Cycle Analyst V3 preview and first beta release

Get real world experience and user feedback on the electric bicycle products.
gronph
1 mW
1 mW
Posts: 11
Joined: Dec 27, 2016 11:59 am

Re: Cycle Analyst V3 preview and first beta release

Post by gronph » Mar 20, 2018 8:46 am

teklektik wrote:
Mar 20, 2018 8:32 am
Since the Sempu is a supported device that can be selected by menu via the 'Device Sensor Type', all other PAS device settings are made automatically in that category as described in the Help File and Tool Tips. Although you can change these in the CA, there is normally no need to do so.
I have three different Sempu´s and all have different Voltage Scale ;)

User avatar
teklektik
1.21 GW
1.21 GW
Posts: 3724
Joined: Mar 26, 2011 1:15 pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Mar 20, 2018 11:45 am

My Bad. I thought you were looking at the new one offered by Grin, but I see that one gets a scale factor of 50Nm/V.
Oops!
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

kevinscargobike
1 mW
1 mW
Posts: 15
Joined: Nov 01, 2017 6:47 pm

Re: Cycle Analyst V3 preview and first beta release

Post by kevinscargobike » Mar 22, 2018 1:00 pm

With my new Sempu, I've noticed that I don't ever see an Hwt below 200 or 250, even when I soft pedal. Which way should I tweak the Nm/V to bring that number down to the 50-150 that I'm really putting out?

I've tried both lower and higher than 50Nm/V, and the reading seems too high either way, maybe I need to tweak the torque offset?

User avatar
teklektik
1.21 GW
1.21 GW
Posts: 3724
Joined: Mar 26, 2011 1:15 pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Mar 22, 2018 1:53 pm

kevinscargobike wrote:
Mar 22, 2018 1:00 pm
I've tried both lower and higher than 50Nm/V, and the reading seems too high either way, maybe I need to tweak the torque offset?
Exactly. You must zero the torque when you install it. This is used to set the no-torque voltage. Voltage higher than that value in multiplied by the scaling factor (50Nm/V in this case) to convert the voltage to torque. If the zero point is off, the subsequent calculation becomes invalid.

You also can only alter these other PAS Device settings (e.g. the 50Nm/V) in the CA itself. They will not propagate from the Setup Utility when you use a predefined device. You can see this by reading the CA settings after you try to adjust them in the SU. This is a weird quirk that came about because the CA actually works the defaults logic on boot-up. It just slams in new defaults since it doesn't know if you changed things in the SU or not. But, if you change things local to the CA, it knows what happened and preserves your changes. I know - a bit strange, but.... :D
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

User avatar
justin_le
10 MW
10 MW
Posts: 2093
Joined: May 28, 2007 3:27 am
Location: Vancouver
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by justin_le » Mar 25, 2018 5:40 pm

nieles wrote:
Mar 20, 2018 8:08 am
pretty simple. 1Kg.F is 9.80665N.M and 1V is 1000mV
This is not true. 1Kg force is 9.8 Newtons. It is not 9.8 Newton-meters.
step 1 -> 0.25V/98.0665N.M (subtituting NM for KgF)
step 2 ->250mV/98.0665N.M (substituting mV for V)
step 3 ->2.54929mV/N.M (divide both with 98.0665 to get mV per N.M)
The correct form of this requires an assumption on the crank length, and lets assume that's 170mm or 0.17m

step 1 -> 10kg * 9.8 = 98 Newtons
step 2 -> 98N * 0.17m = 16.6 Newton-meters
step 3 ->16.6 Nm / 0.25V = 66.4 Nm/V

In practice, if you set it to this value it will read human watts that are inflated, since any and all force on the cranks shows up as torque even if it's not torque producing. Just stand on the pedal at the bottom of the stroke (=no torque to the rear wheel) and you'll have a huge signal. With a Sempu there is another fudge factor to figure in just what percentage of the forces present on the spindle are actually torque producing, since only those contribute to mechanical power. Hence the default in the CA being a lower value of 50Nm/V, but perhaps even this is being generous.

The actual best guess approximation to accurately show human watts is just to pedal at what you feel like is a power that you can relate to from riding on an exercise bike with a power readout, and then adjust the scaling to that the human watts is roughly consistent with what you expect it to be.
Big Dummy Frame (yes This One, thanks ES!) with prototype 26" Grin all-axle front hub, Phaserunner controller, and 52V 11Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled

kevinscargobike
1 mW
1 mW
Posts: 15
Joined: Nov 01, 2017 6:47 pm

Re: Cycle Analyst V3 preview and first beta release

Post by kevinscargobike » Mar 26, 2018 11:59 am

I managed to open up the screen on the CA for zeroing out the torque sensor, I saw the current setting and the voltage being measured, but holding the right button didn't seem to put me into any kind of edit mode. What's the trick for that?

User avatar
teklektik
1.21 GW
1.21 GW
Posts: 3724
Joined: Mar 26, 2011 1:15 pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Mar 26, 2018 12:09 pm

PasD-ZeroTorq.png
From Setup Utility ToolTips and Help File
PasD-ZeroTorq.png (7.47 KiB) Viewed 776 times
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

kevinscargobike
1 mW
1 mW
Posts: 15
Joined: Nov 01, 2017 6:47 pm

Re: Cycle Analyst V3 preview and first beta release

Post by kevinscargobike » Mar 26, 2018 3:42 pm

Worked this time - maybe I was holding for too long last time? I kept getting 'OK' but not getting a change in value. Getting properly zeroed, and changing to 30Nm/V is giving me much more realistic numbers.

kevinscargobike
1 mW
1 mW
Posts: 15
Joined: Nov 01, 2017 6:47 pm

Re: Cycle Analyst V3 preview and first beta release

Post by kevinscargobike » Mar 26, 2018 4:57 pm

I think I'm going to need another piece of the manual screenshotted for me. How do I enable presets 2 & 3? In the setup utility they're just disabled, and I haven't stumbled across a way to enable them. Is this one of things that I need to do with the CA itself first?

User avatar
teklektik
1.21 GW
1.21 GW
Posts: 3724
Joined: Mar 26, 2011 1:15 pm
Location: CT, USA

Re: Cycle Analyst V3 preview and first beta release

Post by teklektik » Mar 27, 2018 8:38 am

kevinscargobike wrote:
Mar 26, 2018 4:57 pm
I think I'm going to need another piece of the manual screenshotted for me.
How do I enable presets 2 & 3?
In the setup utility they're just disabled, and I haven't stumbled across a way to enable them.
CA3-PresetCnt.png
PresetCnt ToolTip
CA3-PresetCnt.png (56.36 KiB) Viewed 721 times
Visit Grin Technologies at www.ebikes.ca
Build Thread: 2WD Yuba Mundo V4

redline2097
100 W
100 W
Posts: 142
Joined: Jun 18, 2015 1:43 am

Re: Cycle Analyst V3 preview and first beta release

Post by redline2097 » Mar 27, 2018 12:30 pm

justin_le wrote:
Mar 25, 2018 5:40 pm
In practice, if you set it to this value it will read human watts that are inflated, since any and all force on the cranks shows up as torque even if it's not torque producing. Just stand on the pedal at the bottom of the stroke (=no torque to the rear wheel) and you'll have a huge signal.
Is there any tweak to overcome this? When coming downhill standing on pedals motor throttles?

kevinscargobike
1 mW
1 mW
Posts: 15
Joined: Nov 01, 2017 6:47 pm

Re: Cycle Analyst V3 preview and first beta release

Post by kevinscargobike » Mar 27, 2018 4:26 pm

redline2097 wrote:
Mar 27, 2018 12:30 pm
Is there any tweak to overcome this? When coming downhill standing on pedals motor throttles?
I think that's why torque sensing is really always torque + cadence, since torque by itself is a false positive.

redline2097
100 W
100 W
Posts: 142
Joined: Jun 18, 2015 1:43 am

Re: Cycle Analyst V3 preview and first beta release

Post by redline2097 » Mar 28, 2018 2:49 am

kevinscargobike wrote:
Mar 27, 2018 4:26 pm
redline2097 wrote:
Mar 27, 2018 12:30 pm
Is there any tweak to overcome this? When coming downhill standing on pedals motor throttles?
I think that's why torque sensing is really always torque + cadence, since torque by itself is a false positive.
So that sempu 2.3 unit registerer cadence as well?

kevinscargobike
1 mW
1 mW
Posts: 15
Joined: Nov 01, 2017 6:47 pm

Re: Cycle Analyst V3 preview and first beta release

Post by kevinscargobike » Mar 28, 2018 2:03 pm

Yeah, it’s torque plus a 24 pole cadence sensor. I’ve only used torque sensing assist setups before, so I’m actually curious to try out cadence only with max power on the potentiometer and see how it compares.

Tordnado
10 mW
10 mW
Posts: 30
Joined: Nov 10, 2016 9:21 am

Re: Cycle Analyst V3 preview and first beta release

Post by Tordnado » Apr 10, 2018 12:45 am

teklektik wrote:
Jan 19, 2018 9:16 am
What is happening is that the CA has cranked ThO down to the minimum in a futile attempt to slow the bike going downhill and there is a huge error wound up in the speed PID controller. It takes a while to counter that large error - your reported delay.

You are entering the realm of PID tuning - which is not always a happy place to go...
SLim->IntSGain controls the rate at which the past error both accumulates and is reduced. You can try increasing this to make the 'past error correction' change more rapidly. Some adjustment to PSGain will likely be useful as well.

There is an explanation of how this works in the Unoffical Guide section "Appendix E. PID Controllers and Gain Parameters". Basically, you have the situation illustrated below and are getting hosed by a large "I" value. Because we can't stop the integral error term from maxing out on a long descent, one strategy is to make the associated IntSGain term as large as we can (without oscillation) so that the accumulated error is reduced quickly. There's a tuning procedure there that in general may help you get the speed limiting as responsive as possible without oscillating.


speedPid5_small.png
Now that white thing called snow has disappeared so that I can start riding my bike again and trying out the above instructions. I changed the SLim->IntSGain all the way up but it is still fairly bad and also starts oscillating a lot (which is expected) but having that set to max and still getting too much delay after downhills tells me no matter what I do I can't get it right!? I hope I'm wrong.

My old no name chinese controller was not having this problem. It had really smooth power control without an extra controller like CA. The Infeon controller itself is also a lot better at providing a smooth speed control however it is too harsh when applying power.

Please tell me I can get this solved, I must be wrong otherwise people wouldn't be raving about the CA if it can´t even hold a steady speed of 25 km/h without going down to 17 km/h and not applying any power all the way down...

madnut
1 mW
1 mW
Posts: 16
Joined: Nov 16, 2016 12:40 am

Re: Cycle Analyst V3 preview and first beta release

Post by madnut » Apr 11, 2018 1:54 pm

Having a strange issue with my CA3 wiring hoping Justin can help. I opened up my CA to rewire some stuff and removed all existing wires then reconnected V+ and G. Noticed that it doesn't power up unless I apply slight pressure on the board in any place. On the Led connectors LED+ reads +12V but +5v on the board has no power until I apply pressure. What on the board produces the 5v? Seems like something is lose maybe if i knew where to start tracing i could find it.

User avatar
justin_le
10 MW
10 MW
Posts: 2093
Joined: May 28, 2007 3:27 am
Location: Vancouver
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by justin_le » Apr 18, 2018 3:49 am

madnut wrote:
Apr 11, 2018 1:54 pm
Having a strange issue with my CA3 wiring hoping Justin can help. I opened up my CA to rewire some stuff and removed all existing wires then reconnected V+ and G. Noticed that it doesn't power up unless I apply slight pressure on the board in any place.
My hunch would generally be that if it doesn't matter where on the board you apply pressure that it would be on the LCD header. You should have +12V on the + led, and the - LED which is the return for the power flow that should be about 3V less since the voltage is dropped across the backlighting LED before going into the 5V regulator.

If it's not that, then the 5V regulator is the small 5 pin package just to the left of the large yellow tantalum capacitor. Botton left pin should be ~10V, top left pin should be 5.0V
Big Dummy Frame (yes This One, thanks ES!) with prototype 26" Grin all-axle front hub, Phaserunner controller, and 52V 11Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled

FETguy
1 µW
1 µW
Posts: 4
Joined: Mar 30, 2018 9:42 pm

Re: Cycle Analyst V3 preview and first beta release

Post by FETguy » Apr 19, 2018 12:07 am

This is my first post, but I've been reading and learning here for awhile. Real good stuff here!
I'm just a basement tinkerer here in bike-friendly Fairfax, Ca. I've been building up and riding a rear wheel drive conversion: old 1980's Jamis frame, a used ebikekit hub motor, 36V battery pack made out of Bosch 18 V 6.3 AH tool batteries x4, Grinfineon 25A controller and of course CA3 controlling it all. Pretty tame and essentially the stuff you guys have been running for years.
I rode it with a thumb throttle for awhile, but recently added the new Sempu Gen 3 quasi-torque sensing bottom bracket, which is working great, and is much more pleasant to ride than the throttle.
The purpose of this post is to show you some data I took on the current draw of the Sempu at various supply voltages:

Sempu operating current.jpg
Sempu operating current.jpg (17.45 KiB) Viewed 216 times

The Sempu was recently purchased from Grin, and is the T2 version. As you see, the current goes down as voltage goes up - this is expected from what Sempu says in the data sheet. It is made to run off the battery supply and clearly has a buck converter (I observed ripple at 77 kHz with waveform as expected). The downside of this is that if you just plug it into the CA3 with the supplied cable, you will be running it off the 10V linear regulator, and the current draw eats up the whole power budget (per the UUG). I think it is better to run the Sempu off the battery. There are various ways you can wire that, but I chose to do it inside the CA3. I moved the white wire from the "10V" pad to the "Vf" pad. I added a 100 Ohm resistor in series to give a little isolation and to give the Sempu's regulator input capacitor something to work against, but that is not essential, you could just move the wire. Works fine. Saves about one Watt, and keeps that dissipation out of the CA3 regulator.
Anyway, hope this helps. Many thanks to all the posters here and to the hosts and moderators of this board!

madnut
1 mW
1 mW
Posts: 16
Joined: Nov 16, 2016 12:40 am

Re: Cycle Analyst V3 preview and first beta release

Post by madnut » Apr 19, 2018 1:11 am

justin_le wrote:
Apr 18, 2018 3:49 am
madnut wrote:
Apr 11, 2018 1:54 pm
Having a strange issue with my CA3 wiring hoping Justin can help. I opened up my CA to rewire some stuff and removed all existing wires then reconnected V+ and G. Noticed that it doesn't power up unless I apply slight pressure on the board in any place.
My hunch would generally be that if it doesn't matter where on the board you apply pressure that it would be on the LCD header. You should have +12V on the + led, and the - LED which is the return for the power flow that should be about 3V less since the voltage is dropped across the backlighting LED before going into the 5V regulator.

If it's not that, then the 5V regulator is the small 5 pin package just to the left of the large yellow tantalum capacitor. Botton left pin should be ~10V, top left pin should be 5.0V
Justin you rock as always! You answer was perfect, thank you so much!
Checked the (-) on the LCD and it read .5v instead of 9ish. So i cleaned off the pin and added new globs and now it works fine!
Image

BTW while probing around i noticed My Z1 is quite hot to the touch, is that normal?

Thanks!

User avatar
justin_le
10 MW
10 MW
Posts: 2093
Joined: May 28, 2007 3:27 am
Location: Vancouver
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by justin_le » Apr 19, 2018 1:32 am

FETguy wrote:
Apr 19, 2018 12:07 am
This is my first post, but I've been reading and learning here for awhile.
Hey FETguy, a most useful and insightful first post if I've ever seen one, thanks for that!

When I saw the voltage range of the Sempu sensor I also presumed it would be running a basic onboard buck converter have more current at lower voltages, but had no idea that the actual current draw at 10V could be so high. Can you clarify what you used as the current measuring device and what determined the transition between operating and sleep mode?

I admit that we didn't properly characterize this, just installed it on two of our shop ebikes with 52V batteries and found that everything worked perfectly. If it was 40+ mA all the time then I would have thought we would have seen many CA failures with Sempu sensors and 52V packs.

Your suggestion of switching the PAS power from the 10V bus to the battery V+ bus is a super good call if this is true. I'll try to arrange someone to do a current measurement across a small spread of Sempu sensors at our shop to see if your results are anomalously high or more or less the norm.
Big Dummy Frame (yes This One, thanks ES!) with prototype 26" Grin all-axle front hub, Phaserunner controller, and 52V 11Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled

FETguy
1 µW
1 µW
Posts: 4
Joined: Mar 30, 2018 9:42 pm

Re: Cycle Analyst V3 preview and first beta release

Post by FETguy » Apr 19, 2018 10:59 am

Hello Justin and all,
I measured the Sempu current draw on the bench with a variable lab power supply; voltage and current measured with Fluke 27 DVMs. I'm pretty sure the unit is healthy and that this is typical operation. It goes to sleep after 10 minutes and wakes up with slight rotation of the spindle. All this is consistent with what they say in the manual. While it was asleep I could run the voltage up and down and read currents without waking it up.
Yes, I think in 48V or 52V systems the CA3 regulator FET will get pretty warm, and of course more so if there are other external loads. Failures? Maybe when the weather warms up. . .
If you do run the Sempu off the battery voltage, I suggest taking care to stay within the 60V rating. We don't know how rugged their regulator is to overvoltages, load dump, etc. That's an advantage of staying with the 10V supply - the regulator gives protection against overvoltage. If I were running 48 V or higher, I'd put a Zener diode in series as well, say, 12V for a 48V system. The Zener and 100 ohm resistor can go in a bit of heat shrink inside the CA3 case.
BTW, does anyone know how the Sempu measures torque? Is it magnetostrictive, or strain gauges, or. . . ? Just curious.

redline2097
100 W
100 W
Posts: 142
Joined: Jun 18, 2015 1:43 am

Re: Cycle Analyst V3 preview and first beta release

Post by redline2097 » Apr 19, 2018 2:01 pm

I have 20s lipo pack made of 5 pieces of 4s 16ah multistars. Small step down converter between battery and sempu or just steal it from first lipo pack positive and negative wires (14v)?

FETguy
1 µW
1 µW
Posts: 4
Joined: Mar 30, 2018 9:42 pm

Re: Cycle Analyst V3 preview and first beta release

Post by FETguy » Apr 20, 2018 10:12 am

BTW while probing around i noticed My Z1 is quite hot to the touch, is that normal?
I can answer this because I repaired one with this problem. The Zener diode Z1 should not get hot. If it does, it is likely because the regulator FET Q1 is damaged. Measure resistance between the two legs of Q1. Should be 150 kOhm - if it is a lot less than that it is because the gate of Q1 has been damaged. You can buy a new one from DigiKey for $0.64 (p/n DN2450K4-GCT-ND). Do not substitute other FETs for this part. Removing and replacing it without damaging the board is a bit fussy, but if you are up to that, this should fix it. This part is ESD sensitive, which may be how it got damaged in the first place during your wiring and troubleshooting, so be careful with the new one. Good luck.

FETguy
1 µW
1 µW
Posts: 4
Joined: Mar 30, 2018 9:42 pm

Re: Cycle Analyst V3 preview and first beta release

Post by FETguy » Apr 20, 2018 10:44 am

I have 20s lipo pack made of 5 pieces of 4s 16ah multistars. Small step down converter between battery and sempu or just steal it from first lipo pack positive and negative wires (14v)?
I'm the new guy here, and I think it would be good to let Justin and the other experts weigh in before we arrive at the best way to power the Sempu bottom bracket. But offhand, I'd say don't tap your battery pack, except maybe as a temporary measure until you get something better going. DC/DC converter sounds okay, but is more complicated than needed IMO. I like the method I stated above of powering it from the full battery voltage with a series Zener diode and small resistor to drop some of the voltage and stay well below the 60V rating of the Sempu. I can describe that in more detail if the group decides that is the way to go.

User avatar
justin_le
10 MW
10 MW
Posts: 2093
Joined: May 28, 2007 3:27 am
Location: Vancouver
Contact:

Re: Cycle Analyst V3 preview and first beta release

Post by justin_le » Apr 20, 2018 11:20 am

redline2097 wrote:
Apr 19, 2018 2:01 pm
I have 20s lipo pack made of 5 pieces of 4s 16ah multistars. Small step down converter between battery and sempu or just steal it from first lipo pack positive and negative wires (14v)?
It's risky to steel power from one pack in a series string when you don't have BMS protection. DC-DC is best but in a pinch you can just use like a 25V zener to drop the 72V of your battery to something that is always <60V for the Sempu power. As you saw from FETguy's posts, the torque sensor current at these voltages is in the 12-15mA range, so the zener is dealing with less than half a watt to dissipate. That's no problem for most leaded zener diodes and is simpler/smaller/cheaper than the DC-DC.
Big Dummy Frame (yes This One, thanks ES!) with prototype 26" Grin all-axle front hub, Phaserunner controller, and 52V 11Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled

Post Reply