2WD trike pulling left when accelerating

teklektik said:
Please note the CA indicated current draw at WOT for each motor when you do it.

I'm not totally sure how to do that. The CA reads the current draw for the entire system...both wheels. Are you saying I should disconnect one wheel while testing the other? I think I could do that by simply disconnecting the throttle wire from one while testing the other. Would that do the trick?

teklektik said:
The controllers have been switched, the pole count doubled, and now the torque problem is worse than before. This seems to indicate the controller is, in fact, involved somehow.

Or there is something going on inside the wheel/motor that is progressively getting worse. I suppose I could switch back the controllers, return to the initial pole count and see if the 50% current limit would then have an effect. That might tell us SOMETHING.

I did make one other minor change in the programming in the "Motor Nominal Current" setting. (this is the first item on the "Motor" programming screen). This setting is used during the identification angle (phase mapping) process. The default setting is 80 and the manual says to increase it for higher power motors and decrease it for lower power. The scale is 0-1000. I reset it for 200 as a "best guess" of where my 3000 watt motors might stack up. Didn't seem to help in any way but if we want to get back to "original settings" I should change this back to 80.
 
cboy said:
teklektik said:
Please note the CA indicated current draw at WOT for each motor when you do it.

I'm not totally sure how to do that. The CA reads the current draw for the entire system...both wheels. Are you saying I should disconnect one wheel while testing the other? I think I could do that by simply disconnecting the throttle wire from one while testing the other. Would that do the trick?
Yep - that would work fine. Less scary, too...

Yes to your other remarks as well. That setting was one of the things I mentioned that might be tinkered. But for now we need an 'apples to apples' comparison....
 
Test - throttle, wheel speed and amps at WOT

My earlier test of throttle and wheel speed only went to about 80% throttle (146 on the Kelly Monitor scale of 0-255) Here are the test results at WOT (186 on the Kelly scale)

Controller D (driver side)
Throttle input 186
Wheel speed 780
Amp draw (CA3) 4.4 - 4.5 (varied between these while holding throttle)

Controller P (passenger side)
Throttle input 186
Wheel speed 782
Amp draw (CA3) 3.8-4.0 (varied while holding throttle)

Is that a significant difference in amp draw?

[Note: during earlier testing I had set the "pole pairs" at 16 in the kelly program. That was an error and the setting for this test was 32. This accounts for the differences in wheel speed between this test and the earlier testing.]
 
The currents are about 10% off - which is a little odd - but it's not the huge difference we might expect from a false positive phase setup but it's not really identical either. My interpretation here is that the phase/hall setup is okay, but there may be some differences in the motors. Ideally, this unloaded WOT test would be done with the same controller to get that out of the equation, but we were going for grosser discrepancies at the time.

Here's the dyno dope for the V3 3000W 40H 273 version which may be what you intended to purchase:


QS_16inch273-40HV3-3000w_dyno_annotated.jpg


This matches the passenger motor pretty accurately - but the other is off. The driver side motor seems to want an extra 10% of current to do the same rpm which is interesting since that is the motor that you needed to run at 100% and limit the current on the passenger side motor.

Looking at the QS information, it seems they have a number of closely related motors that are off by about 10% in speed/torque. These are different motors with very close torques and Kvs that would certainly misbehave together if mismatched in your side-to-side configuration. I'm not sure what you ordered, but perhaps you can you locate and compare stamped part numbers to verify (apparent) the sex of the motors. This doesn't speak to the actual guts of the things, but this might be something to investigate with QS. I think it would be worthwhile (albeit a PITA) to rerun a WOT test on the driver motor with the passenger controller. This could give you some solid data to pass along to QS if there is indeed a mismatch. Just a thought...


QS_16inWheel273_Versions.png
 
Another experiment.

To try to determine why the "limit phase current" is no longer getting the result it did earlier (the trike accelerated straight) I returned everything to how it was when I was getting that result:

I swapped the controllers back to their original motors
I programed them back to 16 pole pairs (the correct is still 32 but it was 16 when it worked)
I returned the setting for "motor nominal current" back to 80 (the default setting)
I set the passenger side for 50% phase current limit
And then I ran the angle ID program.

The bike pulls left.
 
teklektik said:
I think it would be worthwhile (albeit a PITA) to rerun a WOT test on the driver motor with the passenger controller. This could give you some solid data to pass along to QS if there is indeed a mismatch. Just a thought...

I can do that without much trouble now since I just swapped the controllers back to their original locations.
 
Have you tried taking the CA out of the throttle system and wire a simple hall throttle to both controllers signal wires using only 1x power input

Seems like the only thing you havent tried, also make sure the correct throttle type is selected in program settings
 
skeetab5780 said:
...taking the CA out of the throttle system and wire a simple hall throttle to both controllers signal wires...
...make sure the correct throttle type is selected in program settings
Some thoughts...
  • Both controller throttles are already being driven by the single CA Throuttle OUT voltage which is displayed on the CA Diagnostic Screen. Since the CA is common to both motors and the issue is a motor performance difference, changing to a different common source doesn't seem to lead anywhere.
  • The CA can be 'taken out of the system' by selecting CA Bypass throttle mode - no need to make wiring changes. What comes in, goes out exclusively with no changes.
  • The proper CA throttle mode for torque controllers is PassThru mode. What comes in, goes out after scaling and ramping. Output is affected by other functions like autocruise, PAS, as well as LVC, Amp, Power, Speed, and Temp limiting.
 
skeetab5780 said:
Have you tried taking the CA out of the throttle system and wire a simple hall throttle to both controllers signal wires using only 1x power input

No, haven't done that. But your post sent me off to revisit the throttle wiring. In looking over the wiring diagram Kelly provided for hooking the CA3 throttle wire to the controllers (see below) I noticed something I hadn't paid attention to before. They have a diode in the throttle wire between the CA3 connector and the #3 pin on the controller. But the diode points back to the CA3 not toward the throttle pin. So I don't understand what the diode is supposed to do, is it necessary, and might its absence be the cause of the pulling right on acceleration? It seems to me, as drawn, this diode would prevent throttle input from ever reaching the controller. So can someone explain the electronics of this. And to repeat, I do not have a diode in this position in the throttle wiring. I DO have a diode in the wiring from the throttle itself to the CA3 but my understanding is this is simply to lower the voltage a bit so the CA3 will correctly identify any sitiuation which might cause runaway throttle.

Kelly-KLS-CA3-diode.jpg


Edit: I just took a look at the Kelly diagram for wiring multiple controllers to a single throttle (not a CA3) and they don't include any diodes in the wiring. Is there something about the CA3 signal that would require this diode?
 
cboy said:
In looking over the wiring diagram Kelly provided...
Gee - I can't find squat from Kelly. Where did you get this extra stuff? email?


cboy said:
...for hooking the CA3 throttle wire to the controllers ... They have a diode in the throttle wire between the CA3 connector and the #3 pin on the controller. But the diode points back to the CA3 not toward the throttle pin.
...
Edit: I just took a look at the Kelly diagram for wiring multiple controllers to a single throttle (not a CA3) and they don't include any diodes in the wiring. Is there something about the CA3 signal that would require this diode?
The illustration is for the old old CA2 hookup (pre-rev 12 PCB) where the CA did not supply throttle voltage but could only limit the operator throttle clamping it lower. Hence the diode to allow the CA to sink current but not source it. Later CA2 models have the diode internal. On the other hand, the CA3 sources current and so normally has no diode at all. So - confusion from the general moniker of 'a CA' where there is clear difference between how a CA2 works and a CA3 works. This is illustrated in the [strike]Un[/strike]official Guide section "Appendix C. Summary of CA-DP Wiring for Controllers with Old/New Interface Types".


cboy said:
I do not have a diode in this position in the throttle wiring. I DO have a diode in the wiring from the throttle itself to the CA3 but my understanding is this is simply to lower the voltage a bit so the CA3 will correctly identify any sitiuation which might cause runaway throttle.
Yes, these are as they should be.
 
teklektik said:
Gee - I can't find squat from Kelly. Where did you get this extra stuff? email?

Yes, Fany emailed it as a follow up to a question I sent her.

teklektik said:
The illustration is for the old old CA2 hookup (pre-rev 12 PCB) where the CA did not supply throttle voltage but could only limit the operator throttle clamping it lower.

Doesn't sound like the absence of this diode is related to the problem then. Shoot. I was hoping we had stumbled on something.

Oh, I tried the "Aspirin Test" this morning. To see if I could visually detect one wheel producing more torque on acceleration I put the trike up on blocks and put an aspirin on the top of each wheel. Then I'd jerk the throttle immediately to WOT. Nothin'. Couldn't see a lick of difference in the trajectory of the aspirins. So I ate them instead...

Also tried marking the tires with tape, starting marks in identical positions and then hitting the throttle hard but only for an instant. The marks appeared to move at exactly the same time and same rate. So I ate two more aspirins.
 
Test results

I don't think this shows much beyond what we already have but it is testing with the controllers swapped back to their original positions, pole pairs set at 32 and both controllers programmed for 100 phase current. So it is sort of my benchmark for numbers. WOT is at 186.

test-data-8_17_18.jpg
 
cboy said:
I tried the "Aspirin Test" this morning.
...
Couldn't see a lick of difference in the trajectory of the aspirins. So I ate them instead...

If you had done the Valium test instead, you would have gotten the same results, but you wouldn't care....


Okay - good to pursue the 'due diligence' on the Kv testing.

So - before you reported:
Driver Motor
  • Throttle input 186
    Wheel speed 780
    Amp draw (CA3) 4.4 - 4.5

With the same controller you report for the other motor:
Passenger Motor
  • Throttle input 186
    Wheel speed 785
    Amp draw (CA3) 4.1

Using the same controller, the motors look pretty close but a still a bit different. It's also encouraging to see the controller change had a pretty minor effect - which is good.

I'm pretty puzzled why reducing the passenger phase amps by that huge amount didn't affect the torque unless the actual max phase amps for that acceleration test were very much lower than the pre-configured max - which is 380A - pretty huge. I'm wondering if setting the %phase amps down to a lower number (effectively reducing the 380A value) and then attempting to balance the motors by a small offset in phase current on the passenger side would work out....

So - one way to go at this is to actually get a handle on what the phase amps are for some common situation:
Does the Kelly config app in monitor mode show phase amps? Can you collect phase amp data from each motor while doing your 'accelerate and veer from a dead stop' maneuver with both controllers at 100% phase amps? Maybe with a laptop or the Android app....

Another way would be to reduce the %phase amps for both motors until a perceivable performance reduction occurred, then consider that value + a bit to be the 'measured max' for that particular test scenario.

In either case - next, (still experimenting), try backing off the Passenger %phase amps by about 10% as a start, etc. to try to suppress the veering by balancing the torque. Here you know for certain that your reduction would actually reduce 'usable' phases amps instead of just reducing a fictitious upper value beyond that which the motor actually could pull. I'm thinking that your actual max phase amps might be as low as 250A or less - not 380A.

I have no idea what's up with the "it worked before but now it doesn't" issue, but moving on seems more productive than sitting and musing...
(Boy, this problem is a PITA...)
 
teklektik said:
I have no idea what's up with the "it worked before but now it doesn't" issue...
I occurs to me that perhaps the controllers were tuned to a false positive before with the original 16 pole setup instead of the proper 32 poles making the phase amps improperly large. In that scenario reducing the phase current by 50% had an effect (190A from the max of 380A). Now the autotune might be correct, substantially reducing the phase amps and making the original 190A reduction ineffective (e.g. it might take more than 50% to get in the new 'correct' phase amp range). This doesn't necessarily explain the intermediate tunes back to 16 poles but just looking for dots to connect...
 
teklektik said:
Does the Kelly config app in monitor mode show phase amps? Can you collect phase amp data from each motor while doing your 'accelerate and veer from a dead stop' maneuver with both controllers at 100% phase amps? Maybe with a laptop or the Android app....

The monitor does show phase amps (see right hand column). Unfortunately the numbers only flash there momentarily and they are sort of all over the place. I would say it is unusable data. Certainly can't be used to compare throttle input, motor speed and phase current. And the software program does not allow for capturing the numbers or logging them onto the computer laptop...at least not to my knowledge. I've tried to search this on the web and only found one or two vague reference of some user who was going to attempt to capture...but there was no follow up on how to do it. I've even thought of using a video camera running at high speed to capture and playback in slow motion so you could see throttle input, motor speed and phase current at a readable pace and side by side. Perhaps I should really make an effort to do that. If it works, it might be a usable testing tool for others.

monitor-screen.jpg
 
teklektik said:
In either case - next, (still experimenting), try backing off the Passenger %phase amps by about 10% as a start, etc. to try to suppress the veering by balancing the torque. Here you know for certain that your reduction would actually reduce 'usable' phases amps instead of just reducing a fictitious upper value beyond that which the motor actually could pull. I'm thinking that your actual max phase amps might be as low as 250A or less - not 380A.

If I understand you correctly, you are suggesting adjusting down the "Phase Maximum Current" box as opposed to adjusting down the "Phase Current %". If that is the correct, it can't be done. The Phase Maximum Current box is grayed out (see screen shot below), set by the factory based on the controller identified in the software. So all you can do limit the current as a percentage of that maximum. If you did mean to cut back the "Phase Current %" then that is what I have been doing. I have cut it all the way to 50% with little effect on the pulling (except early on when it DID have an effect) Given all that, I do think you idea of limiting BOTH controllers, say to 50 or 60% and then start to make small adjustments on the passenger side to reduce the percentage even more might accomplish the same end. But even if this did achieve some balance in torque, I would think it less than ideal. I just have a gut feeling it would only be masking a problem that might be getting worse. Fany had suggested very early on when I asked her about the availability of a "torque mode" on this controller that I might try adjusting down the Maximum Forward Speed % for the passenger side. I have not tried this because my thinking was this is a torque issue not a speed issue. When I accelerate I am nowhere near a maximum speed range. In fact just the opposite...I'm at or near the zero end of the speed range. Perhaps I am wrong in that thinking but limiting phase current seemed much more appropriate given the circumstances.

pass-vehicle.jpg
 
Too bad - that pretty much sucks.
If I was going to try to capture that I'd use a screen capture video app like Camtasia or SnagIt, but I'd have to be highly motivated. The Phaserunner is nice - it makes plots....

Best plan in the present situation would be to manually search out the max. Five tries or so should nail it pretty well using a binary search. Add or subtract 1/2 the previous increment depending on whether or not the new phase amps had an effect. For example, we start with a max of 100%:

example search where 'no effect' means 'off the line full acceleration unchanged':

  1. starting, so subtract 1/2 100 = 50%, so 100 - 50 = 50%. Try that.
  2. no effect, so subtract 1/2 50 = 25%, so 50 - 25 = 25%. Try that.
  3. yep had an effect, so add 1/2 25 =13%, so 25 + 13 = 38%. Try that.
  4. no effect, so subtract 1/2 13 = 7%, so 38 - 7 = 31%. Try that.
  5. etc...

Homes in PDQ.
 
cboy said:
If I understand you correctly, you are suggesting adjusting down the "Phase Maximum Current" box as opposed to adjusting down the "Phase Current %".
Sorry - I was talking about the effective Phase Amps, not the control to use. My Bad.

Yes - you would be adjusting the percentage. See the example above on how to find the true max to use as a baseline. Do it for each controller, then balance them from there by reducing side-veered-from...

It looks like there may not be too much difference between the motors, so this might work out okay. The searching business is sort of a hack, but...
 
cboy said:
Anybody know what the "Read Zero" button does? (see lower right of screen shot in post above)
I sort of assumed it was the zero-current calibration button, so you would press it when the controller was idle with zero throttle to get the internal zero current measurement offset correct.

If they didn't explain what this is, give that a try and see if the differences between the currents in the two controllers goes away... (here's hoping)
 
teklektik said:
Best plan in the present situation would be to manually search out the max.
...
example search where 'no effect' means 'off the line full acceleration unchanged':
...
I'm not sure what I was thinking when I wrote that particular post, but the procedure called out to find the true max draw for acceleration on the flat cannot work as described. My Bad. Since the idea is based on balancing the controllers after identifying a workable max, the bike will be veering for the whole ordeal making the described test procedure entirely unworkable.

I still think that balancing the motors from a lower nominal 'phase amp %' is a good approach, but the technique I laid out to do it was severely flawed.


cboy said:
Given all that, I do think you idea of limiting BOTH controllers, say to 50 or 60% and then start to make small adjustments on the passenger side to reduce the percentage even more might accomplish the same end. But even if this did achieve some balance in torque, I would think it less than ideal. I just have a gut feeling it would only be masking a problem that might be getting worse.
I agree to a point. There does appear to be some difference between the two motors. Whatever the cause (different magnet batch, etc) your tests indicate that it's somewhat affecting the transformation of current into torque/speed. So unless the motor difference can be remedied, if you pump the same phase amps into both motors, one is going to make more torque than the other. At some low level this difference may not be consequential, but I think it's unclear whether or not this difference at high current is responsible for the veering issue.

Since the controller is fundamentally regulating phase current then balancing at the phase amp configuration level seems to go after the imbalance the most direct way.

  • However, since the controller is mapping throttle voltage into phase current, the other adjustment that would indirectly have a torque-balancing effect would be to similarly rescale one of the throttle voltages. This would have the effect of changing the apparent % of true throttle voltage on one side and so produce a similar % change in phase current. This is going at it around the barn, but it should have the same effect as balancing the phase current directly.

    There are two ways to go implement the throttle balance approach:
    • adjust the controller Throttle Position Sensor (TPS) max settings so the applied throttle voltage would be interpreted for each controller as different percentages of
      (TPS Dead High) - (TPS Dead Low)
    • Add an external balance control to allow actual max throttle voltage(s) to be tweaked with a knob for on-the-fly veer-balancing.

Anyhow, the throttle stuff was just musing - the phase amp tinkering still looks like the best initial research plan just now.


cboy said:
Fany had suggested very early on when I asked her about the availability of a "torque mode" on this controller that I might try adjusting down the Maximum Forward Speed % for the passenger side. I have not tried this because my thinking was this is a torque issue not a speed issue. When I accelerate I am nowhere near a maximum speed range.
Yep, I agree with your assessment that speed adjustment does not appear to be useful. It would certainly be clarifying to have a description of how the controller actually works - torque with overriding speed limiting, some kind of balanced control, etc. Making too many decisions based on guesswork...[/list]

FWIW: Since the throttle rotation maps into phase amps, getting the max phase amps % down to a realistic setting for your motor should also have the effect of improving the resolution of your throttle since more rotation will be required to achieve a given phase current.
 
WHOA --- POSSIBLE BREAKTHROUGH
OOPS --- NOT disregard this post it was a false alarm. Throttle wire on passenger side was disconnected during the road test...that's why it pulled right and not left. Back to square one.

I'd been going over in my head, and on paper, what differences there might be between the driver side controller and the passenger side controller other than the hub motor they were attached too. And the only differences were slight variations in the wiring for each controller. So I started investigating and experimenting. And one of those "differences" was that the driver side controller was used to send a signal for the CA3 speedometer. The passenger side was not wired to the CA3 speedo. And the driver side powered the speedometer off a controller pin called the "Hall Copy Signal"...which I assume is just the signal off of one of the hall sensor wires. So I disconnect that wire from the harness and lo and behold the bike did not pull left. (Unfortunately, it now pulls a bit to the right...although not as bad as the prior left pull.) I don't quite understand electronically why this would occur but it appears that under acceleration and heavy load (road resistance instead of free wheeling) the CA3 speedo was hogging the signal from whatever hall sensor the "hall copy" is taken from it was interfering with phase current to the driver side wheel. I'm still trying to sort out why I am now pulling a bit right but at least I've found something that is directly linked to the problem. Hopefully this might trip off some additional deductive thinking from those following this thread.
 
Back
Top