So close but no movement

[FFMan]

Prius gen3 inverter, phev rear motor, installed in BMW E46 with BMW 360v pack V5.2.

So today was seeing if we could get it to move under it's own power. I retested the motor with the test box, and test batteries and it was able to spin the wheels. So replaced the test box with the proper ampseal and wired it into the car pedal, brake etc. All good though some trouble calibrating throttle pedal. My lesson learnt seemed to be potnom does not get proper values unless the inverter is in run mode. In waitstart state, pot1 can be plotted properly but potnom behaviour isn't as expected - is this normal ? I disabled all regen (I think) as part of this process.

Anyway, with the inverter running the test batteries (150v) appeared unable to move the car. It's three hoverboard packs in series, so basically a lot of 18650 cells. Interestingly UDC was reported as 427v when the multimeter gave the more plausible value of 358v.

So i hooked up the main pack and contactors etc. It's able to spin the wheels when off the ground, but on the ground i can feel the motor taking up the slack in the transmission but no movement.

Parameter file attached. I wonder if a while back i set some limit for testing. We do see current on IL1 & IL2 but only a handful of amps, and more throttle does not increase it.

At the end of a long day, i left it there, i suspect some parameter issue, i would appreciate any guidance on that one.

thanks

[Bigpie]

Throtcur of 1 will be 100 amps at 100% throttle. You'll certainly want higher. I've got 4.5

[Ev8]

Also worth upgrading to the latest foc software, much better than what your running now

[FFMan]

OK - tried a few things this morning but no change.

I upgraded to 5.24r - behaviour the same, i can spin them motor on the throttle in neutral (noload trace attached). I can spin the wheels in the air (wheels in air trace attached).

Do the il1 & il2 at 60-amps seem a little high for a no load spin. There was some handbrake drag which i freed off, and wheels turn ok by hand. If i lock one side and turn the other the prop turns without excessive force though naturally there is some drag in a cold diff and gearbox.

the param changes i made prior to retest where:-

lqminusld = 0.12
fluxlinkage=75 (though i note bigpie has 26)
curkp=25 (same as bigpie)
curki=200 (same as bigpie)
udcmax=400

and i set udcgain, so voltage reading is correct

I'll get this plot later as requested (A plot of iq, id, uq, ud and resolver angle could be useful too.) Is resolver angle just the 'angle' spot value ?

[Pete9008]

All looks OK, are those plots both with the wheels spinning in the air? Would be interesting to see ones when it is trying to move the car.

Yep, the angle spot value, just so that the amount of motor movement can be seen when its trying to move.

Could this be the wrong syncofs offset? There can be two nulls when tuning, 180degrees (electrical) apart. One generates good torque and one doesn't. Did you find both when tuning?

Edit - the second null should be around about a syncofs of 49268. (note, this is based on theory not experience so could be rubbish!)

[FFMan]

> are those plots both with the wheels spinning in the air?

the noload is in neutral, the other is in 2nd gear wheels in the air. Therefore some load difference but not much.

> There can be two nulls when tuning, 180degrees (electrical) apart.

hmmn - first i've heard of that. i don't recall it being mentioned in the foc tuning video but it was a few months back. You'll gather i didn't find two nulls, so i guess the one i have could be the wrong one. Do i find the other one using the same method, or (clutching at straws here), can it be calculated from the one I have ?

> the second null should be around about a syncofs of 49268.

i note bigpie is 46518 so maybe that is more in the ball park then.

do we know any other forum members with my setup ?

[Pete9008]

I think this one is similar https://openinverter.org/forum/viewt ... ?t=766

Should be possible to find both nulls the same way. I'd guess one would be more stable than the other but as mentioned I have no experience doing this in real life. Seem to remember some discussion of it in a thread somewhere but can't remember where now (might even be the one above?).

Edit - found it, bit of discussion starting here https://openinverter.org/forum/viewto ... p44623

[FFMan]

ok - so in ability to undo the crank nut on the wife's volvo to change the timing belt meant i had some time today to progress this.

I decided to go in search of the second null point. I tried on 5.24r as per the video, and i found the bulb lit at every syncof value, and i didn't feel confident enough to bypass the bulb. So i went back to 5.20 and results were as expected except...

The video describes finding the null point by searching for the point the motor does not turn with largest manualid, by binary chopping in until a best value is found. I did this on the bench before and ended up with syncof of 16500. Now however i get the following results.

There does appear to be two highs were max rpm is achieved, does this indicate correct value or am i still shooting for values of zero rpm on max current. I ask because my previous value of 16500 gives decent rpm now, though if you apply the 'max rpm' approach it's a little adrift of ideal.

This is all running on my 120v test pack.

Should i try syncoff of 55000 on my full 360v pack ?

[Pete9008]

Your 16500 syncofs value doesn't look right. It's the zeros that you are looking for so a syncofs of around 0 or 32768 would be more likely based on that data. A value of zero would also match the value mentioned here https://openinverter.org/forum/viewt ... p44764

I remember you mentioning swapping a couple of phases in your other thread, was that before or after measuring the 16500 value? Whenever you swap phases it's best to recheck syncofs as swapping phases can change it.

Edit - shouldn't make any difference which pack you use.

Edit2 - correction, based on those numbers I'd expect syncofs to be nearer either 5000 or 37768

Edit3 - Also be careful when doing this, there is nothing to limit the motor speed so looking for max speed is not to be recommended! Keep the currents low until you know where the zeros are and then increase gradually to hone it in on the final value while keeping the motor speed low.

[FFMan]

I did swap the motor phasing post install, but i thought this was to correct an inadvertent swap i made during the install. Maybe that was a red herring. I thought 16500 was ok when i did it but best to start from nothing again which is what i started to do today. Slightly thrown by the fact 16500 seems to be nearly the point furthest from where i should be but back to basics makes sense.



thanks Pete - i seemed to find a good null point of 38000 so in line with your edit 2 above.

then i tried to move the car on the full pack and whilst i got a tiny bit of movement i wouldn't say it drove. From the traces looks like there is a 100amp limit coming into play. Might it need more than this to pull away up a small incline ?

is the limiting param ocurlim in this case, it is set to 200, but i'm afraid the explanation in the wiki doesn't help me make sense of it 'Hardware over current limit. RMS-current times sqrt(2) + some slack. Set negative if il1gain and il2gain are negative.'

Still on 5.2, will update at next attempt.

[Pete9008]

It may be that you need the other null now, worth trying a syncofs value of 5232.

The trouble is everything affects everything else. The reference motor position for syncofs is with respect to phase 1 on the inverter. Therefore if you swap 2 and 3 it shouldn't affect the reference position but if you swap phase 1 with either of the other two it will. Unless you managed to exactly reverse the accidental phase swap you won't get back to the same setup. Swapping any two phases will reverse the direction but only swapping 2 and 3 will do it without messing up syncofs (and even then it might pull the value slightly).

The other complication is I think using the 38000 null may also reverse motor direction so might need to be done in collaboration with a phase swap. Worth trying 5232 first.

Edit - worth mentioning that using a syncofs value of 16500 when it should be 5000ish would explain the motor spinning but with very low torque previously. The field angles within the motor would have been way off so it's not too surprising it couldn't move the car.

Edit2 - no idea about the current limit, fairly sure the hardware limit isn't used on the Prius inverter (unless the same message is reused for something else?). Hopefully someone else can shed some light on that.

Edit3 - ignore edit 2, the 100A is just because you have throtcur set to 1 which gives 1A per % throttle, or 100A max. I would have thought that would be plenty to move the car if everything else is right (but it will need increasing at some point).

[FFMan]

> It may be that you need the other null now, worth trying a syncofs value of 5232.

agreed - i'll try that tomorrow

> The trouble is everything affects everything else.

yep, and they're the things i know about ! there are some things at play i haven't nailed down like 5.24r behaving oddly when trying to tune, there seems to be a fair amount of stiction in my drivetrain. takes some oomph to get it turning then it's easier kind of thing.

in terms of phase swapping, as the motor cables are moulded it would have been the outer most cables swapping position, so would that be phases 1 & 3 ?

With 38000 as my syncof the motor is reversed from when i had 16500. i now have 1 forward gear and 5 reverse !

i'll try tomorrow and report back.

thank you

[Pete9008]

I seem to remember a change in the way manual current was done in 5.24 but can't remember the details. It might explain the difference though.

Not sure which phase is which on the inverter, yet to get mine hooked up.

Fairly sure I know how to automate a lot of the setup. Got a bit of it coded but no testing done yet. Hopefully get back to it soon though (although probably not soon enough to help you with this ).

[FFMan]

> Fairly sure I know how to automate a lot of the setup.

yeah seems to me there should be away. Those e-scooter controllers i've bought off ebay have a learn mode and seem to work it out whatever the wiring. I don't doubt our stuff is more complex but feels like it should be possible to auto-tune in time

[Pete9008]

there seems to be a fair amount of stiction in my drivetrain. takes some oomph to get it turning then it's easier kind of thing.

Be worth keeping an eye on this, hopefully its just cold gearbox oil and a bit of drag on the gears but it could also be that you have too much side or end load on a motor bearing. If it is the later then the bearing won't last too long before it overheats!

[FFMan]

We left 10mm float on the coupling to allow for expansion etc so hopefully ok but one to keep an eye on as you say

[Pete9008]

That should be fine then, hopefully its just drag from the cold gearbox oil.

Had a thought on the two syncofs values. There is a way to definitively determine which one is the right one.

If you find the two values and then go back to the sine software that you used for test spin and resolver setup and then plot angle against i1 while test spinning then the the syncofs value that gives an angle of zero while i1 is at the positive peak should be the one you want (the other, wrong, one should read zero degrees when i1 is at the negative peak).

Edit - at least it should work, really need to check the sine code to make sure there are no unexpected offsets on the phase angles!

Edit2 - just checked, should work ok.

[FFMan]

thanks pete - i'll try sine f/w when back from work and post results

[Mhs]

I'm following with interest, I wasn't aware that each motor is unique, I thought (for whatever reason) all Mitsubishi phev rear motors have similar syncofs.

[Pete9008]

It would be nice if they are. Seem to remember someone disassembling one and saying to scribe the resolver to make sure it went back in in exactly the same position so I'm assuming there is some mechanical adjustment built in to them so the could well be factory calibrated.

Also even if they are all factory set the same you still need to get the phase connections and resolver connections exactly the same to use the same syncofs value.

Edit - it all comes down to whether the manufacturer decided it was cheaper/easier to calibrate the resolver mechanically or in software. Would be nice to know what each one does.

[FFMan]

I'm following with interest, I wasn't aware that each motor is unique, I thought (for whatever reason) all Mitsubishi phev rear motors have similar syncofs.

where are you with your build - looks similar to mine

[Mhs]

where are you with your build - looks similar to mine

Car needs a bit of rust repairs and so on before it gets dismantled, I'm using the time to collect the items (and knowledge hopefully). I don't have batteries yet, so planning to bench test soon with probably lead-acid batteries.

[Mhs]

Edit - it all comes down to whether the manufacturer decided it was cheaper/easier to calibrate the resolver mechanically or in software. Would be nice to know what each one does.

It will be interesting to understand how is that done systematically..

[FFMan]

That should be fine then, hopefully its just drag from the cold gearbox oil.

Had a thought on the two syncofs values. There is a way to definitively determine which one is the right one.

If you find the two values and then go back to the sine software that you used for test spin and resolver setup and then plot angle against i1 while test spinning then the the syncofs value that gives an angle of zero while i1 is at the positive peak should be the one you want (the other, wrong, one should read zero degrees when i1 is at the negative peak).

Edit - at least it should work, really need to check the sine code to make sure there are no unexpected offsets on the phase angles!

Edit2 - just checked, should work ok.

is there a syncoff valure in the sine f/w - i don't think there is but i can verify later

[Pete9008]

is there a syncoff valure in the sine f/w - i don't think there is but i can verify later

Just checked and you're right, syncofs only in the foc code, so that idea isn't going to work after all. Back to the trial and error approach