openinverter forum

Hi,

I have question about syncofs searching for Leaf motor + v3 board + leaf inverter interface.
According to procedure written at https://openinverter.org/wiki/Using_FOC_Software
Leaf parameters from wiki and latest FOC firmware
I have setup:
DC voltage: 50V
12 V applied to: pin7-START; pin13-FORWARD

And now: like You wrote should I push "run in manual mode" on web interface. With syncofs=0 and manualid=15A motor is not spinning even I will go with manualid=150A. According what You wrote it should start at some value.
If I push stop inverter, change syncofs for other value and then press start in maual it a little change clicking noise, so probably according to syncofs it changes offset, but still it's not spinning.
Strange for me is, that it should spinning in some manualid value. Mine not - can You look if I'm wrong somewhere?

Have you verified your sin/cos connection is correct. Try swapping those cables. Ex and! Ex are always good, if you change direction manually (by phase cables) you need to account for correct sin/cos pins...

Try changing offset in steps of 5000 until motor starts turning in some direction.

I have cables from reslover connected like it was previously in original harness from Leaf. Phase cables also are in the same direction like Nissan did between motor and inverter. I think if I connected all above cables like producer did, so also Johannes on his mainboard probably doesn't mix them? Otherwise I will swap them?
Strange for me is, that if according to wiki syncofs=0 and manualid increased form let's say 15A to 150A something should change. In my case nothing is changing. Changes are only when I'm changing syncofs and stay with some (doesn't matter what) value for manualid.
Depends of changed syncofs from 0 to 35000 motor clicks less or more and that's all.

damian.lo wrote: ↑Sun Feb 16, 2020 10:00 pm I have cables from reslover connected like it was previously in original harness from Leaf. Phase cables also are in the same direction like Nissan did between motor and inverter. I think if I connected all above cables like producer did, so also Johannes on his mainboard probably doesn't mix them? Otherwise I will swap them?
Strange for me is, that if according to wiki syncofs=0 and manualid increased form let's say 15A to 150A something should change. In my case nothing is changing. Changes are only when I'm changing syncofs and stay with some (doesn't matter what) value for manualid.
Depends of changed syncofs from 0 to 35000 motor clicks less or more and that's all.

Did you also account for RST phase cables? Mark them and then try swapping them across. Just because phase cables are usually the easier to change.
If motor moves in the correct direction leave it at that, if no, then return phase cables and swap sin/cos connection.
I had to swap it like 3 times since i didnt mark cables. Its nothing serious, just logic failing at direction change.

Ok, today I will do like you wrote. I had the same situation with v2 board + mine enc-resolver board and solution was like You wrote. With v3 board and mod of Leaf inverter I was curious if it should be working with connection 1:1.
Later I will give feedback.

Have you applied this configuration: https://openinverter.org/wiki/Configura ... Leaf_Motor ?
In Leaf Gen2 inverter phases 1 & 3 must be swapped - but not physically, just by "pinswap" parameter. It is in the configuration linked above. Resolver must not be swapped. Do you get movement if instead you increase manualiq?

Yes, like you wrote I previosuly uploaded parameters and firmware from link. Resolver and phases are 1:1 connected. With "start in manual by web browser + 12V start + 12V on forward" no matter what current I applied to manualiq, motor is not moving, even not clicking.
Shoul be also another input connected, which I forgotten?
If I press stop inverter, change syncofs and then start inverter in window, sometimes motor clicks, but without rotation.

EDIT 17.02.2020:
One mine fault: probably I pressed too fast "start in manual" via web browser, faster than board get 12V on START input. Now, when I have status RunManual is possible to increase manualiq and finally I found syncofs, where motor stops.
Another thing, which I solved today: I had too big noise (poor contact on connector) on resolver pins, so I had deviation on angle about 15 degrees. Now it's solved, so now only to find good direction of rotation. Thanks for help.

Hi forgive me for hijacking this thread. But it seems a good place to start...

So i made Ampera inverter brain with Rev 2 chip and resolver interface and now it behaves as rev 3 board.
I got a servo motor, 6 pole Stromag FOP32-0200-40-S2
Resolver is Tamagawa TS2640N321E64 and i see it is 6 pole resolver on a first glance.
I thought it might make a good experimental motor. It comes with a brake that you release by applying 24V to its contacts.
I use 350Vdc VW Golf hybrid battery as a test bed.

Long story short i think i got resolver correct according to Stromag schematic. I get a good view on potato Sin and smaller potato Cos.
I also connected phase cables as they state in schematic U,V,W.

Now i went on and started inverter and tried to give it some juice. It tripped OClimit of course...
Well then i went and followed instructions for FOC offset. I connected in manual and rotated shaft untill i could see about 2°on a dial. Is that as 0 as it gets? If i would turm just a bit more angle would jump to the other side.

Then i entered a 1A value to manualid and got a bang and instant OClimit.
[31263]: STOP - OVERCURRENT
[31264]: WARN - HICUROFS1
[31264]: WARN - HICUROFS2

I remember Johannes said if output measuring is incorrect you get instant OClimit with FOC.
Do you think i should switch current sensor channels in code? How to proiceed?

I am back. I went on and read about those errors.
It seems that my Olimex pins and current sensors are not measuring the same phase cables that FOC wants to measure. Of course Johannes did find that out and put in code parameter "pinswap".
I changed this to "currents12" value and immediately after i got fammilliar shaking from the motor.
I now just have to find correct resolver offset.

EDIT:
Hm... i only get the motor rumbling and when i change the offset through steps of 1000 motor sometimes seems to quiet down. I hadnt had any luck in finding correct angle. I remember Leaf had a rather wide area where it started to spin. Span of 8000 points.
Does that mean resolver signal is turned the wrong way?
Do i need to change my phase cables?
Or do i need to change resolver cables?

Johannes remember in code i already selected to swap current sensor positions. Now i cant swap resolver channels without getting OC error. Maybe it would be good to separate the pin swapping parameters into 3 different options. It would greatly help with new boards design.

EDIT2: What did you mean exactly with statement "Keep increasing the value until you notice that the motor starts to spin" Which value? "manualid" or "syncofs" ? Or did you mean increase the value untill motor starts to rock?

Well... i inspected my resolver signal.
I get good square signal at 4.4kHz from PD2.
Then i measured Sin and cos signals. I get good 1.5V supply signal at S1 and S4. But i get much more on S2 and S3. As i rotate axis i get increasing signal from 0V to 5V!!! Is that good for uC? I am powering amplifier with 12V directly behind the input diode. Should i rather power it from 5V?
Did i mess up resolver circuit? I just used the schematic from this thread viewtopic.php?f=2&t=15#p65.
Should i use the latest one that goes with your rev.3 mainboard Johannes?

Why cant i get a good resolver signal like i got here viewtopic.php?f=2&t=15&start=20#p380? I am using the same circuit.

damian.lo wrote: ↑Mon Feb 17, 2020 1:30 pm Another thing, which I solved today: I had too big noise (poor contact on connector) on resolver pins, so I had deviation on angle about 15 degrees. Now it's solved, so now only to find good direction of rotation. Thanks for help.

Hi Damian, can you make a snapshot of your resolver signals with scope on pins R1, S2 and S3? I need to see the shape of signal. I think something is wrong with my resolver circuit.

A

arber333 wrote: ↑Wed Mar 18, 2020 7:11 pm Well... i inspected my resolver signal.
I get good square signal at 4.4kHz from PD2.
Then i measured Sin and cos signals. I get good 1.5V supply signal at S1 and S4. But i get much more on S2 and S3. As i rotate axis i get increasing signal from 0V to 5V!!! Is that good for uC? I am powering amplifier with 12V directly behind the input diode. Should i rather power it from 5V?
Did i mess up resolver circuit? I just used the schematic from this thread viewtopic.php?f=2&t=15#p65.
Should i use the latest one that goes with your rev.3 mainboard Johannes?

Why cant i get a good resolver signal like i got here viewtopic.php?f=2&t=15&start=20#p380? I am using the same circuit.

That is seriously wrong, the maximum magnitude of the voltage signal you get from the COS/SIN feedback windings should be around the same. I wonder if you are accidently driving the wrong pair as the exciter. JB kindly measured the resistance of the coils for us: Exciter 17 Ohms, Sin and Cos 37 Ohms. Don't power it from 5V, find the problem so both Sin/Cos outputs have the same peak to peak voltage range.

kiwifiat wrote: ↑Thu Mar 19, 2020 9:46 pm That is seriously wrong, the maximum magnitude of the voltage signal you get from the COS/SIN feedback windings should be around the same. I wonder if you are accidently driving the wrong pair as the exciter. JB kindly measured the resistance of the coils for us: Exciter 17 Ohms, Sin and Cos 37 Ohms. Don't power it from 5V, find the problem so both Sin/Cos outputs have the same peak to peak voltage range.

I know! But i have schematic for the motor and it states exact wiring of the resolver plug. I have prepared it according to the schematic.
Next i found Tamagawa resolver manual and found wiring schematic for it. Wire colours are a match to connector wiring.
Also i measured exciter and sin/cos pins and i get good values out of them.

EDIT: I measured again and i get:
R1-R2 = 54R
S1-S3 = 45R
S2-S4 = 45R
It does seem that exciter winding has more turns than Sin/Cos.

There is actually something in resolver manual. It declared normal signal from resolver is 1x signal. Then it talks about signal multiplication. I thought his was pole numbers or similar. Now it seems they actually wire resolver to output higher voltage...

I need to get a proper automotive motor here and test its resolver....

I have good resolver reporting from 0 to 360. Tested with wifi dial.
I noticed resolver position jumps by a +/-2°when standstill.
My resolver S2 and S3 signal has more than 5V peak to peak, but i can stil get good reporting from the dial.
I have noticed a lot of hash on resolver line when power stage is active though.
I think resolver has some signal boost, since R1 carries only 3V3 signal. It is Tamagawa TS264ON321E64

Hey! I took apart resolver cover and i saw some markings on it. First two from the center were resolver 0°position and they were connected. The other remaining mark must show stator 0°offset position.
Johannes how do you calculate offset in angular position? Is it ok to calculate 65535 / 360 = 182 for one degree of offset? If i just measure angle then i can input that or its 180deg shifted angle.

arber333 wrote: ↑Fri Mar 20, 2020 9:50 am
I know! But i have schematic for the motor and it states exact wiring of the resolver plug. I have prepared it according to the schematic.
Next i found Tamagawa resolver manual and found wiring schematic for it. Wire colours are a match to connector wiring.
Also i measured exciter and sin/cos pins and i get good values out of them.

EDIT: I measured again and i get:
R1-R2 = 54R
S1-S3 = 45R
S2-S4 = 45R
It does seem that exciter winding has more turns than Sin/Cos.

There is actually something in resolver manual. It declared normal signal from resolver is 1x signal. Then it talks about signal multiplication. I thought his was pole numbers or similar. Now it seems they actually wire resolver to output higher voltage...

I need to get a proper automotive motor here and test its resolver....

If you are getting 1.5Vp-p on one set of windings and 5Vp-p on the the other as you indicate above then the ADC in the STM32 will be saturated on the 5Vp-p signal and it is not possible to calculate an accurate angle from the data. The fact that your resolver has more windings on the exciter than the feed back coils in not important. All that matters is that the Vp-p of both sin and cos signals are close and the magnitude is such that the ADC does not max out. That is to say that after the ~1.6V offset that is added to the feedback signals by the circuitry on the controller board the voltage seen by the STM32 does not exceed 3.3V. Unless you isolate and resolve the issue that is causing such a huge disparity between the Vp-p Sin/Cos signals there is no chance of getting that motor running with inaccurate angle data.

Maybe johu could chime in please?

Why then i can get a good position feedback if i rotate motor by hand?
The way Tamagawa manual is stating you can order resolver from them to have up to 4x the signal boost. So that would mean 1.5V * 4 = 6V. This is close to what i am seeing when i measure signal with scope.

This motor is weird...
I think soon i will get a "normal" hybrid motor from Remy. It should run better. If not for the fact i already had it on my desk running at 10000rpm. DOH!

Hi,
Sorry Arber for late reply. For now I can't send You any scopes from my resolver because I have big mess under hood. I'm finishing now main harness so in next days will be finished and then some trials.
I will remember about this and send You some screens if You still need them.

arber333 wrote: ↑Fri Mar 20, 2020 6:54 pm Why then i can get a good position feedback if i rotate motor by hand?
The way Tamagawa manual is stating you can order resolver from them to have up to 4x the signal boost. So that would mean 1.5V * 4 = 6V. This is close to what i am seeing when i measure signal with scope.

This motor is weird...
I think soon i will get a "normal" hybrid motor from Remy. It should run better. If not for the fact i already had it on my desk running at 10000rpm. DOH!

That is a good question, it should be mathematically impossible to calculate the angle correctly if the one of the signals exceeds 3.3V for any significant time. Did you scope the output of the resolver circuit when it was actually driving the load? It makes no rational sense that one of the 45R feed back coils would output 5Vp-p and the other 1.5Vp-p. The fact that they have the same resistance would rule out a short in one of them. Did you scope both outputs before connecting to the controller? An interesting mystery

Off topic: I got a resolver to HALL converter working for the Lebowski. Tested on a Hyundai HSG with a 3 pole pair resolver. Test on a Leaf resolver next week.

kiwifiat wrote: ↑Sat Mar 21, 2020 1:49 am

arber333 wrote: ↑Fri Mar 20, 2020 6:54 pm Why then i can get a good position feedback if i rotate motor by hand?
The way Tamagawa manual is stating you can order resolver from them to have up to 4x the signal boost. So that would mean 1.5V * 4 = 6V. This is close to what i am seeing when i measure signal with scope.

This motor is weird...
I think soon i will get a "normal" hybrid motor from Remy. It should run better. If not for the fact i already had it on my desk running at 10000rpm. DOH!

That is a good question, it should be mathematically impossible to calculate the angle correctly if the one of the signals exceeds 3.3V for any significant time. Did you scope the output of the resolver circuit when it was actually driving the load? It makes no rational sense that one of the 45R feed back coils would output 5Vp-p and the other 1.5Vp-p. The fact that they have the same resistance would rule out a short in one of them. Did you scope both outputs before connecting to the controller? An interesting mystery

Off topic: I got a resolver to HALL converter working for the Lebowski. Tested on a Hyundai HSG with a 3 pole pair resolver. Test on a Leaf resolver next week.

Is it a hardware based converter (like the PGA411) or software based (like an Arduino/Teensy or similar)?

Well i guess it was really problem of ADC and resolver signal boost.

On second try when i got a trusty Remy HVH250 back on my desk it started to run immediately.
I wired resolver wires as they were declared and i connected phase wires and i was done!
Here is a video of the first run and some pics of assembly.


I am done here i think. Next comes the dual Ampera connection!

slow67 wrote: ↑Sat Mar 21, 2020 5:05 pm

kiwifiat wrote: ↑Sat Mar 21, 2020 1:49 am

arber333 wrote: ↑Fri Mar 20, 2020 6:54 pm Why then i can get a good position feedback if i rotate motor by hand?
The way Tamagawa manual is stating you can order resolver from them to have up to 4x the signal boost. So that would mean 1.5V * 4 = 6V. This is close to what i am seeing when i measure signal with scope.

This motor is weird...
I think soon i will get a "normal" hybrid motor from Remy. It should run better. If not for the fact i already had it on my desk running at 10000rpm. DOH!

That is a good question, it should be mathematically impossible to calculate the angle correctly if the one of the signals exceeds 3.3V for any significant time. Did you scope the output of the resolver circuit when it was actually driving the load? It makes no rational sense that one of the 45R feed back coils would output 5Vp-p and the other 1.5Vp-p. The fact that they have the same resistance would rule out a short in one of them. Did you scope both outputs before connecting to the controller? An interesting mystery

Off topic: I got a resolver to HALL converter working for the Lebowski. Tested on a Hyundai HSG with a 3 pole pair resolver. Test on a Leaf resolver next week.

Is it a hardware based converter (like the PGA411) or software based (like an Arduino/Teensy or similar)?

Based on this post: https://www.diyelectriccar.com/forums/s ... stcount=41
Credit to Dr Andrew Pugh for the original design.

Good job Arber getting the Remy running.

@kiwifiat, which chip are you using for resolver to 6 step conversion?

I have the circuit drawn for AU6802N1 chip on my Lebowski board, but i am lacking incentive since Leaf motor is just as good with my RLS modified UVW sensor.
Of course i have AU6802N1 chips scavenged from Prius main board!

Well it took no time to find offset of 9500pt. Motor is 10pole with matching resolver.
I can rotate it quite smoothly in CW direction. I can set max RPM to 600Hz and i get 7000rpm...
Then if i select REVERSE motor spins CCW. But there it starts to growl and has some powerful torque that wants to jump the motor at smallest of throttle change.
Would that be something left to setup?

Hm... when i tried to fine tune the offset like here:

"You have found 1 of 2 possible offsets. Now you can enter a small value for "manualiq" and reset "manualid" to 0. The motor should spin clockwise".

Well i tried to order "manualiq" and i dont get any reaction from inverter.
My motor rotates CW when i order it by throttle,
If i change direction by switch, motor would spin CCW. However here motor is rough and when i return throttle to 0 it starts to rattle until it stops.

Also i tried to use regen with throttle at 1/3. Motor spins under throttle and when i press brake switch motor starts rumbling and shaking until stop not dissimilar to CCW rotation above.

Any thoughts? Is there some parameter that is too large...

A

Well, I'm back 1 week early (even though I would have preferred to hide out in Sweden for a few months).
I spotted a mistake in the instructions: PWM is disabled with 0 "id". So set manualid to a small non-zero value.

Regarding the rumbling I don't get this with the Leaf motor. What happens if you add 180° (32768) to synofs? Does it spin the other direction smoothly? Otherwise you might still have a twist in current or resolver channels.