2WD trike pulling left when accelerating

[methods]

WOOOHOOO

Finally got the controllers re-flashed for pure current mode. ....

And the best news of all, the bike runs straight and true ....

Wow...I'm just elated. And a huge huge shoutout to Fany and Kelly Controllers. ...

Kelly is a good company. They treated me well over the years. They make good stuff that works.
So - congrats on finding your solution.

You are now commanding a specific current to go to both controllers... to drive both motors equally....
We will leave it to the experts to determine what was going on while you were in speed mode

Current mode is nice


-methods

 

[BVH]

I have nothing to do with this type of build but I followed it with interest. It's a great documentation of the development of your diagnostic skills, your perseverance and ultimate overcoming of a very significant problem! Congrats!

 

[wturber]

Yeah. This was a fun one to follow along with.

 

[cboy]

Yeah. This was a fun one to follow along with.

Ha. Maybe fun for YOU. :lol:

 

[wturber]

Yeah. This was a fun one to follow along with.

Ha. Maybe fun for YOU. :lol:

OK. Maybe "fun" wasn't the right word. Let's say "interesting." But yeah - I'm sure it was frustrating from your end.

 

[phate]

WOOOHOOO

Finally got the controllers re-flashed for pure current mode. Long hard battle getting the install program to work. Totally a problem with my old XP laptop that doesn't go beyond 1024 x768 screen resolution. And the install program was written with wider screen resolution. So of the most important "buttons" to do the install were not visible on my screen. So it took me almost a day to figure out how to get a big screen monitor hooked up and working to get wide screen resolution. But finally got it going this morning and did the re-flash.

And the best news of all, the bike runs straight and true with full, 100% phase and 100% speed in both controllers. What a difference. It is like a totally different vehicle now. Have to do some much longer road test runs but I can't imagine needing anything more than a few tiny tweaks. One side note to anyone doing this in the future. When you reflash the controllers it resets everything you previously programmed into the controllers, like reversing the wheel direction (don't ask how I discovered that). And you have to rerun the angle identification program. So basically you are starting over in terms of any programming...but that part is fairly easy.

Wow...I'm just elated. And a huge huge shoutout to Fany and Kelly Controllers. They really came through even though I had not purchased the controllers directly from them (but rather through QSMotors) they really went the extra mile to see if this change would work for my setup. Also a huge shoutout to Teklektik and everyone else who contributed to this rather long and difficult thrashing of the problem. I learned a ton along the way. And hopefully this thread will help others who might use a dual hub motor configuration with the kelly KLS controllers. NOW I'm going for a nice lone ride...to enjoy the breeze...and not pull to the left!

Wait, you can reflash these things now?!?

I feel like pure current mode would be a lot better for my application anyways. Does the controller still have the "soft start", or will they pretty much go full blast from the start?

 

[Chalo]

Now it's time to put an encoder on your steering axis so you can bias power to the outside wheel.

 

[amberwolf]

Now it's time to put an encoder on your steering axis so you can bias power to the outside wheel.

That's something I am working on (iin hte back of my mind) for SB Cruiser, since powering thru turns with the outside wheel makes them a lot sharper, so I can take the turns at a higher speed and get out of traffic's way better when needed

Right now it uses separate throttles...but at some point it will use just one, and be able to be controlled by a torque-sensing BB on the pedals.

 

[teklektik]

Now it's time to put an encoder on your steering axis so you can bias power to the outside wheel.

No need. Torque controllers deliver the torque specified by the throttle regardless of speed. With torque controllers and the trike side to side drive geometry, the inner and outer wheels provide equal thrust but are free to turn at different speeds when turning. This contrasts with PWD or speed controllers which by design attempt to make the inner and outer wheels turn at the same speed - attempting to force the vehicle to travel straight.

So, for this trike with torque controllers, the mechanical steering works normally without fighting the rear drive or tire scrubbing. Since power is torque times rpm and the wheels have the same torque while the outer wheel has higher rpm when turning, the trike naturally applies more power to the outside wheel without extra electronics. The controllers sense the increase or decrease in current from the changing inner/outer wheel load when turning and twiddle the current (torque) as required. Turn the bars and the rest is automagic. Pretty cool really...

 

[phate]

Now it's time to put an encoder on your steering axis so you can bias power to the outside wheel.

No need. Torque controllers deliver the torque specified by the throttle regardless of speed. With torque controllers and the trike side to side drive geometry, the inner and outer wheels provide equal thrust but are free to turn at different speeds when turning. This contrasts with PWD or speed controllers which by design attempt to make the inner and outer wheels turn at the same speed - attempting to force the vehicle to travel straight.

So, for this trike with torque controllers, the mechanical steering works normally without fighting the rear drive or tire scrubbing. Since power is torque times rpm and the wheels have the same torque while the outer wheel has higher rpm when turning, the trike naturally applies more power to the outside wheel without extra electronics. The controllers sense the increase or decrease in current from the changing inner/outer wheel load when turning and twiddle the current (torque) as required. Turn the bars and the rest is automagic. Pretty cool really...

I think they were getting at implementing some form of torque vectoring.

I already have an Arduino running the speed controllers (brushed RC car parts) in my daughter's power wheels, so that idea has been bouncing around my mind as well.

 

[teklektik]

No need.

I think they were getting at implementing some form of torque vectoring.

Yep - I couldn't tell if it was a joke, or something as fanciful (for a delta trike) as torque vectoring.

From the outset in this thread, the issue of speed vs torque controllers has been a point of confusion for some so it seemed useful to better explore that from the perspective of basic operational 'need', not steering 'enhancement'.

• The problem in the OP's build was actually resolved way back on page four when the controllers were discovered to run in speed mode. What may not have been clear was that all of the effort after that was skirting the issue of a bad controller type in attempts to understand why the speed controllers couldn't even go straight (we know they don't like to turn), and perhaps make the best of a bad situation by reducing power or other compromise. It was just the wrong control mode and this business of steering was one of the big reasons.

Anyhow - it seemed useful to spend a little time on a point of basic handling. It was mentioned in passing a few times but was largely overshadowed by the veering discussions. As for torque vectoring a rear-heavy lead-powered delta trike so it can corner sharper - well...

 

[amberwolf]

Now it's time to put an encoder on your steering axis so you can bias power to the outside wheel.

No need. Torque controllers deliver the torque specified by the throttle regardless of speed. With torque controllers and the trike side to side drive geometry, the inner and outer wheels provide equal thrust but are free to turn at different speeds when turning.

While it may not be necessary from a controller standpoint, if one wants sharper turns at the same speed, biasing the outer wheel to a higher torque than the inner one will do that.

In an extremely hard turn, actively de-powering the inner wheel can enable even sharper cornering, as long as the vehicle characteristics don't allow it to tip over at that speed.

I use this to my advantage in traffic with the SB Cruiser delta trike, which while not lead powered is certainly rear-heavy. :lol:

 

[cboy]

In an extremely hard turn, actively de-powering the inner wheel can enable even sharper cornering, as long as the vehicle characteristics don't allow it to tip over at that speed.

This made me think back to tractor driving during my youth where the right and left brakes are used to help turn very sharp corners. While braking seems a little self defeating, what if on a dual motor vehicle one had separate hand levers to control the variable regen for each controller. The rider would turn the handlebars while simultaneously squeezing the regen lever for the inside controller. The inner wheel would brake slightly while the outer wheel provided full thrust. Based on my brief experience with the stopping power of regen, I can imagine making a pretty sharp turn that way. This might even improve stability a bit since the front wheel would not have to be turned quite as far because the regen braking to taking on a good part of the turning chores.

 

[amberwolf]

Unless you were in a chase or a race, it's unlikely that would be necessary. Just increasing power to the outside and decreasing or removing it from the inside is sufficient on SB Cruiser. While it's a totally different beast from yours, I think the same principle would apply to yours.

You could try it and see, by temporarily adding a second throttle on the other handlebar end, and having each of them go directly to the Kelly on it's own side, and operating them as needed.

WWould take some practice, but not much before you could try the test itself.

SBC is narrow enough and at slow enough speeds (20mph) that it also rolls fine with just one motor at a time, and the amount of steering correction needed is small and is just automatic to me. (not really any worse than just dealing with that from the rroad crown slopes).

 

[cboy]

During some major road testing I have noticed an occasional "pull left" on acceleration. To attempt to kill off this gremlin for good I went back in this thread to the "make the ground wires equal length" theory. I made all the main (4 awg) ground wires from both controllers equal length. And a very unexpected thing happened. The hub motors would grunt a bit but wouldn't turn. When I hooked the grounds back up in their original configuration, the motors run fine. Here is a sketch of the two wiring configurations.

The one on the left, where the passenger controller ground connects to the driver controller ground and then goes on to the shunt works...this is basically the original configuration but a bit shorter wire connecting the two controllers. The one on the right, where the passenger controller ground goes directly to the shunt and the driver controller ground goes directly to the shunt, will not turn the wheels. This is just baffling to me but I'm sure someone can reason it out...and maybe offer a remedy.

 

[amberwolf]

If one works and the other doesn't, then there's some other difference you may not have noticed. It could be a bad crimp, high resistance connection, or another unrelated wire that got disconnected, but there will be something.

It would work in either of those configurations, otherwise.


Do the dots in the lines represent additional connection points? (like where controller wires bolt to or plug into the shunt wires). If so, perhaps its at one of those points.

 

[cboy]

I put a 4awg "jumper" in the wiring as shown by the red line in sketch below and the wheels operate normal. If the jumper is removed that configuration does not work. But with the jumper in place I don't really know how the current is actually flowing. My purpose was to make the ground wires equal length but I don't know that this achieves the purpose since the current could still be flowing from controller P to controller D and then to the shunt rather than traveling the equal length wires which connect each controller to the shunt.

 

[cboy]

Do the dots in the lines represent additional connection points? (like where controller wires bolt to or plug into the shunt wires). If so, perhaps its at one of those points.

The dots are the connection point between the ground wire from the controller and the 4awg wire that runs to the shunt. At one point I had the ground buss connected to that connector on the Driver side controller. The wheels would work with that connection with the configuration shown on the left. It would not work with the configuration on the right. To try to eliminate that as the problem, I connected the ground buss directly to the shunt. Again, it would work in configuration on the left but not the one on the right. One of the grounds on the ground buss is from my auxiliary 12 volt battery. If I disconnect the ground from that battery then the wheels will operate in either the left or right configuration WITHOUT the red jumper wire in place.

 

[amberwolf]

I put a 4awg "jumper" in the wiring as shown by the red line in sketch below and the wheels operate normal. If the jumper is removed that configuration does not work.

Then that means a connection is being made by that jumper that is not being made without it.

If both controllers work with the jumper but only one works without it, then the current path is from the controller that works thru it's wiring to the ground point, and there is either a high resistance or an open in the connection point for the other controller that's not working.

If both work with the jumper but neither works without it, then with it in place it is making a connection to ground that the other wires aren't, which implies there is another connection at one or both of those points that isn't shown on your drawing. It also implies that the crimps or connections of the new wire(s) are not good enough to handle the current.

Perhaps the cable insulation wasn't stripped before crimping?

But with the jumper in place I don't really know how the current is actually flowing. My purpose was to make the ground wires equal length but I don't know that this achieves the purpose since the current could still be flowing from controller P to controller D and then to the shunt rather than traveling the equal length wires which connect each controller to the shunt.

Current is flowing in an unknown path, thru whatever wire(s) you are not showing in that drawing. It can't be flowing thru either of the other wires to ground, that are between shunt and red jumper, if it won't work without the red jumper on either controller.

If it works with one controller without the jumper then it is flowing thru the jumper into that controller's wire to the shunt, and is thus the same as it was previously with unequal wire lengths.

 

[amberwolf]

One of the grounds on the ground buss is from my auxiliary 12 volt battery. If I disconnect the ground from that battery then the wheels will operate in either the left or right configuration WITHOUT the red jumper wire in place.

That changes things from my post above, assuming I am correctly visualizing your wiring changes.

This probably means there is a voltage difference between the 12v ground and the controller system ground, and the 12v ground is "lifting" the controller ground, interfering with controller operation.

If it used to work with the 12v ground connected to the controller ground, but doesn't now, it means there is some other change in the system, probably wiring, that is causing this ground lift.

I don't know that these are the cause, but:

I've had wiring issues caused by "upgrades" to wiring where I was dead certain that I had wired everything correctly, only to find that some casing for some switch or light or controller or whatever, was connected to something internally (intentionally or not), and causing a short (full or partial) to some other system via the bike frame or the panel metal, or whatever.

I've also had problems caused by my certainty I'd wired it correctly when in fact I'd swapped wires somewhere....

And I've also had them caused by abraded or split insulation, where it rubbed or pushed against something that was electrically connected to something else. :/

 

[cboy]

This probably means there is a voltage difference between the 12v ground and the controller system ground, and the 12v ground is "lifting" the controller ground, interfering with controller operation.

If it used to work with the 12v ground connected to the controller ground, but doesn't now, it means there is some other change in the system, probably wiring, that is causing this ground lift.

I don't quite understand what "lifting" is but if it is something like stealing the ground or a portion of the ground perhaps this might also have something to do with the entire issue of the bike pulling left under acceleration. If "lifting" means some or all of the current could be diverted where it is not intended to go (but not a short or fuse blowing situation) AND this situation could be exacerbated upon acceleration, the two issues may somehow be linked or related. So YES, the motors ran with the 12v battery ground connected to the controller ground BUT one of them did not run properly when under acceleration and the trike would consistently pull to the left. That situation has been vastly improved with the re-flashing of the controllers for pure current mode, but I'm still getting a bit of it on occasion.

I'm done for the night but tomorrow I hope to wire things up in a number of different configurations and then draw up the wiring sketches for each, with ALL connections identified, and the results of each wiring configuration (whether motors run or not). With what I've provided so far it requires some shooting in the dark. So hopefully tomorrow I can put this all in sharper detail and provide some better (more complete) info to go on.

 

[amberwolf]

I don't quite understand what "lifting" is

Basically it sounds like you have the right idea.


Specifically, ground should be, in a system with no negative voltages, the point at which all other voltages would measure positive from. It'll always measure 0V relative to any other ground point in the same system, and normally would also be 0V relative to the battery negative.

When a ground is "lifted", it means it's no longer 0v, it's some positive value. (it could also be pushed negative in a system with negative voltage, but yours shouldn't have that).

This happens when some positive voltage source (lighting battery) separate from the ground's own votlage source (traction battery) gets connected into the ground, in a way or at a point that has sufficient resistance between the ground's own voltage source (traction battery negative) that a current flowing for whatever reason, generates a voltage across the resistance.

It can also happen if there is a large enough current flowing thru the ground, and the ground's connection to battery negative is simply insufficient to handle that, and it's own resistance creates a voltage across the resistance...so ground isn't really ground everywhere. But that shouldn't be the cause, since the disconnection of lighting ground from traction ground at that point makes things work correctly.

I'm done for the night but tomorrow I hope to wire things up in a number of different configurations and then draw up the wiring sketches for each, with ALL connections identified, and the results of each wiring configuration (whether motors run or not). With what I've provided so far it requires some shooting in the dark. So hopefully tomorrow I can put this all in sharper detail and provide some better (more complete) info to go on.

I'll be waiting to see what you find out.

 

[teklektik]

One of the grounds on the ground buss is from my auxiliary 12 volt battery.

This is sort of a red flag...

Ideally, an EV 12V aux system should be isolated from the traction battery. Sometimes for smaller (low voltage) vehicles this is not done because isolated DC converters are hard to come by inexpensively, but with a separate battery there is no reason to violate this isolation strategy.

If the 12v aux system shares a common GND with the traction battery then the 12V aux system should not use chassis ground. This is a de facto legacy wiring practice for gassers that should not be followed for lighting or any 12V accessory in this common ground situation. Even if there is no common GND with the traction battery, keeping the chassis neutral is arguably desirable in part because it renders many common chassis shorting failures benign instead of catastrophic.

• If your chassis has a live ground, then it may possibly be that your driver motor has a phase short to the stator which is making its way though the stator, to axle, through the swing arm bearings, to chassis ground, to the 12V aux system, and finally to the common traction battery GND to semi-short that controller phase. With your revised wiring the controllers may be seeing a high phase current and shutting down. Such a motor phase short to stator is clearly undesirable but has no operational effect with an isolated ground - not so in the chassis/common ground case. Other possible wiring phase shorts could have similar effects.
  
  Checking the controller blinking LED error code may give a hint as to why they are shutting down - and may possibly reveal a 'phase wiring' error code which would support the shorted phase idea.

If I disconnect the ground from that [auxiliary 12 volt battery] battery then the wheels will operate....

This seems a plain indicator that you have a traction/aux 12V ground issue. Even if you do not have a live chassis ground you clearly have issues with unwanted traction current finding its way though your accessory wiring.

At this point, I would defer mapping out your wiring for analysis. Instead take the time to separate the 12V aux system from your traction system in all respects. With isolated grounds, it becomes optional to eliminate a live 12v chassis ground if that is presently in place. This will get the 12v system out of the picture and the need for a lot of wiring analysis will go away. From your report above this will go directly to remedying your immediate controller issue regardless of whether the cause is a phase short as postulated. If problems persist, the traction system will be simplified and wiring diagrams, etc will not be as complex.

If things get working, it would be good to then investigate possible phase shorts, etc, but without the common ground, the matter may not materially interfere with operation and you at least will have the option of addressing it or not (i.e external wiring or inside motor).

So - some suppositions here, but maybe there is something in the post that will pan out....

 

[cboy]

At this point, I would defer mapping out your wiring for analysis. Instead take the time to separate the 12V aux system from your traction system in all respects. With isolated grounds, it becomes optional to eliminate a live 12v chassis ground if that is presently in place.

Sounds like a good plan. Just to be clear, there are actually two 12V systems. One runs off the 12V circuit provided by the controllers and the other runs off the 12V auxiliary battery and the dc/dc converter (thus far I have not turned on the converter and have just been running off the 12v battery). Am I correct that the 12V activities that are powered off the controller 12V output pin (such as the reverse switch) are NOT separated out. Instead these should ground back to the controller and/or directly to the shunt, correct? As I understand it what I need to isolate are any things (like the lights, turn signals, horn, etc.) that run off the auxiliary 12V system and not off the controller provided 12V system.

I think I can accomplish this separation without a huge amount of work. But down the road it seems I would want at least the dc/dc converter grounded to the shunt/battery pack in order for the CA to measure and record that amperage usage. But we can cross that bridge once I get the systems isolated and have a better idea of where the problem lies.

 

[teklektik]

...there are actually two 12V systems. One runs off the 12V circuit provided by the controllers and the other runs off the 12V auxiliary battery and the dc/dc converter (thus far I have not turned on the converter and have just been running off the 12v battery).

Am I correct that the 12V activities that are powered off the controller 12V output pin (such as the reverse switch) are NOT separated out. Instead these should ground back to the controller and/or directly to the shunt, correct?

As I understand it what I need to isolate are any things (like the lights, turn signals, horn, etc.) that run off the auxiliary 12V system and not off the controller provided 12V system.

Yes on both counts - that's the idea and is what I would recommend.

But down the road it seems I would want at least the dc/dc converter grounded to the shunt/battery pack in order for the CA to measure and record that amperage usage.

Yep. The CA presents some well-known grounding challenges - particularly for high voltage vehicles (not your case). I'm guessing you have a live chassis Gnd hooked to the converter and that may be involved in the present issue. That can be addressed.

• I'm not claiming to have the specific answer to the peculiar behavior you are experiencing, just recommending smallish changes that 'may' cure the symptoms, but will certainly simplify/reduce the electrics and connections that are in play. This is useful in its own right but will also help if additional debugging is needed. Simpler is nice...