2WD trike pulling left when accelerating

cboy

100 W
Joined
Sep 18, 2017
Messages
179
Location
CA Central Valley
I have a three wheeler shown in this build thread with dual QS Motors hub wheels at the rear. Dual Kelly KLS7230S controllers. Resistance throttle to Cycle Analyst 3 and a single throttle control wire from the CA3 to each controller. From a dead stop the trike pulls hard to the left (driver’s side) under moderate to heavy acceleration.

* If the trike is rolling (5-10 mph or more) the trike goes straight under heavy acceleration.
* If accelerate slowly from dead stop, the trike goes straight.
* When accelerating heavily and pulling left, a quick release and then apply full throttle - trike goes straight.
* When trike is cruising, it tracks and runs straight...no dog tracking, and no wandering in either direction.

What I’ve checked so far:

* Throttle signal from the CA3 is identical at both controller connection pins. (.884 volts at rest, 3.665 volts at WOT)

* The programming in both controllers is absolutely identical except Controller P has the box checked to “change direction” of this wheel. Controller D does not have this box checked. The “change direction” function allows both wheels to run in sync in the same direction. To insure this wasn’t the problem, I reversed the settings in the two controllers and the trike still pulled left.

* The Kelly software monitor indicate that during operation both controllers are receiving a throttle signal of 46 (on a scale of 0-255) while at rest. At WOT both controllers are receiving a throttle signal of 185 when the wheels are not in gear.

* The monitor can also be used to compare throttle input to motor speed with the wheels in gear and up on blocks so they can spin freely. I ran a series of three tests at each of three different throttle inputs on each controller. The average results were nearly identical for both controllers.

Controller D
Throttle at 80 Ave. Motor speed is 208
Throttle at 100 Ave. Motor speed is 357
Throttle at 146 Ave. Motor speed is 1096

Controller P
Throttle at 80 Ave. Motor speed is 211
Throttle at 100 Ave. Motor speed is 357
Throttle at 146 Ave. Motor speed is 1098

* During the above throttle/motor speed tests, both wheels visually appear to increase speed directly in response to throttle input without lulls or glitches.

* Since the pulling left issue is only during moderate to heavy acceleration the "throttle/motor speed" tests may be of limited value since they don't show what's happening at acceleration.

So at this point I’m looking for other potential explanations, troubleshooting hints and/or test procedures to determine what is going on. This may also be chassis related rather than electronics but I would think a chassis problem would have a number of additional symptoms which I’m not seeing thus far. The chassis design is shown in the build thread link above and is pretty straightforward. But theories are always welcome to explain how the chassis might be the culprit.
 
cboy said:
From a dead stop the trike pulls hard to the left (driver’s side) under moderate to heavy acceleration.

* If the trike is rolling (5-10 mph or more) the trike goes straight under heavy acceleration.
* If accelerate slowly from dead stop, the trike goes straight.
* When accelerating heavily and pulling left, a quick release and then apply full throttle - trike goes straight.
* When trike is cruising, it tracks and runs straight...no dog tracking, and no wandering in either direction.

I have no personal experience with this controller but I understand it can be programmed for Speed, Torque, or Balanced modes. Operation of the first two is clear, but the last is a proprietary crap shoot. You should be using torque mode to ensure the torque on each side is proportional to the throttle signal. It's a little unclear if balanced mode will work appropriately.

If the issue is not the operating mode then:
Since this is only happening under high acceleration, it smacks of different phase current limits. These limits are only in play under high load at low speeds. If the controllers are set the same but one is pulling harder then the controllers may not have identical phase current calibration. Try reducing the phase current limit on the right or increasing it on the left so the controllers max out phase current at the same level (in spite of having different configuration values).
 
Thanks for the quick response teklektik. I'll sort through the manual to see if I can follow up on the settings you suggest. In the meantime another little tidbit of test info. I just did a quick road test where I disconnected the throttle from Controller P (passenger side wheel) and powered only the drivers side wheel. With just moderate throttle this caused the trike to go hard to the right. When I reversed the set up, connecting only to the passenger side controller, it caused the trike to go hard left. What that tells me is that I think YOU are on the right path and that it is a controller mismatch and not a chassis design issue. So I'm off to follow up on your reply.
 
teklektik said:
I have no personal experience with this controller but I understand it can be programmed for Speed, Torque, or Balanced modes.

I checked the specs and it turns out the KLS series of controllers is NOT programmable for torque mode...or speed or balanced. I'm next looking into your suggestion of tinkering with the phase current limit. My question is, would adjusting phase current for balanced acceleration have a negative effect during "cruising" when there is little or no acceleration? Edit...oops, just re-read your post and now see that it will have little or no effect when not accelerating. So disregard that question.
 
Just did some tinkering with the phase current limit on controller P (passenger side) and the adjustments are working. I currently have that controller set at 50% and the other controller at 100% and it is accelerating straight. I'll do some more fine tuning tomorrow to get it really dialed in (max current without pulling). But things are on the right track. As always, thanks for your wealth of knowledge and willingness to help.
 
Could you possibly have a bad connection in one of the phases that's burning off some of the amps as heat during heavy acceleration?
 
Voltron said:
Could you possibly have a bad connection in one of the phases that's burning off some of the amps as heat during heavy acceleration?

The external connections all appear solid. But what it looks like inside the motors or the controllers I don't know. Is there a way I can test for this possibility?
 
It's unlikely, because at the levels of current you're probably dealing with, you'd probably see some results of the heat at the wires or connections.

However, you can use an IR thermometer, focusing it's laser dot on each connection point for each phase, and see if any are warmer than the others.

That would indicate a higher resistance, which would cause some loss of phase power at that connection.


The most likely thing is either a poor connection at a phase, or simply unbalanced operation of the controllers--one of the controllers simply has a different curve for output vs the other one, either because of a manufacturing issue, or the way it's programmed from the factory (which may not be something you can change).

It could also be something different in one motor vs the other, but that should show up even in lower accelerations.

To test for motor differences vs controller differences, you can connect the phase wires opposite, meaning connect the phases from the righthand controller to the lefthand motor, and vice-versa. If there is no change in the issue, then the motor is the cause. If the problem swaps sides, then the controller is the cause.
 
amberwolf said:
To test for motor differences vs controller differences, you can connect the phase wires opposite, meaning connect the phases from the righthand controller to the lefthand motor, and vice-versa. If there is no change in the issue, then the motor is the cause. If the problem swaps sides, then the controller is the cause.

If this would require switching the large phase wires (relatively difficult on my setup) and the smaller phase sensor wires (very difficult on my set up) then it would probably be easier to just swap the controllers from side to side. Then all the wiring harnesses would just hook up without having to fabricate a bunch of jumpers etc. Wouldn't swapping the controllers provide the same clues as to which is a fault...wheel or controller? I'm not sure I want to tackle this at all if I can get the thing to run straight and maintain satisfactory overall performance by just adjusting the phase current limits to balance out the torque of the two controllers. It would be interesting to know exactly where the difference originates, but a lot of effort to find out. If my overall performance drops off as a result of adjusting down the phase current limits, then that's a different story and I'd definitely go digging deeper to find an alternative solution.
 
cboy said:
I checked the specs and it turns out the KLS series of controllers is NOT programmable for torque mode...or speed or balanced.
The Kelly specs are essentially worthless regarding specific available features and programming options. You need to run the configuration program and snoop around.

Sadly, there are no means to run the app without an RS-232 adapter and controller - it won't even come up. I have been unable to locate any screen shots of the program for your controller specifically, so there is no way to actually know what options are available. Importantly, we don't know how your controller is specifically configured so this is sort of wandering in the dark at this end. You are getting generic advice because of the absence of problem specifics.

Posting up screen shots of your KLS-S controller config would be nice and might shed some light on matters.
But, in any case, moving on...


cboy said:
I currently have that controller set at 50% and the other controller at 100% and it is accelerating straight.

This is an indicator that something is seriously wrong. This is way beyond a controller internal phase current measurement calibration error.


cboy said:
Wouldn't swapping the controllers provide the same clues as to which is a fault...wheel or controller?

Yes - swapping controllers will certainly localize the problem. This is a critical step. I do not think you want to proceed to 'just get it working' because you have a clear indication of a 50% phase current screwup somewhere - this could possibly have unpleasant consequences if you don't correct or at least identify the issue.

At this point, my knee jerk reaction is that it's much more likely that the problem lies in the controller than the motors.

If we assume the controllers are accurately measuring the phase current (which directly maps into torque), then It appears that the controller that you are leaving at 100% is passing a huge amount of current but getting little torque in return. This suggests a phase mismatch or hall timing error. I believe this controller has a setup or autolearn mode - this sort of mode is designed to eliminate the need to get phase wires connected in a specific order as with many controllers. I would start by re-running the auto-setup for the 100% motor and see if matters improve. Repeat the autotune on the other motor and check the results there. (Importantly, this must be done every time you randomly connect phase wires or swap controllers and motors.)

If this does not give any joy, I would go to each motor in turn and swap two phase wires and re-autotune on the off chance the controller is having a hard time with those model motors for some reason. Theoretically, they should perform the same after this Hail Mary wire-swap, but it's something you can easily do and autotune issues are known to exist so this is not as wacky as it sounds.

If still no joy, you can now be reasonably certain the controller phase/timing is the best the controllers can do - probably correct. Now you can move on to swapping controllers to see if the problem moves with the controller. If it does, this should give you enough information to contact Kelly and get their take on the unbalanced performance. If it doesn't then you have a motor issue and using the configured/mismatched phase current limits will probably be the way to go (at least for now) since the controllers won't be harmed and pursuing motor investigation or replacement may not be in the cards just now.
 
teklektik said:
Posting up screen shots of your KLS-S controller config would be nice and might shed some light on matters.

Here are screen shots taken before any adjustments to the phase current limits. There are three screens for each controller: "vehicle", "control", and "motor". I identify the two controllers as P for passenger side and D for driver side.

D Vehicle

driver-vehicle.jpg


D Control

driver-control.jpg


D Motor

driver-motor.jpg


P Vehicle

pass-vehicle.jpg


P Control

pass-control.jpg


P Motor

pass-motor.jpg
 
Interesting. This is quite different from screen shots I've seen for config apps for other kelly controllers.

I see no explicit controls for an autotune or learning mode. How do you set up the phase/hall connections?
 
teklektik said:
I believe this controller has a setup or autolearn mode - this sort of mode is designed to eliminate the need to get phase wires connected in a specific order as with many controllers. I would start by re-running the auto-setup for the 100% motor and see if matters improve. Repeat the autotune on the other motor and check the results there. (Importantly, this must be done every time you randomly connect phase wires or swap controllers and motors.)

I'm not seeing anything in the program like autolearn or autotune. On the screenshots in my prior post you can see a "setup wizard" on the left side of the screen. This is apparently just a logo. It is not a clickable function. The only button I am not certain of is the "read zero" button on the vehicle screen. I have not used this or clicked it. The pages shown in the screenshots are the only programming available. There is a "monitoring" screen, shown below, which looks like this when the throttle is at rest. In the second column there is an item "phase current". This does flash numbers as the throttle is applied but the numbers change too quickly for me to make any sense of it. The only other screen is the COM screen which simply identifies the communication port.

Monitor Screen

monitor-screen.jpg
 
teklektik said:
I see no explicit controls for an autotune or learning mode. How do you set up the phase/hall connections?

I just connected the wires to match the color coding...yellow, blue and green. The hall sensor wires (the small wires from the hub motor) connect together in a pre-configured connector (on the QS Motor hub) to a matching connector on the kelly controller. According to the Kelly manual, one of the first this to do in setting up is to run their "Identification Angle" program. It is unclear in the manual exactly what this does, perhaps it is setting the phase/hall connections. I don't know. Maybe this is even the "autotune" function you mentioned. I did run this for both controllers according to the instructions and it went thru the process (it briefly turns the wheels in both directions) and it posted an "85" on the "vehicle" screen, which Kelly says is the indicator that everything is a-okay and ready to be driven.
 
cboy said:
teklektik said:
I see no explicit controls for an autotune or learning mode. How do you set up the phase/hall connections?

I just connected the wires to match the color coding...yellow, blue and green. The hall sensor wires (the small wires from the hub motor) connect together in a pre-configured connector (on the QS Motor hub) to a matching connector on the kelly controller. According to the Kelly manual, one of the first this to do in setting up is to run their "Identification Angle" program. It is unclear in the manual exactly what this does, perhaps it is setting the phase/hall connections. I don't know. Maybe this is even the "autotune" function you mentioned.
Yep - that's it. There is no 'standard color coding' - running that program sorts out the phases and halls and does some positioning angle measurements for the halls. It will work even if you connect the colors every which-a-way.

Okay - so I would rerun the id program, etc as mentioned in the above post where I called out 'learning' or 'autotune'. This is getting down to the last resorts for standard procedures to verify all is well, but if you contact Kelly about the torque mismatch, they are pretty surely going to ask you to do the same.
 
FWIW, there have been threads around here where the automatic "learn" routine simply wouldn't work on some controller/motor combinations; I don't remember the brands though. Sometimes the only solution was to manually determine the phase / hall wiring combination for the lowest off-ground battery current that gets the motor running in the correct direction.

If rerunning the routine on yours doesn't fix the problem, you may have to do the same.
 
teklektik said:
Okay - so I would rerun the id program, etc as mentioned in the above post where I called out 'learning' or 'autotune'. This is getting down to the last resorts for standard procedures to verify all is well, but if you contact Kelly about the torque mismatch, they are pretty surely going to ask you to do the same.

I reran the ID program. No change, still pulled left. Came back in the garage and upon close inspection noticed that the pre-wired (by QS Motors) connectors for the hall sensor wires had the yellow and green wires reversed. I hadn't read you post at that point saying the wiring really doesn't matter and the ID program will sort it out. But I did reverse those wires on both controllers so that the colors matched up. I then reran the ID program once again and took it out for a test run. No change...still pulls left.

So after lunch I'm heading back out to swap the controllers, run the ID program again and then take a test drive.
 
amberwolf said:
FWIW, there have been threads around here where the automatic "learn" routine simply wouldn't work on some controller/motor combinations; I don't remember the brands though. Sometimes the only solution was to manually determine the phase / hall wiring combination for the lowest off-ground battery current that gets the motor running in the correct direction.

I'll try a search for those threads...maybe some other clues in the posts. Thing is, I can get the motors running in the correct direction so would the off ground battery current tell us anything helpful?
 
cboy said:
...I can get the motors running in the correct direction so would the off ground battery current tell us anything helpful?
Somewhat. If these are roughly the same at full throttle and relatively small then the controllers are set up okay - which is sort of what it looks like just now. (Those are big motors with big tires - I'm not sure what 'relatively small' would be: somewhere between 3A-5A I would think - guessing here.) That result would suggest that the normal running mode is okay but the phase current limiting is somehow fouled up in one since that is a separate mechanism not in play in the unload full throttle mode. If the currents are different or large, then a bunch of things could be awry so that result is not as helpful.

Instead, I would move on to swapping the controllers first instead of fussing with trying to identify a particular kind of controller failure. If swapping moves the problem, there's no need to drill into the specifics of controller issues.
 
Well, looks like the hub motor is the culprit. I swapped controllers side to side, reran the Angle ID program and took it for a test ride. Pulls hard left, just as before. Anyone have ideas for troubleshooting at this point?
 
If you touch two of the three phase together, while you spin the wheel by hand in the air, is should suddenly start cogging ( having a " notchy" feeling to the turning). Do all three combos.
If it doesn't then one of the windings is shorted.
 
cboy said:
Well, looks like the hub motor is the culprit. I swapped controllers side to side, reran the Angle ID program and took it for a test ride. Pulls hard left, just as before. Anyone have ideas for troubleshooting at this point?

Good troubleshooting
I was going to suggest:

1) lift and listen... Is either motor louder? (Bad combo)

(I was very alarmed to hear 50% and 100% made it go straight.. )

2) measure freewheel current and speed on each side by removing one controller at a time.

Sorry if those were suggested - I skimmed

...

Other obvious:

3) exact same rolling diameters? So rim, tire, inflation?

4) dragging brake on one side?

Its been isolated to a -side- but always return and double check simple stuff.
Good luck!

Edit: same kv on motors? Same cogging feel when hand turning?

-methods
 
Back
Top