Cycle Analyst V3 preview and first beta release

[Cowardlyduck]

AW, that is something I've occasionally thought about solving as well. Also because it's just nice to ride without a screen in the face sometimes.

If I find the time, I might design a 3D printed flip down cover. Printed in black, it should block 90%+ of the light. That would be the easiest way IMO.

Cheers

 

[amberwolf]

Hmf. I never even thought of printing a cover***.... I would rather do that than modify the electronics. If you ever make a design, lemme know and I'll try printing it out.

If it's made as a cover kinda like a phone case where it has sides to go down over the CA's sides, it'll block light pretty completely, if the hinge itself is below the rounded edge of the forward CA side.

If someone needs to use it on a system that uses presets changed by the front panel buttons, then if it's made thin enough around the button area with thicker bumps where the buttons are it shouldn't need any holes, if it's printed with soft enough TPU. I have a roll in black shore95A and another in gray (95A), which are hard but flexible enough for this if printed thinly.

If I ever run into the time to design one, i'll post it up here. (but don't hold your breath :lol: )


***which is pretty wierd since I once wanted to print a new back cover that would have connectors "molded" into it similar to the ebikesolutions cover; that was long ago before I actually had a printer.

 

[Cowardlyduck]

I screwed up the wiring on a PAS sensor and plugged it in...let out the magic smoke

Any chance this can be repaired?



Or does anyone want it in Australia?

Cheers

 

[amberwolf]

There are a couple of alternatives.

One is to power it via a lower voltage, bypassing the regulator FET, as long as the secondary regulator past Q1 still works. If you have a "lighting pack" for 12v lights and such you can run it from that, by connecting the 12v to the Q1's output pin, and change the shutdown voltage in the CA menus to below that. If you don't have a lighting pack you can use a tiny DC-DC to get 12v from your battery.

Then if the Vf input is fried and it doesn't read your battery anymore, connect battery voltage (<100v? check your manual!) to the Vex pin on the Divider input so it still gets a voltage reading on the traction pack.

I did this over a decade ago after blowing up Q1 on my first CA because it didn't have any warning (other than on the website or in this thread if you knew about it and dug around) that using the THUN BB torque sensor with it (that IIRC came with mine) on anything above a 48v pack would put too high a load on Q1 and overheat it. IIRC they have changed the modern CAs to deal with this issue, but I don't know that for certain.

Or you can replace Q1 itself, and the resistor. This thread has some info on parts you may be able to use, if yours is the same version:

Cycle Analyst Repair

I'm sure I have FETs that will work; I can probably use an IRFB4110 from a dead controller (or a new one if I have to--I have a few) just to test if it is all that is wrong, but I'm pretty sure it's not. Oh, and when i wire it in or solder in a drop-in-replacement, I'll have to run the gate...

endless-sphere.com

 

[Cowardlyduck]

There are a couple of alternatives.

One is to power it via a lower voltage, bypassing the regulator FET, as long as the secondary regulator past Q1 still works. If you have a "lighting pack" for 12v lights and such you can run it from that, by connecting the 12v to the Q1's output pin, and change the shutdown voltage in the CA menus to below that. If you don't have a lighting pack you can use a tiny DC-DC to get 12v from your battery.

Then if the Vf input is fried and it doesn't read your battery anymore, connect battery voltage (<100v? check your manual!) to the Vex pin on the Divider input so it still gets a voltage reading on the traction pack.

I did this over a decade ago after blowing up Q1 on my first CA because it didn't have any warning (other than on the website or in this thread if you knew about it and dug around) that using the THUN BB torque sensor with it (that IIRC came with mine) on anything above a 48v pack would put too high a load on Q1 and overheat it. IIRC they have changed the modern CAs to deal with this issue, but I don't know that for certain.

Or you can replace Q1 itself, and the resistor. This thread has some info on parts you may be able to use, if yours is the same version:

Cycle Analyst Repair

I'm sure I have FETs that will work; I can probably use an IRFB4110 from a dead controller (or a new one if I have to--I have a few) just to test if it is all that is wrong, but I'm pretty sure it's not. Oh, and when i wire it in or solder in a drop-in-replacement, I'll have to run the gate...

endless-sphere.com

Thanks AW! Will give it a try.

 

[amberwolf]

Thanks AW! Will give it a try.

Any luck?

 

[Gab]

Hi i have a weird.random throttle behaviour. I had a look inside.and didnt see or smell anything burnt.and no water. What.it does is.randomly sometimes it puts like 1/4.throttle on its own! But on the screen the input and output bars on the .in screen show nothing?? The other weird thing it does if i do.full trhottle off takeoff and the wheel off the ground it only ramps up very very slowly like 5 seconds to get to full throttle.which is not the cav3 setting then other times it ramps up.normal faster. Its like their is a bad connection of the output mosfet to have this random behaviour.sometimes its ok. I know its the cav3 becuase if i bypass the throttle direct to controller it works perfectly fine. The cav3 is on passthrough mode hence why i think their is some issue with the output mosfet or wiring any ideas on what to check ?

 

[amberwolf]

Broken ground on the would be the most common.

Anywhere between throttle and CA, or CA and ocntroller.

If you're using JSTs it's probably at the back end of a crimp on a contact, or a pin that has backed out of the housing as they were pushed together, and when you unplug them look like theyr'e all seated but if you push on the tips one moves back.

 

[Gab]

I have.plugged an unplugged everything and i have.not had the random 30pcnt throttle. The only weird behavior it now does...when i turn it on and change.mode to.full power it takes like 30.seconds to ramp to full throttle. I can see.on the display it shows full throttle but why does it ramp up.so slow? Is their.a.soft.start feature.after u change to high power mode? This doesnt sound like a ground problem ?

 

[amberwolf]

I have.plugged an unplugged everything and i have.not had the random 30pcnt throttle.

That means it *is* a connection or wiring fault...but it doesn't mean it's fixed. I would closely examine hte ground pins and wires of all connector related to throttle. If you don't make *sure* the problem is really fixed, you could have this while riding, and be surprised / unable to do anything about it for just long enough to get hurt or cause a serious problem for someone.

The only weird behavior it now does...when i turn it on and change.mode to.full power it takes like 30.seconds to ramp to full throttle. I can see.on the display it shows full throttle but why does it ramp up.so slow?

What ramping settings do you have in each throttle control source (CA, and controller)?

Is their.a.soft.start feature.after u change to high power mode?

Where are you changing this mode?

Are you testing without the CA in the loop? If not, you need to do this to isolate the problem. Youc an do it via hardware direct throttle connection to the controller, or by using the right mode in the CA to bypass the whole CA's control of throttle output.

If you have two separate things that modify how the system behaves (the CA, and the controller's own setup), it's going to take you forever to get it working if you have to guess which one is causing any specific behavior.

This doesnt sound like a ground problem ?

No, a ground problem is the kind of thing you saw before.

 

[flug32]

I've recently installed a Cycle Analyst attached to a Hilltopper Horizon controller & motor. It has generally been working fine, but today I ran a battery down for the first time and observed what I think is puzzling behavior.

It is a 36 volt battery with a 350 W motor.

I am running CA 3.2b3 so that I can use the 'instant power on start' with the torque sensor.

So here is the issue: As the battery was running down, just above 31V I observed reduced wattage going into the motor, and noticeably reduced power. This is what I had expected as I'd entered 31V as the low-voltage cutoff.

So then it went along for not much further, maybe a mile or two, and when the voltage was showing somewhere around 30.5V it stopped providing power.

So that is what I expected also. It seems to be the CA cutting off the power, not the motor controller or battery pack - although I might be wrong about that. I'm not sure how to tell the difference for certain.

But then I thought, well let's see what happens when I set the Low Voltage Cutoff to 30 instead.

Also, if the controller/battery is cutting off at around 30.5-31V, where is the smart place to set the CA's Cutoff Voltage setting?

I got a little further, but not far - basically just as far as when you let the battery recover for a bit and the voltage bounces back up a little.

So then I reduced the voltage cutoff to 29, then 28, then 19. Same behavior each time.

So I guess this might be the battery or motor controller cutting out due to the battery voltage being low. But when I've experienced that before, I couldn't get it to turn on again at all, let alone go a half mile, and then turn on again for a while 2 or 3 times.

So what do you think? Could it be the CA cutting off power due to lower voltage, or is it the controller or battery? How can I tell the difference?

Part of my concern is that the battery seems to be running out sooner than it should based on its amp-hour rating and amp-hours used as shown on the CA, the little battery-remaining display seemed to be running down faster than I expected given the amp-hours and mileage I expect out of this battery - yet then the battery gave up and quit pretty well when I would have expected given that display. (Unfortunately I re-flashed the firmware on the CA midway through this battery, and didn't think to write down the battery status before erasing it with the re-flash, so I don't have the complete stats for the full battery run from the CA.)

 

[chuyskywalker]

So what do you think? Could it be the CA cutting off power due to lower voltage, or is it the controller or battery? How can I tell the difference?

The CA "limits the motor" by adjusting the throttle output signal. It uses fancy math to figure out all the inputs (real throttle, brakes, motor power, pedals, etc, etc) and then adjusts the throttle output to the controller to get you where you want according to how all of those signals mesh together.

One of the screens shows actual throttle and output throttle signal levels, I believe. On that screen while the CA is reducing go-juice, you should see the output throttle values looking quite a bit reserved.

 

[flug32]

FYI I tried the CA 3.2 beta 2 firmware, because this installation is on a recumbent, so I thought maybe I could use the "instant torque when pedaling after coasting" feature. I recently installed the Fischer BB torque sensor.

(Apparently the problem found in beta 2, and the reason this feature was removed in beta 3, is that in some positions of the pedals for some bikes, a high torque is reported, even though the rider isn't really pushing the pedals at all. This in turn causes the motor to turn on unexpectedly, possibly with full throttle. So this feature was removed in beta 3 for safety reasons.)

I really like the "instant power on startup when you torque the pedals" feature and so I thought maybe I would like the instant power after coasting feature as well. And I thought, perhaps I wouldn't have the "phantom torque" problem since this is a recumbent and I don't generally need to put any force on the pedals when coasting.

However, I can report that this experiment was a failure and strange, random, very large torque readings are reported in some certain positions of the pedals, even when no particular force is being put into them. It is odd because it is infrequent and with no particular pattern that I could find. But it would happen with some regularity. When this happened, of course the CA would input full throttle and so I would be off to the races even if I didn't really want to be. So it definitely is a safety and control issue.

I think a feature like this could be made to work, but it would require quite a lot more controls and fussing about to make it work properly and be safe. But (based on my experiments) I do think more power could be provided, safely, about half a pedal stroke faster than it currently is. My typical cadence is around 60RPM, so that means the power could be felt maybe 1/2 second after pedal input rather than the current about 1 second (on my setup).

(Some of the delay in my setup is because the throttle of my motor controller ramps up fairly slowly - so the CA has to deal with that delay regardless of everything else.)

What I think it would take:

- Only do this after pedaling has stopped for a while, and then resumed
- Require something like minimum 1/4 or maybe 1/6 or 1/8 pedal revolution as indication of resumption of pedaling (ideally this would be user-settable)
- Also require the torque indication. So maybe it is torque + 1/4 (or 1/6 or 1/8 or whatever) revolution rotation, or maybe it could be on rotation alone. (Or maybe the user could choose.). But NOT torque alone - it is too unreliable.
- Limit the initial boost to maybe 100 or 150 watts (again, ideally user-settable). This would help with the safety aspect - even if this does fire off inadvertently, it is not this H-U-G-E unexpected burst of power, but more of a small bump.

Whether it would be worth messing around with all this to get it working, I can't say. Or whether there is enough space left to work with to make this work, since it is a fair bit more complex than simply linking the torque output to the throttle input. I do think if it could be made to work reliably, it would be a moderate improvement in responsiveness to torque input over 3.2b3.