowing to 16mph or less before the motor spools up and attempts to regain 20mph.

I have tried many combinations of increasing or decreasing the gain parameters, and while some combinations tighten this "overshoot from above", in particular increasing IntSGain, I could not find a setting that did not compromise performance when approaching the limit from below or while cruising in steady state.

In general I found that I could change IntSGain and DSGain together, e.g. both to 400 or both to 100 and get similar performance but with slightly more oscillation or over-damped response when approaching the limit from below, but 200 seemed to be the best compromise. Changing IntSGain and DSGain in opposition, i.e. increasing only one parameter or increasing one while decreasing the other resulted in poorer performance. Increasing PSGain had the throttle pumping more while cruising at the limit, and decreasing PSGain introduced low-frequency oscillation about the limit speed while cruising.

Anyone have any suggestions how I might tighten the limit overshoot from above while maintaining my current damping when approaching the limit from below? Thanks.
Bill Bushnell
http://mrbill.homeip.net/
Bike build thread can be seen here.

User avatar
izeman   10 GW

10 GW
Posts: 4929
Joined: Jun 21 2011 8:25am
Location: vienna, austria
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by izeman » Aug 17 2014 6:14pm

i got the 'power up bug' as well. sometimes my ca boots normally, sometimes nothing happens, and sometimes i can see the backlight flash once.
if i remember correctly i read about a work around or alternative firmware or something similar. as i didn't have those problems at the time being i just overflew that.
i now reread all posts of this topic back to dec 2013 but couldn't find it. searching the forum for 'boot, power, ca, ...' gives endless results.
could anyone please enlight me? ;)

User avatar
teklektik   10 GW

10 GW
Posts: 4020
Joined: Mar 26 2011 1:15pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Aug 17 2014 6:46pm

izeman wrote:i now reread all posts of this topic back to dec 2013 but couldn't find it. searching the forum for 'boot, power, ca, ...' gives endless results.
could anyone please enlight me? ;)
Please see the Post Index (section 2) of the Guide download post.

Pretty much all Justin's posts, posts with downloadable files, or posts of other technical interest are indexed there.
Last edited by teklektik on Aug 18 2014 10:13am, edited 1 time in total.
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
izeman   10 GW

10 GW
Posts: 4929
Joined: Jun 21 2011 8:25am
Location: vienna, austria
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by izeman » Aug 18 2014 9:18am

teklektik wrote:
izeman wrote:i now reread all posts of this topic back to dec 2013 but couldn't find it. searching the forum for 'boot, power, ca, ...' gives endless results.
could anyone please enlight me? ;)
Please see the Post Index (section 2) of the Guide download post.

Pretty much every post by Justin, with a downloadable CA file, or of technical interest is indexed there.
man i'm getting old. in my memory justin post was month ago, but it was only 2 weeks, and i think i stopped searching 2-3 posts before justin's post :( thanks for your help. i will try out and report back if this solved my issues.

User avatar
snellemin   10 kW

10 kW
Posts: 837
Joined: Apr 03 2011 12:24am
Location: Spring TX

Re: Cycle Analyst V3 preview and first beta release

Post by snellemin » Aug 18 2014 11:03am

Anybody got the 2.3 firmware? I screwed up mine by accident. I had flashed my CA3 to fix the boot up issue and was just curious to see if I could read the settings off the CA 2.3. Well I didn't close the firmware application and when I turned on my bike, the firmware automatically loaded up. Stupid of me to mess around with my bikes on an empty stomach.
Silent speed Racer

User avatar
izeman   10 GW

10 GW
Posts: 4929
Joined: Jun 21 2011 8:25am
Location: vienna, austria
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by izeman » Aug 18 2014 11:51am

you mean v3b23? http://endless-sphere.com/forums/viewto ... 22#p718722
all fw downloadable are in the link teklektik posted just 2 posts above. i'm not aware of and 2.3 fw so i hope this is the one you asked for.

User avatar
snellemin   10 kW

10 kW
Posts: 837
Joined: Apr 03 2011 12:24am
Location: Spring TX

Re: Cycle Analyst V3 preview and first beta release

Post by snellemin » Aug 18 2014 12:13pm

No, I need the firmware for the V2 CA. I couldn't find a link anywhere to one. I've send an email to Grin and no answer yet. I figure they are busy during the summer season.
Silent speed Racer

User avatar
izeman   10 GW

10 GW
Posts: 4929
Joined: Jun 21 2011 8:25am
Location: vienna, austria
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by izeman » Aug 18 2014 1:24pm

you can program the v2? i thought only grin can do that ...

User avatar
izeman   10 GW

10 GW
Posts: 4929
Joined: Jun 21 2011 8:25am
Location: vienna, austria
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by izeman » Aug 18 2014 1:43pm

so this is my experience with flashing the "bootfix hex": i only had the old 1.0 firmware updater. installing the bootfix hex worked fine. it rebooted, and i set it to "1". power cycled several times and it always rebooted fine with the bootfix.
then i tried to reflash pre6_noeprom but that didn't work as the updater couldn't find any CA. so i tried updater 1.2. same issue here. flashing pre5 worked. but after rebooting it it did hang with a backlit screen again. trying to flash ANY hex file didn't work as i couldn't find a CA with updater 1.2.
so i went back to updater 1.0 and was able to load CA3B19. after that i could flash "bootfix" with updater 1.2 and then flash "pre6" with the same tool.
then i rewrote my settings with ca3setuptool.
the result is quite strange. everything looks fine on the first view. but i could NOT change the "aux pot" setting. press/holding the button to enter it's setup menu just does nothing. the batteries internal resistance value also is WAY off. it shows 0.8ohm as where it's really 0.02ohm.
riding the pain was a PITA. throttle control was SUPER jerky. i thought it would be a good idea to resetup the throttle low/high values etc. all fine.
setting up 3 presets (600w, 2200w and 1000w) worked in principal, but power application was (as i said) from bad to VERY bad. intermittent power and power loss. the lower the powersetting the worse it was. but even removing ALL limiting factors (checking with the limit screen on the CA as well) gave a jerky throttle response it not WOT.
so i think there as some settings messed up and corrupt. what would i need to do to TOTALLY RESET the CA and start from scratch?
i hope this long thread helps others that may run into the same issues - mainly not being able to access the CA with various versions of the updater.

User avatar
snellemin   10 kW

10 kW
Posts: 837
Joined: Apr 03 2011 12:24am
Location: Spring TX

Re: Cycle Analyst V3 preview and first beta release

Post by snellemin » Aug 18 2014 2:14pm

Yea, I found out the hard way that the V2 can be reflashed.

I had retry the boot fix a few times before it went through(where you can enter the numerical value). After that everything went fine with my CA. All my settings remained the same after the CA3_Prelim6_NoEeprom.hex was flashed into it.
Silent speed Racer

Post Reply