Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[MPaulHolmes]

It's controlling torque rather than rpm, and since the current range is like -400amp to 400amp max using a lem Hass 200-s, it can only really hold at like +/- 4 amps (1%). Can you hold it still with your hand?

It always has a little funny behavior when the intent is to control torque in a 2000 pound vehicle, but is unloaded.

[motorulf]

Yes i can hold it with my hand, it is very weak.

I ran pi-test2
i got these results:

+0030,+0529,+0529,+0499,+0463,+0510,+0482,+0482,+0522,+0510,+0493,+0510,+0569,+0510,
+0493,+0563,+0510,+0493,+0504,+0475,+0493,+0533,+0463,+0493,+0475,+0493,+0464,+0483,
+0475,+0475,+0475,+0493,+0466,+0467,+0467,+0468,+0445,+0488,+0488,+0453,+0493,+0472,
+0469,+0477,+0455,+0475,+0478,+0476,+0477,+0462,+0479,+0464,+0495,+0481,+0496,+0494,
+0494,+0481,+0481,+0482,+0470,+0471,+0468,+0483,+0473,+0475,+0475,+0463,+0477,+0473,
+0461,+0474,+0481,+0487,+0476,+0476,+0477,+0477,+0470,+0471,+0477,+0472,+0471,+0463,
+0465,+0474,+0465,+0466,+0466,+0466,+0450,+0469,+0459,+0463,+0453,+0468,+0456,+0462,
+0462,+0469,+0451,+06,+0447,+0452,+0453,+0454,+0440,+040410,+0393,+0395,+0377,+0416,
+0362,+0382,+0402,+0364,+0376,+0378,+0410,+0369,+0382,+0360,+0324,+0363,+0347,+0366,
+0399,+0337,+030,+0330,+0327,+0347,+0330,+0332,5,+0267,+0248,+0249,+0236,+0255,+0256
,+0243,+0230,+0233,+0237,+00198,+0202,+0205,+0193,+0195,+019,+0192,+0160,+0156,+0211
,+0252,+0166,+0180,

I am not sure how to graph these though, also i think my terminal emulator is messing up the line endings, or adding line endings maybe.

[MPaulHolmes]

Oh you don't need to graph it. I wasn't thinking. Just notice that it REALLY SLOWLY approaches zero. Zero means Id and Iq have officially converged to their commanded values. Your PI values are too small. Would it be practical to do the testing with like 50 volts instead of 25v? Now that it's being a little better behaved it might be safe. Either way, I'd try maybe doubling both Kp and Ki and run
run-pi-test2 again. The goal is to get the numbers to race to zero in maybe 20 or 30 or whatever iterations. like this:

511, 490, 400, ..., 50, 0, -50, 0, 20, (hurray, it converged!)

[motorulf]

I connected 75V.
I tried to manually tune the kp and ki. i am not very good at it.
I ran pi-test, and pi-test 2. Looks like this:

run-pi-test
The test was a success.  If you are staying with this bus voltage, type 'save':
If you change your bus voltage, you should rerun the command 'run-pi-test'.

config
motor-type=1
kp=17422 ki=00281
current-sensor-amps-per-volt=0320
max-regen-position=0360
min-regen-position=0420
min-throttle-position=0500
max-throttle-position=0600
fault-throttle-position=0020
max-battery-amps=0015 amps
max-battery-amps-regen=0015 amps
max-motor-amps=025 aps
max-motor-amps-regen=025 amps
precharge-time=0050 tenths of a sec
rotor-time-constant=030 ms
pole-pairs=002
max-rpm=06000 rev/min
throttle-type=0
encoder-ticks=0256 ticks/rev
pi-ratio=062
raw-throttle=0479

run-pi-test2
+0019,+0539,+0520,+0492,+0465,+0373,+0253,+0151,+0177,+0209,+0235,+0248,+0223,+0203,
+0184,+0128,+0096,+0055,+0168,+0188,+0145,+0117,+0067,+0128,+0115,+0088,+0091,+0104,
+0087,+0056,+0065,+0064,+0058,+0066,+0071,+0076,+0090,+0083,+0080,+0057,+0056,+0082,
+0074,+0036,+0060,+0067,+0062,+0017,+0020,+0028,-0022,+0038,+0029,+0034,+0059,+0103,
+0089,+0037,+0022,+0030,+0077,+0021,+0012,+0033,+0039,+0038,+0026,+0008,+0006,+0012,
-0018,-0006,+0033,+0062,+0063,-0007,-0034,+0023,+0024,+0039,+0016,+0003,+0032,+0070,
+0016,+0004,+0020,-0018,+0003,+0015,+000+0025,+0018,+0043,-0002,-0034,-0018,+0048,+0
018,-0014,+0003,-0056,-0072,-0027,+0010,+0040,+0042,+0061,+0025,-0022,+0001,+0015,+0
005,+0006,+0005,+0001,-0009,-0006,-0008,-0036,-0044,-0004,+0017,+0045,-0016,-0022,-0
021,-0006,+0033,+0032,+0002,-0003,-0006,-0010,-0003,-0021,-0012,+0004,+0023,+0027,+0
022,+0011,-0001,-0002,+0002,+0002,+0011,+0013,-0023,-0056,-0086,-0042,-0019,+0023,+0
049,+0017,+0022,+0008,+0000,-0011,-0005,-0012,-0022,-0040,-0020,-0016,-0006,+0028,+0
013,-0007,-0041,-0034,-0035,-0014,+0041,+0052,+0012,-0034,-0082,-0018,+0001,-0002,+0
024,+0014,-0009,-0004,-0018,-0035,-0005,+0026,-0034,-0062,-00017,+0020,-0018,-0054,-
0038,+0033,,-0007,-0016,-18,-0015,-0016,-0007,-0058,-0033,-0049,-0053,+0020,+0078,+0
032,-0041,-0019,-0013,-0017,-0017,-0025,-0049,-0035,-0023,+0027,+0027,+0011,-0067,-0
085,

Then i was trying to do the rotor-test.
It never does anything. no humming, no movement.
So i tried altering the rotor-time-constant myself. I find that hard too.
After much testing i decided that somewhere around 50-60 gives the fastest acceleration.
Is that what to look for?
And does the ki and kp look resonable?
How do i calculate the rotor-time-constant if i have resistance and inductance?
I have an old scope so i guess i can calculate the inductance of the motor.

My testing had to be put on hold for a while, -35°C outside now.
I had to fetch some fire wood and try to heat up this place a bit. The cat was complaining hehe

[motorulf]

I hade some success with the reverse.
Now there is full power going backwards too, but thats just thrilling
Not quite ready for public release yet, but soon.

I made a small adapter that connects to the ICSP.
Thats because in the newest source code it looked like all other pins are in use.
The negative side of that is that you can no longer debug the code.
I took the LED output and had it show when the reverse is enabled or not.
So you can see that it will not let it change into reverse unless it is rotating really slow.

In the video the throttle is setup with clockwise=accelerate, anticlockwise=regen.

www.youtube.com/watch?v=kkGfkSDkq8Y

[motorulf]

What about having a momentary switch for the reverse. Then we could also use it for cruise control

Like this:
Motor is not rotating and the brake is applied, you push the button, it switches between forward and reverse. Still with a LED indicating the choice.
Motor is rotating, you push the button, the cruise control is enabled (maybe not while reversing).
You apply brakes, cruise control is disabled.
You apply a lot of throttle, or a fast rise, cruise control is disabled.

Good or bad idea?

[MPaulHolmes]

Sounds like a real cruise control to me!

[MPaulHolmes]

Originally Posted by motorulf

I connected 75V.
I tried to manually tune the kp and ki. i am not very good at it.
I ran pi-test, and pi-test 2. Looks like this:

run-pi-test
The test was a success.  If you are staying with this bus voltage, type 'save':
If you change your bus voltage, you should rerun the command 'run-pi-test'.

config
motor-type=1
kp=17422 ki=00281
current-sensor-amps-per-volt=0320
max-regen-position=0360
min-regen-position=0420
min-throttle-position=0500
max-throttle-position=0600
fault-throttle-position=0020
max-battery-amps=0015 amps
max-battery-amps-regen=0015 amps
max-motor-amps=025 aps
max-motor-amps-regen=025 amps
precharge-time=0050 tenths of a sec
rotor-time-constant=030 ms
pole-pairs=002
max-rpm=06000 rev/min
throttle-type=0
encoder-ticks=0256 ticks/rev
pi-ratio=062
raw-throttle=0479

run-pi-test2
+0019,+0539,+0520,+0492,+0465,+0373,+0253,+0151,+0177,+0209,+0235,+0248,+0223,+0203,
+0184,+0128,+0096,+0055,+0168,+0188,+0145,+0117,+0067,+0128,+0115,+0088,+0091,+0104,
+0087,+0056,+0065,+0064,+0058,+0066,+0071,+0076,+0090,+0083,+0080,+0057,+0056,+0082,
+0074,+0036,+0060,+0067,+0062,+0017,+0020,+0028,-0022,+0038,+0029,+0034,+0059,+0103,
+0089,+0037,+0022,+0030,+0077,+0021,+0012,+0033,+0039,+0038,+0026,+0008,+0006,+0012,
-0018,-0006,+0033,+0062,+0063,-0007,-0034,+0023,+0024,+0039,+0016,+0003,+0032,+0070,
+0016,+0004,+0020,-0018,+0003,+0015,+000+0025,+0018,+0043,-0002,-0034,-0018,+0048,+0
018,-0014,+0003,-0056,-0072,-0027,+0010,+0040,+0042,+0061,+0025,-0022,+0001,+0015,+0
005,+0006,+0005,+0001,-0009,-0006,-0008,-0036,-0044,-0004,+0017,+0045,-0016,-0022,-0
021,-0006,+0033,+0032,+0002,-0003,-0006,-0010,-0003,-0021,-0012,+0004,+0023,+0027,+0
022,+0011,-0001,-0002,+0002,+0002,+0011,+0013,-0023,-0056,-0086,-0042,-0019,+0023,+0
049,+0017,+0022,+0008,+0000,-0011,-0005,-0012,-0022,-0040,-0020,-0016,-0006,+0028,+0
013,-0007,-0041,-0034,-0035,-0014,+0041,+0052,+0012,-0034,-0082,-0018,+0001,-0002,+0
024,+0014,-0009,-0004,-0018,-0035,-0005,+0026,-0034,-0062,-00017,+0020,-0018,-0054,-
0038,+0033,,-0007,-0016,-18,-0015,-0016,-0007,-0058,-0033,-0049,-0053,+0020,+0078,+0
032,-0041,-0019,-0013,-0017,-0017,-0025,-0049,-0035,-0023,+0027,+0027,+0011,-0067,-0
085,

Then i was trying to do the rotor-test.
It never does anything. no humming, no movement.
So i tried altering the rotor-time-constant myself. I find that hard too.
After much testing i decided that somewhere around 50-60 gives the fastest acceleration.
Is that what to look for?
And does the ki and kp look resonable?
How do i calculate the rotor-time-constant if i have resistance and inductance?
I have an old scope so i guess i can calculate the inductance of the motor.

My testing had to be put on hold for a while, -35°C outside now.
I had to fetch some fire wood and try to heat up this place a bit. The cat was complaining hehe

Sorry for the delay, I missed this message.
run-rotor-test did nothing?! It does take a long time to run, but yes, you are looking for best acceleration. 50-60 sounds reasonable. There is a range of values that work best. Big motors tend to be in the 25-80 range. 50-60 is fine. There is a way to compute the rotor time constant involving the stator inductance/resistance and mutual inductance, but the true value is going to be in the range of 50-60 anyway. All this is doing is figuring out what causes the best acceleration, so the "true" value isn't going to be better. Also, the true value varies a little bit with temperature, so you could compute 55, but while driving it will vary some anyway.

Yikes, -35degC is ridiculous. I've never been in anything like that!

[ale0502]

Hello everyone, I have my controller finished to test, or at least I think. I have several questions for Paul:
What would be the steps to be performed for the first ignition?
I have a load of 3 resistors in star, not connected to a motor, so the encoder does not rotate.
How do you use the routines run-pi-test, runrotortest, etc?

Thank you very much

[MPaulHolmes]

For the PI test, do you know your stator resistance? we can skip this test if you have that info. By stator resistance, I mean 0.5*resistance between any 2 of the 3 motor cables.

Did I solder this, or did you? If you did it, you will first have to verify that the PWM can make it to the output. Do you have it hooked up to IGBTs and a bus capacitor? Could you take a picture of it and post it here?