2WD trike pulling left when accelerating

[cboy]

The "Whack-A-Mole" continues...

I now have the controller 12V wiring (I'm calling it C12V) and the auxiliary 12V battery wiring (call it B12V) totally separated. In addition I have the B12V system totally shut off - the 12V battery negative terminal is totally disconnected from everything and the positive terminal is switched off at the relay. Below is a sketch of the wiring. The main 4awg positive wires are now identical lengths and the 4 awg ground wires are now identical lengths. In this configuration the motors run. HOWEVER, the trike now pulls hard to the RIGHT on acceleration (this is opposite of the previous situation) and it pulls right consistently.

A little addition explanation of the wiring:

C/P is controller passenger side
C/D is controller driver side

The C12V is connected from ONE controller (the passenger side) to a fuse and then to five switches on the dash. These switches each have an LED indicator light and each switch has a ground pin to allow the LED to come on when the switch is on. Four of the switches feed power to the coil side of four relays. These four switches are all in the off position. The fifth switch is the forward/reverse switch. This switch feeds C12V power to pin 12 on the two controllers for forward and pin 14 on the two controllers for reverse. This switch also has an LED indicator light which has a ground pin. This switch must be in either the forward or reverse position for the motors to turn. The ground pins for all five switches are connected to a small buss and the buss, at present, is wired to the shunt. The B12V ground buss shown in the sketch is NOT connected to the shunt or to either of the main controller ground wires and the 12V battery is not connected to the buss.

Highly frustrating. I seem to be back where I started, just pulling in the opposite direction.

 

[teklektik]

This trike has seriously bad juju.
Just time for a quick note, but try running autotune (id program) again now that power environment has been altered. May not have material effect, but a fresh start without old baggage can't hurt...

 

[cboy]

This trike has seriously bad juju.
Just time for a quick note, but try running autotune (id program) again now that power environment has been altered. May not have material effect, but a fresh start without old baggage can't hurt...

Yup, tried that. And it comes up "85" which means the autotune was successful. I also started tinkering with the phase current limits, this time on the driver side. I can get the bike to run straight at 62%. So I guess we might claim a small victory on that note. Earlier I could only get it to run straight at around 50% current limit.

I really thought Amberwolf was onto something with the current "lifting" theory. Because that is how it feels when you are on the bike. It is like one wheel is being robbed of current and if you let off on the throttle just a wee bit the slower wheel will "catch up" and then run even and straight even if you hit the throttle as hard as possible after that. And it is also intriguing to me that with this new wiring set-up the problem has shifted from one side to the other. It's like that must be a clue of some sort...I just can't figure out what it is telling us.

At this point, however, I just don't see how it can be the 12V system interfering with the 72 volt system. I have now disconnected everything from the 12V side of things including the battery (bolt plus and minus terminals, all grounds to the buss bar, and all fuses except those on the 72 volt side. I have also disconnected all three wires to the dc/dc. I don't believe it was ever grounded to the frame but if it was...it's not now.

 

[wturber]

Time to "frog" the controllers (swap positions) and see if the trouble follows? Possibly do the same with the motors.

I know you did this long ago, but things have changed since then...

 

[teklektik]

And it is also intriguing to me that with this new wiring set-up the problem has shifted from one side to the other. It's like that must be a clue of some sort...I just can't figure out what it is telling us.

With the 12v system disconnected and the controller wiring reduced to the bare minimum the bike pulls to the right - differing from the controllers just shutting down and refusing to power the motors. With the ground cabling squared away either the driver motor is now running as it should and faster, or some new gremlin has popped up and the passenger side is now mysteriously running poorly. I'm being optimistic and voting for the driver side being cured rather than the coincidence of something new becoming broken...

At this point, however, I just don't see how it can be the 12V system interfering with the 72 volt system. I have now disconnected everything from the 12V side of things including the battery (bolt plus and minus terminals, all grounds to the buss bar, and all fuses except those on the 72 volt side. I have also disconnected all three wires to the dc/dc. I don't believe it was ever grounded to the frame but if it was...it's not now.

The whole episode with the ground wires made no apparent sense and suggested that traction current was running through the 12V circuitry somehow. Something pretty serious was going on to shut down the controllers. Controller error codes might have been useful, but this 'fix' of removing the 12V system doesn't really fix anything in particular, it just takes the offending 12V system out of the equation and removes the symptoms of the failure.

Hopefully what you are seeing now is some difference in the motors that was previously masked by the 12V problem.

I also started tinkering with the phase current limits, this time on the driver side. I can get the bike to run straight at 62%. So I guess we might claim a small victory on that note.

Actually, I think this is not a bad thing and reflects a good resolution, but just winging it here, so...

Harkening back to page four when the mode was assumed to be torque mode, you were running tests to examine the current draw to see what the max phase amps for each really were. This was to obtain the 'Max phase current' draw per motor. Hopefully, this is where you are again - balancing motor differences.

Stepping back a minute to look at the limiting process: By it's very nature the idea of torque throttle suggests that the phase current that the motor would draw if it could is being restrained by the controller according to the throttle setting. With the default crazy high 380A phase limit, the motors are essentially unlimited (there is no 'torque throttle' in play) and the performance you see is the basic motors running flat out without restraint. If the motors are perfectly matched we would expect them to develop the same torque and everything would rocket away straight as an arrow. In that unfettered situation, the motors will be drawing some unknown max phase amps and we are pretty sure that it's less that the giant 380A figure. This means you can reduce that 100% of 380A max phase amps setting quite some way before it actually has any effect.

The 62% figure you determined empirically sets the phase current limit at 380A * 62% = 235A which is about 3 times the max battery current limit of 80A. Presumably, this tones down the left motor to more or less match the torque of the other one running flat out - which certainly draws less than the full 100% of 380A. This suggests that you could set the right motor to 'about' the same value and things would work pretty much the same in your WOT off-the-line experiment. Here you are balancing the max phase currents to accommodate the actual L/R drive differences (basically reducing the passenger phase amps until it starts to pull left - this is where that side is actually being limited instead of running flat out).

We don't really know how the Kelly maps the throttle, but I expect it maps throttle rotation 0-100% to 0-100% of (Current Percent)*(PhaseCurr Max). I would think that setting both 'Current Percent' values to the minimum to give balanced operation (i.e. no unachievably high phase amps) is going to make the L/R scaling at lower throttle settings map to more closely balanced torque. For example: set L/R 'Current Percent' to a realistic 62% and 70% max values instead of 62% and default 100%.

Also, the lower you set the max phase amps, the more headroom you give the controllers to control torque at higher speeds. This happens because motor torque falls off as rpm increases. This means that at higher speeds the torque throttle will eventually be commanding a torque (phase current) that cannot be achieved.... So, setting the max phase current to a lower value extends the rpm range where torque throttle can be in play. If you experience pulling at higher speeds, you may need to drop the matched 'Current Percent' settings somewhat lower.

Anyhow - I think a reasonable next step is to back down the phase amps on the passenger side until it begins to have some noticeable effect just to get the throttle scaling equalized as mentioned above. This might turn out to be 70% with the difference from the driver side being the actual L/R difference not the present 100% vs 62%.

The C12V is connected from ONE controller (the passenger side) to a fuse and then to five switches on the dash. These switches each have an LED indicator light and each switch has a ground pin to allow the LED to come on when the switch is on. Four of the switches feed power to the coil side of four relays.

Am I correct that these relays are powered by the aux 12V system not the C12V supply? If there is any electronics on the 12V side that is powering the relays, then the relays should have snubber diodes to avoid killing the sensitive parts with high voltage spikes.

 

[cboy]

Some more head scratching testing...but some of it in a good way.

I disconnected absolutely everything, including the final ground wire from the reverse/forward switch leaving ONLY the two ground wires from the controllers connected to the ground shunt. I then set both controllers at 100% phase current. And then I tested three wiring scenarios shown in diagrams A, B, and C


Wiring A, which in theory ought to be the best since all the ground wires are equal length, made the trike pull to the right.

Wiring B also made the trike pull to the right but maybe a bit less noticeably.

Wiring C is the original wiring configuration from way back when but NOW with all but the controller ground wires disconnected from the system. With this configuration the bike pulls straight.

While I'm ecstatic I may have found a solution witch allows the bike to accelerate straight while retaining equal (100%) phase current in both controllers, I'd really like to understand why in the world these three wiring configurations can make the bike operate so differently. These are absolutely the only wires I am altering...in fact there are no other wires whatsoever attached to the grounding system (except for the CA wires attached to the shunt). Just the two controllers. And to my untrained eye, there is virtually no difference in the way these three wiring scenarios ought to work. But the fact is, they do make the motors and/or controllers behave quite differently.

What I'll be doing next is finalizing the wiring to get the 12 volt system (lights, turn signals, horn etc) operating as an isolated system. Once that is done I can get back on the road for some lengthy testing to insure the "pull on acceleration" problem is totally eliminated. I'd still like to incorporate the dc/dc converter into the system...maybe try doing that without the battery in the system at all. But that will be another chapter to be addressed later. I'm keeping my fingers crossed that this may finally be the solution. Not a solution I totally understand. But a solution.

 

[Alan B]

I would wonder if all those high current cables are "good". They don't act like it.

 

[teklektik]

I would wonder if all those high current cables are "good". They don't act like it.

No, they don't.

What is screwy about this is that the shift in ground reference - however small - from introducing the jumper in Case C should reduce the apparent throttle voltage and torque to the passenger motor making the trike pull more - not less - to the passenger side.

Hey cboy-
If you're not exhausted yet swapping cables around, could you restore Case A with equal length cables then increase the driver controller "TPS Dead High" a bit - say from 80 to 85 and see if things go straight. This altered setting will reduce the apparent throttle rotation to the driver controller and so reduce the torque (or power) applied to the driver side.

And BTW - great news on your experiments - even if they are 100% befuddling...

 

[Alan B]

One other thought.

When a controller is reduced to get things "even" it might be interesting to try reducing the other controller to see if it has any effect. The top of the range is probably not doing much, but reducing both might produce better throttle span, or have some interesting effect, perhaps leading to a more nearly equal setting of both controllers but less than 100%. Or just confuse things more.

 

[Alan B]

I would physically test those cables, pulling hard on the conductors, etc to see if there's a "faulty" crimp, or internally damaged conductors or ???

I would also measure them with a milli-ohmmeter to see if they are the resistance they should be, but most people don't have that meter.

If the battery or motor currents strayed into the wrong pathways they may have damaged traces in the controllers and cause some havoc. It is easy to have problems like this when the controller has multiple grounds, each signal ground is subject to damage if it presents a lower impedance path than the main battery ground. So any problem with the external power grounding can cause damage that may be hard to figure out.

 

[teklektik]

cboy-
Reviewing your earlier schematics, the only thing that concerns me a small bit is tying together pin 20 on the two controllers. This appears to be the only case where you bridged controller 'RTN' pins. This will cause a small bit of motor current to flow between the two controllers and through Gnds in the PCBs in each side. By design we expect that the Gnds in the two controllers will be similar but different and to work independently according to their own GND reference voltage. Tying the RTNs together tries to make these two different GND references equal by having some current flow between the controllers via this connection.

This may be inconsequential, but if it's not too much of a PITA, perhaps you could disconnect one of the pin 20 connections. To be honest, I don't expect any effect, but this is a connection I would not make and we seem to be grasping at straws to locate the cause of this weird behavior. That said, if you just want to get this on the road, no worries...

 

[teklektik]

The top of the range is probably not doing much

Raising the upper voltage limit will have the effect of stretching the working voltage range so an unchanged input throttle voltage will appear as a lower percentage of the new increased range. This will directly reduce the torque or power from that controller depending on the limiting region in which the controller is operating.

 

[cboy]

... could you restore Case A with equal length cables then increase the driver controller "TPS Dead High" a bit - say from 80 to 85 and see if things go straight. This altered setting will reduce the apparent throttle rotation to the driver controller and so reduce the torque (or power) applied to the driver side.

Might as well add that to the list of experiments on this beast. Could be another tool for fine tuning things. I'll try it in the morning.

 

[teklektik]

Excellent. Thanks!

 

[cboy]

Reviewing your earlier schematics, the only thing that concerns me a small bit is tying together pin 20 on the two controllers.

Tying together these two pins comes directly from Fany. I couldn't get the regen to work in both controllers and she said the two grounds need to be tied to accomplish that.

 

[cboy]

I would physically test those cables, pulling hard on the conductors, etc to see if there's a "faulty" crimp, or internally damaged conductors or ???

I have some new lug ends being delivered tomorrow and will make up totally new cables. So I can test the theory a bit that way as well.

 

[cboy]

Another observation I've made over the past couple days is that after shutting down the motors, turning off the ignition and turning off the emergency button to disconnect the battery pack, I have voltage remaining in the system of 50V or more. This voltage steadily declines but takes a few minutes to reach zero. I assume this is the caps. But it is something I had not observed previously and I only started testing for it when I saw a little spark while moving around some wires for testing. Is this normal?

 

[teklektik]

Tying together these two pins comes directly from Fany. I couldn't get the regen to work in both controllers and she said the two grounds need to be tied to accomplish that.

Yep - and that wasn't bad advice.

An alternative would be to tie the regen throttle Gnd to either the CA GND or the common Gnd where the CA is grounded. This would give a GND reference without cross-connecting the controller RTNs. Each controller RTN is already connected through it's existing B- to that common GND point.

Similarly, stealing the 5V for the regen throttle power from the CA instead of from one controller would make the regen voltage independent of either controller as well. Since the regen control is on the bars with the CA anyway, snatching the 5V and GND from the CA would leave only the single regen signal wire to run back to the controllers. This would then be the exact analogy of the means by which the CA throttle signal and GND work with both controllers without tying RTNs together or getting power from one controller or the other - same problem exactly.

Again, this is likely not the problem here, but given two ways to wire this from scratch, I would go with the independent power supply and ground strategy from the CA to get consideration of controller interactions off the table - just because I'm a little paranoid about ground loops...

 

[teklektik]

I assume this is the caps. But it is something I had not observed previously and I only started testing for it when I saw a little spark while moving around some wires for testing. Is this normal?

Yep. In fact, if you rummage around in the Kelly manual they tell about bleeding off charge before tinkering.

It may be that you never saw this before because you just didn't notice or because some of that 12V aux stuff you've disconnected was sucking power and running the caps down pretty quickly on power off (DC converter or the mystery wiring gremlin).

 

[cboy]

It may be that you never saw this before because you just didn't notice or because some of that 12V aux stuff you've disconnected was sucking power and running the caps down pretty quickly on power off (DC converter or the mystery wiring gremlin).

Makes sense that the charge might have been getting bled off. I'll have to look for the reference in the Kelly manual.

 

[cboy]

... could you restore Case A with equal length cables then increase the driver controller "TPS Dead High" a bit - say from 80 to 85 and see if things go straight.

More good but confusing results. I returned the cables to diagram A and then set the TPS Dead High to 90 (just to see if it would have any effect at all). And it did, the trike went straight. So came back and set it at 85 per your original suggestion. Bike ran straight. So just to put a final exclamation point on this as a solution, I set the Dead High back to it's original 80%. And to my chagrin, the bike still ran straight. The only conclusion I can come to is that on alternative Mondays, Wednesdays and Fridays the bike runs straight for the two weeks following a full moon. Otherwise, it runs crooked.

Actually, I'm beginning to doubt my own sanity a bit. But at least I'm happily insane...the bike is going straight. Granted, these results are based on just a couple runs up and down my street. We'll see what happens when I start cranking up the miles. Or when the moon goes out of alignment.

 

[teklektik]

 

[teklektik]

The reason there wasn't any change by adjusting the 'TPS Dead High' settings is that the phase current is still ridiculously high (unless you backed the 'Current Percent' settings down in secret). So, WOT is commanding 380 phase amps and any throttle setting above 65% rotation is probably going to give identical results since the achievable max phase current seems to only be around 250A. The small throttle change I suggested would have no effect because it's effectively moving the throttle just a tad in a giant dead zone. (My Bad on this experiment.)

That said, I'm with you on the other business with the previously inoperative cabling scheme suddenly working - baffling...

Anyhow, for now it's doing what you want, so....

 

[cboy]

Now that the "bare bones" system seems to be working I am attempting to get the peripheral systems back in operation. And I ran into an immediate snag. As you may recall my switches (five of them) operate off the 12V supply provided by the passenger side controller. Four of the switches are connected to relays and the fifth (forward/reverse switch) is wired to the controllers without a relay. Each switch has a ground to run its LED indicator light and each relay coil has a ground. I wired things up as shown in this sketch (I cut out a couple switches/relays to simplify things). The problem is that when I connect these grounds at the shunt it sometimes (but not always) results in the bike pulling to the right. So I think this is creating a potential "lifting" situation. I'm wondering if I put a diode in that dotted orange line would it possibly prevent the 72V current from wandering into that 12V system? As an alternative is there a way to wire the 12V supply from BOTH controllers into this switch/relay portion of the system. The bike is now pulling towards the passenger side which is the side supplying this 12V current.

 

[teklektik]

As you may recall my switches (five of them) operate off the 12V supply provided by the passenger side controller. Four of the switches are connected to relays and the fifth (forward/reverse switch) is wired to the controllers without a relay. Each switch has a ground to run its LED indicator light and each relay coil has a ground.

This matter concerned me earlier and I asked you above if the relays were being powered by the controller 12V supply, but the question may have been lost in the post...

Am I correct that these relays are powered by the aux 12V system not the C12V supply?

However, from this diagram and your description it seems the controller is powering the relays - which is not good.

The problem is that when I connect these grounds at the shunt it sometimes (but not always) results in the bike pulling to the right.

I'm certain the problem is that that the relays are dragging down the passenger (right) controller power supply which was only sized to power the controller, not other stuff. If those are conventional automotive relays they are probably drawing about 100-150ma each. This means that with four relays engaged you are pulling between 400mA to 600mA from a supply that was designed to run a few throttles and switches that might draw a total of 10ma. (You could check with Kelly on this, but I'm very certain about not having an extra 1/2A capacity.)

The reason the bike is pulling to the right is almost certainly because the controller electronics are failing to operate properly and only partially driving the power FET output stage.

So I think this is creating a potential "lifting" situation.

I think 'lifting' is AW's terminology for a shifted ground reference - but that's not it. With the relays powered from the controller, the controller is just power-starved and is bungling control of the motor power.

I would recommend using 72V forklift relays like JD2912's and just power them off the traction battery. These have the familiar 'C' (SPDT) configuration. You can still run the LEDs off the controller supply. Alternatively, you can power the relays from the aux 12V supply which might be easier in your present situation although it entangles the switching logic with the 12V system. But - if the relays are involved in the 12V stuff anyway then that's so bad and are not really 'entangled'...

Also, the relays should have snubber diodes to avoid killing the sensitive parts with high voltage spikes. No necessary if there are no electronics being powered by the relay supply...

• FWIW: I use a 48V relay in a similar situation on my cargo bike to switch a DC converter with a wicked startup current surge.